EASTERN AND SOUTHERN AFRICA UNITED REPUBLIC OF TANZANIA Estimating the Impact of Climate Change in Tanzania World Bank Group November 2024 © 2024 The World Bank Group 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org This work is a product of the staff of the International Bank for Reconstruction and Development (IBRD), the International Development Association (IDA), the International Finance Corporation (IFC), and the Multilateral Investment Guarantee Agency (MIGA), collectively known as The World Bank Group, with external contributors. The World Bank Group does not guarantee the accuracy, reliability or completeness of the content included in this work, or the conclusions or judgments described herein, and accepts no responsibility or liability for any omissions or errors (including, without limitation, typographical errors and technical errors) in the content whatsoever or for reliance thereon. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of any of the organizations of The World Bank Group concerning the legal status of any territory or the endorsement or acceptance of such boundaries. The findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of IBRD/IDA, IFC and MIGA, their respective Boards of Executive Directors, and the governments they represent. The contents of this work are intended for general informational purposes only and are not intended to constitute legal, securities, or investment advice, an opinion regarding the appropriateness of any investment, or a solicitation of any type. Some of the organizations of The World Bank Group or their affiliates may have an investment in, provide other advice or services to, or otherwise have a financial interest in, certain of the companies and parties named herein. Nothing herein shall constitute, or be construed, or considered to be a limitation upon or waiver of the privileges, and immunities of any of IBRD/IDA, IFC and MIGA, all of which are specifically reserved. Rights and Permissions The material in this work is subject to copyright. Because the World Bank Group encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given, and all further permissions that may be required for such use (as noted herein) are acquired. The World Bank does not warrant that the content contained in this work will not infringe on the rights of third parties, and accepts no responsibility or liability in this regard. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. UNITED REPUBLIC OF TANZANIA Estimating the Impact of Climate Change in Tanzania COUNTRY CLIMATE AND DEVELOPMENT REPORT Table of Contents Acknowledgmentsvii Abbreviations, Acronyms and Glossary of Key Terms viii 1. Introduction 1 2. Methodological Overview 2 2.1. Analytical approach 2 2.2. Development and adaptation scenarios 4 2.3. Climate information 4 2.3.1. Background and historic climate 4 2.3.2. Selection of future climate scenarios 7 2.3.3. Expected changes by 2050 11 3. Impact Channel Results 14 3.1. Human capital 14 3.1.1. Heat and labor productivity 14 3.1.2. Human health 19 3.2. Natural capital 25 3.2.1. Land use and land cover 25 3.2.2. Crop production 26 3.2.3. Soil erosion 33 3.2.4. Livestock production 37 3.3. Physical capital 42 3.3.1. Inland flooding 42 3.3.2. Bridges 47 3.3.3. Roads 49 4. Conclusion 55 ii  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania References58 Appendix A: Climate Scenario Selection 61 A1. Climate data and projections used 61 A2. Processing of climate information 63 A3. Selection of climate scenarios 65 A4. Daily interpolation of monthly data 67 A5. Results of the climate scenario selection process 67 A6. References 71 Appendix B: Impact Channel Methods 72 B1. Human capital 72 B1.1. Labor supply model 72 B1.2. Heat and labor productivity 72 B1.3. Human health  74 B2. Natural capital 81 B2.1. Crop production 81 B2.2. Livestock production 84 B2.3. Soil erosion 87 B3. Physical capital 89 B3.1. Infrastructure and capital stock model 89 B4. References 97 Appendix C: Data Sources 103 C1. Heat and labor productivity 103 C2. Human health 103 C3. Crop production 104 C4. Livestock production 104 C5. Soil erosion 105 C6. Inland flooding 106 C7. Bridges 106 C8. Roads 107 Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  iii List of Tables Table 1: Overview of impact channels evaluated 3 Table 2: Selected climate scenarios 11 Table 3: Change in average national temperature by decade relative to 1995–2020  12 Table 4: Percentage change in average national precipitation by decade relative to historical baseline 13 Table 5: Proportion of occupations per sector 15 Table 6: Development scenario assumptions for the labor heat stress channel  15 Table 7: Outdoor workforce exposure by sector and development scenario 16 Table 8: Historical death and incidence rates by disease 19 Table 9: Development scenarios evaluated for the human health channel  21 Table 10: Change in mortality and morbidity rates (per 100,000 people) by disease, 2041–50  22 Table 11: Development scenarios evaluated for the crop production channel  29 Table 12: Development scenarios evaluated for the erosion channel  34 Table 13: Development scenarios evaluated for the livestock channel  38 Table 14: Development scenarios evaluated for inland flooding channel  43 Table 15: Adaptation scenarios evaluated for the bridges channel  47 Table 16: Adaptation scenarios evaluated for the roads channel  50 Table 17: Summary of estimated shocks by channel by the 2040s 55 Table A1: List of GCMs and the associated SSPs available for this study  62 Table B1: Possible climate effects on roads and adaptation measures 95 List of Figures Figure 1: Impact channel modeling approach  2 Figure 2: Country map 5 Figure 3: Koppen-Geiger climate classification map, 1991–2020 6 Figure 4: Mean precipitation and temperature by 0.5-degree grid cell, 1995–2020 6 Figure 5: Monthly mean temperature and precipitation, 1995–2020 7 Figure 6: Annual precipitation between TMA and CRU data, 1995–2020 8 Figure 7: Monthly maximum temperature between TMA and CRU data, 1995–2020 9 Figure 8: Monthly minimum temperature between TMA and CRU data, 1995–2020 9 Figure 9: Projected climate variables across a range of SSP-RCPSs 10 Figure 10: Change in mean temperature from 2041–50 relative to 1995–2020, by climate scenario 12 Figure 11: Change in mean precipitation from 2041–50 relative to 1995–2020, by climate scenario 13 iv  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 12: Labor productivity shocks, 2041–50 16 Figure 13: Labor productivity shocks, 3-year moving average 17 Figure 14: Breakdown of disease mortality and morbidity for 2015–19 21 Figure 15: Average labor supply shock by disease under the baseline, 2041–50 22 Figure 16: Fraction of waterborne diseases deaths attributable to temperature increase for 2041–50, relative to 1995–2020 23 Figure 17: Malaria transmissibility coefficient, 1995–2050 23 Figure 18: Labor supply shocks, all development scenarios, 3-year moving average 24 Figure 19: Land use allocation (count, thousands of hectares) by development scenario 25 Figure 20: Change in cropland coverage relative to 1995–2020, by region, all development scenarios26 Figure 21: Share of rainfed crops by area, production, and revenue, 2018/2019 average 27 Figure 22: Rainfed crop production shock under the baseline, 2041–50 27 Figure 23: Banana yield impact by region under the baseline, 2041–50 28 Figure 24: Tropical fruit yield impact by region under the baseline, 2041–50 28 Figure 25: Historical banana production distribution by region, 1995–20 29 Figure 26: Historical tropical fruit production distribution by region, 1995–20 29 Figure 27: Rainfed crop production shock under BAU scenario, 2041–50 30 Figure 28: Rainfed crop production shock broken down into heat and water effect, 2041–50 31 Figure 29: Rainfed crop production shock, under a BAU scenario, 3-year moving average 31 Figure 30: Rainfed crop production shock, under different scenarios, 3-year moving average 32 Figure 31: Erosion during historical period, 1995–20 34 Figure 32: Erosion risk by 2040 under two futures, BAU scenario 35 Figure 33: Crop production shock due to erosion, 2041–50 36 Figure 34: Crop production shock due to erosion, under different development scenarios, 3-year moving average 36 Figure 35: Share of livestock revenue, by product, 2017–21 average 38 Figure 36: Livestock production shock under a BAU scenario by product, 2041–50 39 Figure 37: Livestock production shock under all development scenarios, 2041–50 40 Figure 38: Livestock production shock under all development scenarios, 3-year moving average 41 Figure 39: Percent of capital exposed to flooding, 1995–2020 43 Figure 40: Inland flooding impacts expressed as percent of capital stock, 1995–2020, by region 44 Figure 41: Mean annual percent of total capital damaged by return period, under a BAU scenario 45 Figure 42: Change in capital damage for 100-year flood in each development scenario relative to 1995–2020, by region, with difference caused by climate change under SSP3–7.0 45 Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  v Figure 43: Change in capital damage for 100-year flood under SSP3–7.0 relative to BAU, by region 45 Figure 44: Annual average shock to capital by development and climate scenario under all development scenarios, 2041–50 46 Figure 45: Average annual bridge damage cost, by return period, under BAU 48 Figure 46: Expected annual incremental bridge damage by development scenario 48 Figure 47: Labor supply shock due to bridge delays, by adaptation scenario 49 Figure 48: Annual incremental road costs and delays under BAU for 2041–50  51 Figure 49: Additional annual damage in 2041–50 relative to 1995–2020, by region, under BAU 52 Figure 50: Projected annual incremental road costs by development scenario, 2041–50 52 Figure 51: Projected annual incremental delays by adaptation scenario, 2041–50 53 Figure A1: CMIP6 SSPs  61 Figure A2: Spatial downscaling methodology for ACCESS-CM2-SSP2–4.5 64 Figure A3: Climate scenario selection process 66 Figure A4: GCM selection results for dry/hot and wet/warm futures 68 Figure A5: GCM temperature (°C) and precipitation (%) change trajectories 69 Figure A6: GCM temperature (°C) and precipitation (%) change by 0.5-degree grid cell, 2041–50 vs. 1995–2020 70 Figure B1: Work capacity as a percentage from wet bulb globe temperatures 74 Figure B2: Fuzzy functions for malaria 76 Figure B3: Fuzzy functions for dengue 76 Figure B4: V-shaped function of excess mortality due to high temperatures 77 Figure B5: Change in waterborne disease incidence from changes in temperature 79 Figure B6: Fecal Contamination composite index indicators 80 Figure B7: Diarrheal disease relative risk 80 Figure B8: Relationship between precipitation and crop yield response 82 Figure B9: Illustrative relationship between temperature and yield response 83 Figure B10: Relationship between temperature and THI 86 Figure B11: Curve number model: relationship between precipitation and runoff 90 Figure B12: Routing of hydrographs schematic 91 Figure B13: Depth-damage functions 92 List of Box Box 1: Migration in Tanzania  17 vi  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Acknowledgments The Tanzania Country Climate and Development Report (CCDR) was prepared by a multisectoral World Bank Group team led by Diji Chandrasekharan Behr (Lead Environmental Economist, East Africa Environment Department), and William Battaile (Lead Economist, Macroeconomics, Trade & Investment, Eastern and Southern Africa), under the supervision of Paul Jonathan Martin (Manager, East Africa Environment Department) and Abha Prasad (Manager, Eastern and Southern Africa Macroeconomics, Trade and Investment Department), and the direction of Iain Shuker (Regional Director, Planet vertical, Eastern and Southern Africa). This background report on Estimating the Economic Damage of Climate Change in Tanzania was a key input to the Tanzania CCDR. It was prepared by a team in Industrial Economics Incorporated (IEc) led by Brent Boehlert, and involved Diego Castillo, Ken Strzepek and Kim Smet as key experts. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  vii Abbreviations, Acronyms and Glossary of Key Terms CCDR Climate Change and Development Report CGE Computable General Equilibrium (model) CMIP Coupled Model Intercomparison Project CRU Climatic Research Unit CRU TS Climatic Research Unit gridded Time Series FAECI fecal contamination composite index FAO Food and Agriculture Organization of the United Nations FCEGCM General Circulation Model (also known as Global Climate Model) GDP gross domestic product GHG greenhouse gas GRIP Global Roads Inventory Project LULC land use and land cover RCP Representative Concentration Pathway RUSLE Revised Universal Soil Loss Equation SSP Shared Socioeconomic Pathway THI temperature-humidity index TMA Tanzania Meteorological Authority All dollar amounts ($) are US dollars Climate projections: Simulated response of the climate system to a scenario of future emissions or concentrations of greenhouse gases (GHGs) and aerosols, and changes in land use. GCM: A modeled representation of the physical relationships of the global climate system that are used to generate climate projections. These capture atmospheric and ocean dynamics, and other water and biogeochemical cycles. Typical outputs include variables such as precipitation and temperature. Emission scenarios: A plausible representation of the future development of emissions of substances that are radiatively active, such as GHGs and aerosols. These are used in the form of illustrative RCPs. RCPs: Scenarios of emissions and concentrations of GHGs, aerosols, and land use/land cover. They represent different intensities in the additional radiative forcing caused by human activities. SSPs: Different possible evolutions of the world in terms of demography, technology, economy, and so on, with these socioeconomic conditions in turn achieving certain RCPs. viii  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania 1. Introduction This report outlines the process of estimating the economic damage of climate change for Tanzania, with these damage estimates subsequently informing the Tanzania Country Climate and Development Report (CCDR). Developing this CCDR provides an opportunity to better understand the benefits and costs of climate action and cross-sectoral policy priorities to manage climate risks effectively. Within the activity documented in this report, climate projections are run through biophysical and economic models to assess the country’s vulnerability to climate change under different policy scenarios: a business- as-usual (BAU) scenario that projects current growth trends in the country; an aspirational (ASP) scenario that assumes higher levels of growth and development; as well as scenarios that overlay nationally determined contribution (NDC) commitments on the other scenarios such that the country’s development takes into account climate change, and climate mitigation as well as resilient growth are pursued. This assessment of climate change impacts is done by first selecting a representative set of climate scenarios, used to assess the macroeconomic effects of climate change. Macroeconomic shocks arising from relevant “channels of impact” under climate change are then explored, with these shocks serving as input for a country-specific macroeconomic model. With this chapter introducing the project, chapter 2 goes on to provide an overview of the analytical approach used to estimate economic damage from climate change. Chapter 3 presents results for each impact channel assessed and chapter 4 presents a summary of the key findings as well as conclusions. Appendixes A and B present details on the methodologies used for the development of climate scenarios and impact channels, respectively. Appendix C documents the data sources used to conduct the analysis. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  1 2. Methodological Overview As introduced above, this study aims to estimate the economic damage caused by climate change to Tanzania’s economy, with these estimates to be used in the development of the CCDR. The analysis is built around different “impact channels” through which climate change will result in shocks to the country’s macroeconomy. This study considers the following different types of shocks, namely shocks to: • Human capital, including heat and labor productivity, and human health impacts on labor supply. • Natural capital, including from changes in crop production, livestock production, and soil erosion. • Physical capital, including roads and bridges and from inland flooding. This chapter presents the overarching analytical framework used to model impact channels and their linkages to the macroeconomic model, the climate scenarios considered, and the development and adaptation policy scenarios evaluated. Appendix A presents further detail on the methods used to process climate data and select scenarios, while appendix B presents detailed methodological information for each individual impact channel. Appendix C documents the various sources of data used in the analysis. 2.1. Analytical approach Within this study’s analytical framework, developing impact channels involves four stages (figure 1): 1. Obtaining gridded historical and projected climate data for a set of climate scenarios. 2. Selecting, tailoring, and/or developing biophysical models that convert changes in climate data into biophysical shocks for each of the impact channels evaluated for the country. 3. Aggregating grid-level biophysical shocks to national and/or sectoral scales using high-resolution geospatial data. 4. Producing shocks to be fed into the country’s macroeconomic and poverty model(s). Results are aggregated either to national scale inputs, such as capital or labor, or to economic sectors, such as agriculture, to match the macroeconomic model’s resolution. Figure 1: Impact channel modeling approach 2  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania For this analysis, we consider eleven channels of impact. Table 1 provides a high-level description of each channel broken down by the categories introduced above (human capital, natural capital, and physical capital). The shocks caused by climate change through each impact channel are calculated based on changes in climate variables—such as monthly precipitation or daily maximum temperature—for the 30-year period from 2021 to 2050, relative to a historical climate baseline from 1995 to 2020. These shocks are then used as input to the existing country macroeconomic model. Table 1: Overview of impact channels evaluated Name of channel Description of how climate change translates to damage Human capital 1 Heat and labor productivity Shock to labor productivity from daily heat stress to both indoor and outdoor workers. Considers occupation-specific work ability curves from the International Labor Organization. 2 Human health Shock to labor supply from changes in the incidence and mortality of vector-borne (malaria and dengue), waterborne (i.e., diarrheal), and temperature-related diseases. Natural capital 3 Crop production Shock to rainfed crop revenues through changes in yields. Based on the Food and Agriculture Organization’s (FAO) crop-specific yield response functions to water availability and heat stress. 4 Livestock production Shock to livestock revenues through changes in productivity by animal and product type. Considers extreme heat and feed availability effects through animal-specific impact curves. 5 Soil erosion Shock to crops from topsoil erosion. Impacts on erosivity from changes in rainfall are based on the Revised Universal Soil Loss Equation (RUSLE) model. Physical capital 6 Inland flooding Shock to capital from changes in the recurrence of peak precipitation events that result in fluvial (riverine) flooding. Models streamflows and floodplains, with damage estimated using depth-damage curves. 7 Bridges Shock to capital due to damage to and increased maintenance of bridges from changes in the recurrence of peak precipitation events that result in fluvial (riverine) flooding. 8 Roads Shock to capital due to damage to and increased maintenance of roads, as modeled using the Infrastructure Planning Support System model. Also considers labor supply effects of road disruptions. As summarized in table 1, individual impact channels rely on stylized biophysical models that are capable of accepting climate information and projections, and simulating changes in biophysical (such as streamflow or infrastructure conditions) and/or socioeconomic (such as labor supply hours) variables under these altered climatic conditions. The biophysical models are customized to the country context by using country-specific inputs, obtaining key assumptions from country experts and available literature, and calibrating outputs using local data. Where locally collected data are not available, we rely on global sources. Scenarios that consider different possible development and adaptation policy decisions and investments are captured in the modeling by modifying inputs and assumptions. Detailed descriptions of the impact channel methodologies, including the models used, resolution of the analysis, and key assumptions and limitations are presented in appendix B, while appendix C documents the specific data sources utilized in the impact channel analysis. The outputs of these biophysical models are then translated into inputs to a Computable General Equilibrium (CGE) model. These models are used to simulate the effects of policy changes and external shocks on the economy as a whole. CGE models incorporate data on the interactions between different sectors and agents in the economy and can be used to analyze the distributional impacts of policy changes on different groups. The economy is then divided into different sectors, such as agriculture, manufacturing, Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  3 and services. These sectors are further divided into subsectors, depending on the level of detail required for the analysis. The Social Accounting Matrix is an important component of the model, used to represent the structure of the economy being studied and containing information on the flows of goods and services between different sectors, households, and the government. Typically, the Social Accounting Matrix would consider multiple sectors—for example, agriculture, industry, and services—as well as labor and capital as factors of production. These variables are shocked by results from the biophysical models. CGE models can be used to extract poverty-related outputs. These models can estimate changes in household income, household consumption, employment, and welfare measures—such as poverty headcounts or gaps— that can be used to assess the impact of different policies on poverty and inequality. Poverty microdata is generally obtained from household surveys, which provide information on the characteristics and behavior of households. These data are used for (1) calibrating and validating the parameters used in the biophysical models and (2) calculating household-level shocks derived from the impact channels. 2.2. Development and adaptation scenarios Given the large degree of uncertainty associated with future conditions, the biophysical modeling conducted for each impact channel considers different possible future development and adaptation scenarios that vary in terms of socioeconomic development, sectoral expansion, and adaptation efforts. As such, we consider the following scenarios: • A BAU scenario assumes that Tanzania’s development continues to follow recent historical averages, projecting current growth trends in the country out to 2050. The economy is assumed to undergo little by way of structural transformation nor are significant investments made to combat climate change. • An ASP scenario assumes higher levels of economic growth and development are achieved over the coming decades, resulting in structural transformation of the country’s economy, in line with socioeconomic development objectives. • A BAU + climate action scenario assumes the same socioeconomic development outcomes as the BAU scenario, but the country’s development considers climate change. Climate change adaptation as well as resilient growth are pursued. • An ASP + climate action scenario assumes the same socioeconomic development outcomes as the ASP scenario, but the country’s development considers climate change. Climate change adaptation as well as resilient growth are pursued. When it comes to the scenarios that consider the country’s NDC, there is explicit consideration of investment in climate adaptation interventions, with these interventions aiming to reduce the negative effects of climate change experienced through a particular channel. The interventions considered under the NDC scenario are channel-specific and were selected based on modeling feasibility, applicability to the specific country context, and the degree of projected climate change impacts. Adaptation costs are based on unit cost estimates obtained from international and, where available, local sources. The modeling assumptions for individual adaptation measures are detailed in the corresponding impact channel sections provided in chapter 3. 2.3. Climate information 2.3.1. Background and historic climate Tanzania is in east Africa, bordering the Indian Ocean (figure 2). Its legislative capital, Dodoma City, is located in the center of the country, with its former capital, Dar es Salaam, located on the east coast. Tanzania is bounded by Kenya, Uganda, Rwanda, Burundi, Democratic Republic of Congo, Zambia, Malawi, and Mozambique. 4  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 2: Country map Source: Worldometer. Tanzania is characterized by very varied topography. The East African Rift System cuts through mainland Tanzania in a north-south direction, resulting in narrow depressions that form lakes, such as Lake Tanganyika on the Western Rift Valley branch and Lake Nyasa on the Eastern Rift Valley branch. A central plateau stretches between these two rift valleys. The southwestern corner of the country is comprised of highlands, which stretch across the country towards the northeast, culminating in Mount Kilimanjaro, Africa’s highest mountain. The country is home to many large as well as smaller inland lakes, notably Lake Victoria, on the border with Uganda and Kenya. Rivers in the country include the Rufiji, Ruvuma, Wami, and Pangani, all of which drain into the Indian Ocean. Figure 3 shows the country’s Köppen-Geiger climate classification. The majority of the country experiences a tropical savanna climate, with generally warm temperatures and a clearly defined wet and dry season. Central regions of the country are dominated by a hot semi-arid climate, which is characterized by hot summers and warm to cool winters, with some to minimal precipitation. Localized regions in the northeast and southwest experience either a monsoon-influenced humid tropical climate with dry winters and hot summers or a subtropical highland climate with dry winters and warm summers. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  5 Figure 3: Kóppen-Geiger climate classification map, 1991–2020 Am Tropical monsoon climate As/Aw Tropical savanna climate BSh Hot semi-arid climate Cwa Monsoon-influenced humid tropical climate Cwb Subtropical highland climate Source: World Bank 2021. Rainfall in Tanzania varies from around 550 millimeters per year in the drier central parts of the country to more than 3,600 millimeters per year in portions of the southwestern highlands. The country experiences two rainy seasons, from October through December and from March to May (World Bank 2021). Figure 4 shows the spatial variability of mean average precipitation and temperature across the country. Generally, precipitation is concentrated in the southeast and northwest of the country, with more central regions characterized by lower precipitation totals. The temperature is generally highest in the southeastern portion of the country. Figure 4: Mean precipitation and temperature by 0.5-degree grid cell, 1995–2020 a. Precipitation b. Temperature 6  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 5 shows the monthly variability of precipitation and temperature for the period from 1991–2020. Monthly average temperatures ranged between 20.6˚C in July and around 24˚C from December to March, while monthly average precipitation varied from 5 millimeters in June to 166 millimeters in March. Figure 5: Monthly mean temperature and precipitation, 1995–2020 32°C 180 mm 28°C 150 mm 24°C 120 mm Precipitation Temperature 20°C 90 mm 16°C 60 mm 12°C 30 mm 8°C 0 mm Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Average Minimum Surface Air Temperature Average Mean Surface Air Temperature Average Maximum Surface Air Temperature Precipitation Source: World Bank 2021. 2.3.2. Selection of future climate scenarios Future climate is inherently uncertain, due to variability in the earth’s physical responses, uncertainty in future greenhouse gas (GHG) emissions trajectories, as well as uncertainty across different climate model projections for the coming decades. As such, a total of eight climate scenarios were selected from among a larger set of available scenarios to explore the impacts under a range of possible future climatic conditions. Available climate scenarios were first obtained from the World Bank’s Climate Change Knowledge Portal for 29 General Circulation Models (GCMs) from the Coupled Model Intercomparison Project 6 (CMIP6) suite of model outputs. Each GCM has up to five combinations of Shared Socioeconomic Pathway (SSP) and Representative Concentration Pathway (RCP) emissions scenario runs available. For each GCM-SSP combination, a modeled history from 1995 to 2014 and projections from 2015 to 2100 were available, for monthly mean temperature and precipitation at a 1x1 degree grid resolution. Given that GCM output is biased relative to observed climate conditions, bias-correction and spatial disaggregation was conducted, before then interpolating monthly variables to a daily timestep. The resulting data is available for a 0.5-degree resolution consistent with the Climate Research Units (CRU) from the University of East Anglia. Appendix A provides further details on the climate inputs utilized as well as the bias-correction, spatial and daily interpolation processes. To ensure accuracy in capturing regional climate trends, baseline monthly precipitation and maximum and minimum temperatures used in this study were compared with data from the Tanzania Meteorological Authority (TMA) for 20 weather stations around the country. Overall, the majority of weather stations compared favorably to CRU data. Larger differences between the datasets were found at weather stations located at higher or lower elevations than the surrounding landscape. The high elevations of these stations, such as Kilimanjaro, Moshi, and Sumbawanga stations, are not captured by the CRU data, which represents the average climate from a wider 0.5-degree grid area. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  7 Figure 6 shows that the precipitation differences from weather stations to CRU are limited, not exceeding a 20 percent deviation, and generally restricted to stations with a large elevation difference relative to the 0.5-degree grid cell. Trends are similar across maximum and minimum temperatures, where CRU also shows differences in prediction at coastal weather stations. Since CRU represents weather at a 0.5-degree grid level, it is unable to account for localized coastal climates, therefore minimum temperatures (figure 8) are slightly different between the datasets at the islands of Zanzibar and Pemba, and maximum temperature (figure 7) varies at the Bukoba station on Lake Victoria and Sumbawanga near Lake Rukwa. Despite these small, localized differences, both CRU and TMA data show similar regional climate patterns. These regional patterns and trends are the major factors for this climate analysis. Figure 6: Annual precipitation between TMA and CRU data, 1995–2020 Weather Station 3.50 Arusha Bukoba Dar es salaam Dodoma 3.25 Iringa Kigoma Weather Station Log Precipitation Kilimanjaro 3.00 Morogoro Moshi Mtwara Musoma 2.75 Mwanza Pemba Same 2.50 Shinyanga Songea Sumbawanga Tabora 2.25 Zanzibar 2.25 2.50 2.75 3.00 3.25 3.50 CRU Log Precipitation 8  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 7: Monthly maximum temperature between TMA and CRU data, 1995–2020 Weather Station 33 Arusha Bukoba Dar es salaam Dodoma Iringa 30 Weather Station Maximum Temperature Kigoma Kilimanjaro Morogoro Moshi Mtwara 27 Musoma Mwanza Pemba Same Shinyanga 24 Songea Sumbawanga Tabora Zanzibar 24 27 30 33 CRU Maximum Temperature Figure 8: Monthly minimum temperature between TMA and CRU data, 1995–2020 Weather Station 25 Arusha Bukoba Dar es salaam Dodoma Iringa Weather Station Maximum Temperature Kigoma 20 Kilimanjaro Morogoro Moshi Mtwara Musoma Mwanza Pemba 15 Same Shinyanga Songea Sumbawanga Tabora Zanzibar 15 20 25 CRU Maximum Temperature Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  9 Figure 9 shows the projected mean temperature (panel a) and precipitation (panel b) in the country through 2100 across a range of SSP-RCP combinations. The bold lines are averages across GCM projections for each of the four SSP-RCPs, and the shaded zones surrounding those lines are the full range of GCM projections within an RCP. As can be seen, while GCM ensemble averages for precipitation (the bold lines in the right panel) do not change significantly relative to historical baseline precipitation, the precipitation projected across the full range of GCMs (the shaded zones in the right panel) varies widely. This emphasizes the importance of selecting a set of climate scenarios that capture a wide range of possible future conditions. Figure 9: Projected climate variables across a range of SSP-RCPSs a. Mean temperature b. Precipitation Projected Average Mean Surface Air Temperature Projected Precipitation Tanzania; (Ref. Period: 1995-2014), Multi-Model Ensemble Tanzania; (Ref. Period: 1995-2014), Multi-Model Ensemble 32 1800 30 1600 28 1400 26 1200 24 100 22 800 20 600 1960 1980 2000 2020 2040 2060 2080 2100 1960 1980 2000 2020 2040 2060 2080 2100 Hist. Ref. Per., 1950-2014 SSP1-2.6 SSP2-4.5 Hist. Ref. Per., 1950-2014 SSP1-2.6 SSP2-4.5 SSP3-7.0 SSP5-8.5 SSP3-7.0 SSP5-8.5 Source: World Bank 2021. Following World Bank (2022a), two of the eight scenarios included in this study were selected to allow for comparisons across emissions scenarios; these are referred as mitigation scenarios. The guidance specifies the following two mitigation scenarios: • Ensemble average of SSP3–7.0 GCMs: Pessimistic case; scenario in which warming reaches 4°C by 2100, due to lax climate policies or a reduction in ecosystems and oceans’ ability to capture carbon. • Ensemble average of SSP1–1.9 GCMs: Optimistic case; represents reductions in GHG emissions in line with limited 1.5°C of warming by 2100. In addition to enabling comparison across emissions scenarios, climate scenarios were also selected in such a way as to capture the broadest range of climate change effects across GCMs. In doing so, the vulnerability of the economy and the performance of adaptation options under possible wet vs. dry and hot vs. warm GCM outcomes can be assessed. We selected the following set of scenarios, based on changes from the historical baseline climate as compared to the period between 2051 and 2060: • Dry/hot scenarios: Three scenarios around the 10th percentile of mean precipitation changes (dry) and the 90th percentile in mean temperature changes (hot), across SSP2–4.5 and SSP3–7.0 GCMs. Final channel results will also include the average climate impact of the three dry/hot GCMs selected. • Wet/warm scenarios: Three scenarios around the 90th percentile of mean precipitation changes (wet) and the 10th percentile in mean temperature changes (warm), across SSP2–4.5 and SSP3–7.0 GCMs. Final channel results will also include the average climate impact of the three wet/warm GCMs selected. 10  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania The selected climate scenarios are summarized in table 2, with further detail provided in appendix A. As noted, mitigation scenarios are relevant for comparing channel results across scenarios of different global mitigation efforts, while dry/hot and wet/warm scenarios are appropriate for exploring individual realizations of climatic conditions, providing a low- and high-end estimate of impacts. Given that the differences between the two mitigation scenarios are relatively minor in the time period leading up to 2050, the results presented in this report focus primarily on the projected impacts under the dry/hot and wet/warm futures. Table 2: Selected climate scenarios Type Scenario Mitigation SSP1–1.9 mean SSP3–7.0 mean Dry/hot future SSP2–4.5 CNRM-ESM2-1 SSP3–7.0 EC-EARTH3-VEG SSP2–4.5 GFDL-ESM4 Wet/warm future SSP2–4.5 EC-EARTH3 SSP2–4.5 INM-CM4-8 SSP2–4.5 CMCC-ESM2 These same climate scenarios are utilized across all impact channels except for the following: • The inland flooding and bridges analyses rely on data for the peak single-day precipitation magnitude and frequency, rather than mean precipitation volumes. Since these data are only available for SSP ensemble aggregates in the Climate Change Knowledge Portal, we do not consider dry/hot or wet/warm futures and instead perform the analysis for SSP1–1.9, SSP2–4.5, and 3–7.0 median (50th percentile) results. • The roads channel evaluates impacts under 55 SSP2–4.5 and SSP3–7.0 climate futures (that is, all SSP2–4.5 and SSP3–7.0 GCMs contained in the Climate Change Knowledge Portal). By running this large set of future scenarios, we can characterize both emissions uncertainty (SSPs) and climate model uncertainty (through different GCMs). 2.3.3. Expected changes by 2050 Figure 10 shows the projected change in mean temperature throughout the country in the period from 2041–50 relative to 1995–2020 for the selected dry/hot and wet/warm climate scenarios (previously shown in figure 4). Generally, projected changes in temperature are relatively uniform across the country for each of the different climate scenarios. As expected, changes in temperature relative to 1995–2020 are expected to be greatest under the selected dry/hot GCMs (shown in figure 10a), with changes in temperature peaking at around 1.6˚C across portions of the country. Table 3 presents the change in average temperature by decade throughout the country under the dry/hot mean and wet/warm mean scenarios relative to historical baseline conditions. Overall, we expect temperatures to increase in each decade relative to historical conditions, with average temperatures peaking by the 2040s. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  11 Figure 10: Change in mean temperature from 2041–50 relative to 1995–2020, by climate scenario a. Dry/hot climate scenario b. Wet/warm climate scenario Table 3: Change in average national temperature by decade relative to 1995–2020 Scenario 2020s 2030s 2040s Dry/hot mean +0.57˚C +0.85˚C +1.23˚C Wet/warm mean +0.34˚C +0.62˚C +0.82˚C Figure 11 shows the projected change in mean precipitation throughout the country in the period from 2041–50 relative to 1995–2020 for the selected dry/hot and wet/warm climate scenarios (previously shown in figure 4). Under the selected dry/hot scenarios (panel a), changes to mean precipitation vary by climate projection, with most showing drying across most of the country, though some wetting in the north appears possible. Under the selected wet/warm scenarios (panel b), we expect wetter conditions across most of the country, with increases of more than 20 percent possible. Table 4 presents the change in average precipitation by decade throughout the country under the dry/hot mean and wet/warm mean scenarios relative to historical conditions. Generally, we expect precipitation to increase in each decade under the wet/warm mean scenario, with declines in average national precipitation possible in the 2020s and 2040s under the dry/hot mean scenario. 12  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 11: Change in mean precipitation from 2041–50 relative to 1995–2020, by climate scenario a. Dry/hot climate scenario b. Wet/warm climate scenario Table 4: Percentage change in average national precipitation by decade relative to historical baseline Scenario 2020s 2030s 2040s Dry/hot mean -1.5% +1.1% -4.3% Wet/warm mean +4.9% +5.9% +11.8% Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  13 3. Impact Channel Results This chapter presents the results of the impact channel assessment by which climate change damage is estimated. For each impact channel considered, we first provide a brief overview in table 5, which summarizes key findings for this channel. This is followed by a description of how the channel was modeled, with further methodological detail available in appendix B and data sources described in appendix C. Finally, the results are presented across relevant development scenarios. 3.1. Human capital Climate change can negatively impact human capital in a variety of ways. Increased heat stress to workers can reduce labor productivity, while the increased incidence of infectious diseases can reduce labor supply. We use a labor supply model to calculate the total labor hours in the country under different future conditions, with the estimated impacts to Tanzania’s human capital due to climate change presented below. 3.1.1. Heat and labor productivity As of 2022, Tanzania’s total gross domestic product (GDP) was estimated at around $76 billion (World Bank 2022b). Significant areas of economic growth include transportation, infrastructure, communications, manufacturing, electricity, trade, tourism, real estate, and business services (International Trade Administration 2022c). However, over 70 percent of the country’s population is rural, where meaningful economic growth has been less substantial due to low productivity changes in the agricultural sector (USAID, n.d.). Agriculture makes up approximately 64 percent of the country’s labor force, representing the largest share of employment relative to the services and industry sectors at 28 and 7 percent of the labor force, respectively (ILOSTAT 2022a, 2022b, 2022c). Agricultural sector improvements are a priority, with the government’s transformation strategies and budget bolstering a target of 10 percent agricultural sector growth by 2030 (UNDP 2023). In 2020, Tanzania was designated as a lower-middle-income country, an improvement from its previous low-income status. The country’s industrialization objectives aim at achieving middle-income economy status by 2030, highlighting a focus on industry sector improvements as well. Overview of impact channel Climate change can impact labor productivity by increasing workday temperatures and decreasing the number of hours an individual can perform work. To estimate labor heat stress impacts due to climate change, we calculate workday wet bulb globe temperatures as a measure of heat stress to derive labor productivity loss curves across three levels of physical activity. Our quantification of wet bulb globe temperatures assumes that relative humidity and solar radiation will remain constant into the future and that all occupations will be performed evenly throughout the year. A detailed description of this method and its limitations are presented in appendix B. For each sector, the labor productivity shocks are based on the relative labor hours by occupation and the extent to which these occupations are exposed to outdoor work. Employment by occupation and sector is obtained from the 2021 National Labor Force Survey reported in the International Labor Organization’s Labor Force Statistics and summarized in able 5 for groups of occupations. Both agriculture and industry have a large proportion of skilled and unskilled workers who perform activities of high physical activity. Total labor hours are then calculated based on average weekly hours worked by sector, which in Tanzania are 40.6 hours per week for agriculture, 46.2 hours per week for industry, and 53 hours per week for services. 14  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Table 5: Proportion of occupations per sector Occupation group Agriculture Industry Services Managers and professionals 0% 0% 0% Service, sales, and clerical support workers 0% 2% 41% Skilled workers 98% 72% 26% Elementary occupations 2% 25% 33% Technicians and associate professionals 0% 0% 0% Outdoor exposure by occupation is estimated from the Occupational Requirements Survey from the United States Bureau of Labor Statistics by occupation (BLS 2022). Elementary occupations are the most exposed to outdoor conditions at 94 percent outdoor exposure; skilled occupations range between 50 to 81 percent outdoor exposure; and all other occupations range between 5 to 30 percent outdoor exposure. While both indoor and outdoor workers are exposed to heat stress, outdoor workers are generally exposed to higher wet bulb globe temperatures at full sun exposure. In contrast, a portion of indoor workers are assumed to perform work in air-conditioned environments. For Tanzania, this is estimated at 1.1 percent of the indoor workforce, based on World Bank analyses conducted using the Household Budget Survey. Overall, the effect of heat stress intensifies for labor types that are outdoors and conducting more intense physical work. The total impacts to labor productivity are estimated for the agriculture, service, and industry sectors for different development scenarios, as summarized in table 6. We use available data on labor supply, relative heat and air conditioning coverage, and labor force distribution. The specific data sources are listed in appendix C. Table 6: Development scenario assumptions for the labor heat stress channel BAU ASP BAU+ climate action ASP+ climate action Assumes the current Assumes mechanization of the agriculture Assumes increased Assumes both the sectoral breakdown of the sector increases from current levels (0.2%) to investment in air conditioning mechanization of the workforce does not change 30% by 2050. As a result, a portion of labor- from current coverage (1.1%) agriculture sector from in the decades to 2050, intensive occupations in agriculture shift to other to 30% coverage of workers ASP as well as increased with limited mechanization occupations with lower intensities and lower for the industry and services investment in cooling from and cooling being realized. exposure to outdoor temperatures. sectors by 2050. Costs of BAU+ climate action. adaptation are summarized Changes in the labor force mix are not with results. considered. This measure is assumed to be cost-neutral. Results Table 7 summarizes the share of indoor and outdoor workers by sector for the different development scenarios. Mechanization of agriculture results in elementary occupations shifting to skilled occupations, which have lower outdoor exposure. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  15 Table 7: Outdoor workforce exposure by sector and development scenario Development scenario Sector Indoor Outdoor BAU and Agriculture 19% 81% BAU + climate action Industry 67% 33% Services 40% 60% ASP and Agriculture 34% 66% ASP + climate action Industry 67% 33% Services 40% 60% In the historical baseline period (1995–2020), we observe a mean maximum workday wet bulb globe temperature in the country that ranges from about 20°C to 30°C. Generally, across all levels of physical intensity, work ability reductions are only noticeable above 25°C. At 32°C, maximum work ability for low, medium, and high physical activity corresponds to 83, 78, and 75 percent (of a full hour of work) for indoor workers. For outdoor workers, maximum physical activity at the same temperature is 75, 63, and 52 percent, respectively. These percentages represent a multiplier that captures the percentage of an hour that laborers can perform work at a certain temperature. The results below describe labor productivity shocks relative to a historical baseline period (1995–2020) by sector until 2050. Across macroeconomic sectors, agriculture is expected to experience the highest labor productivity shock due to elevated temperatures by 2041–50 under the BAU scenario, followed by industry and services, respectively (figure 12). The shock to agriculture ranges from -2.5 to -4.1 percent under BAU and BAU + climate action conditions. These shocks are reduced to between -0.8 and -2.2 percent under the ASP and ASP + climate action scenarios due to the increased mechanization of agricultural labor, which lessens exposure to the rising temperatures anticipated in Tanzania. Impacts on agriculture of such mechanization are significant because agriculture has a high proportion of both skilled and unskilled workers that perform high physical intensity activities, which are also more heavily exposed to the projected higher temperature outdoor conditions. Impacts on the industry and services sectors are projected to range between approximately -1.1 and -3.4 percent under BAU, and do not experience any changes under the ASP scenario. However, investment in increased cooling technologies under the BAU + climate action and ASP + climate action scenarios sees decreased shocks for these sectors, with industry and services experiencing residual shocks between -0.9 and -2.1 percent, depending on the sector and climate scenario. Figure 12: Labor productivity shocks, 2041–50 BAU ASP BAU with climate action ASP with climate action 0.0% -1.0% -2.0% -3.0% -4.0% -5.0% Agriculture Industry Services Agriculture Industry Services Agriculture Industry Services Agriculture Industry Services Dry/Hot mean Wet/Warm mean Dry/hot GCM range Wet/warm GCM range 16  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 13 shows the aggregated results across all three sectors. Note that these aggregated results do not consider changes in the sectoral breakdown over time. Generally, shocks from the dry/hot mean scenario are higher than under the wet/warm mean scenario. The spread between the wet/warm and dry/hot mean climate scenarios gradually increases over time under all three of the development scenarios considered. By mid-century, the wet/warm and dry/hot mean climate scenarios result in a -2.3 percent and -3.7 percent shock, respectively under BAU. As a result of agriculture mechanization adaptations under the ASP scenario, adoption of air conditioning under the BAU + climate action scenario, and both under the ASP + climate action scenario, the impacts of projected wet bulb globe temperatures on productivity are lessened to shocks between -1.2 percent to -2.4 percent, -2 to -3.3 percent, and -0.9 to -2.1 percent, respectively. These results would suggest that a structurally reformed economy where fewer people are employed in labor-intensive outdoor jobs and cooling technology is more widely available is less vulnerable to expected labor heat stress shocks from climate change in Tanzania. Figure 13: Labor productivity shocks, 3-year moving average BAU ASP BAU with climate action ASP with climate action 1.0% 0.0% -1.0% -2.0% -3.0% -4.0% -5.0% 2020 2030 2040 2050 2020 2030 2040 2050 2020 2030 2040 2050 2020 2030 2040 2050 Dry/Hot mean Wet/Warm mean Dry/Hot GCM range Wet/Warm GCM range While anticipated labor force migration from populous eastern coastal regions like Dar es Salaam and Zanzibar towards the Lake Victoria Basin could impact the labor productivity effects from climate change (box 1), this analysis does not consider migration dynamics. Box 1: Migration in Tanzania The World Bank Group report “Groundswell: Preparing for Internal Climate Migration” (Rigaud et al. 2018) projects that internal climate migration throughout Sub-Saharan Africa and other regions will increase by 2050 due to impacts such as water stress, crop failure, and sea level rise. Out-migration regions are areas expected to be increasingly susceptible to climate impacts and will likely experience resulting declines in population growth. In-migration hotspots are locations with more suitable climatic conditions for agriculture and economic opportunity. In Tanzania, out-migration areas include the eastern coast near Dar es Salaam, where deteriorating water availability and crop yields combined with sea level rise, storm surges, and declining land availability are anticipated to increase migration towards the Lake Victoria basin and its surrounding areas. These in- migration regions are projected to experience comparatively less severe climate impacts, though existing high-density populations could pose additional resource challenges as migrants relocate. This migration has the potential to significantly impact labor force distributions of the country. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  17 Figure B1.1: Migration in Tanzania, 2030–50 Costs of adaptation The BAU + climate action and ASP + climate action development scenarios consider increased investment in suitable cooling technologies for the industry and services sectors, reaching 30% coverage of workers by 2050. In this case, the cooling technology considered is the installation of air-conditioning systems, which are assumed to only be run to ensure workability in indoor spaces, not for comfort. Adaptation costs include the capital cost of the new air-conditioning units required each year (which is driven by labor force growth based on average population prospects and increase in air conditioning coverage), as well as annual energy consumption costs. The capital costs per unit are assumed to be the average of low-end and high-end air- conditioning systems—that is, a window unit and split system. In workspaces, one unit is assumed to cover the cooling needs for as many people as are in two average households. Capital costs are annualized over 15 periods at a 3 percent discount rate. Total costs (capital and operating costs) due to an increase to 30 percent of air conditioning coverage for workers by 2050 are expected to reach around $5.85 million annually by 2041–50. From 2024–50, these costs total $90.9 million, of which $28.3 million are capital expenditures and $62.6 million are operational expenditures. Mechanization of the agriculture sector as an adaptation measure under the ASP and ASP + climate action scenarios is not costed in this analysis. 18  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Discussion and key takeaways The impacts to labor productivity from heat stress are relatively high in Tanzania due to its hot climate, even in the historical baseline. While more moderate labor productivity effects are projected for the industry and services sectors, agriculture is consistently more severely impacted than other sectors due to a higher proportion of workers performing high physical activity tasks, as well as higher outdoor temperature exposure. Agriculture’s large share of the workforce in Tanzania results in the sector shock heavily affecting national shocks. That said, the industry sector in Tanzania is also likely to experience larger labor productivity shocks from heat stress relative to similar countries; this is primarily due to industry labor being concentrated on the coast near Dar es Salaam, an area subject to hot and humid conditions. The analysis presented above represents the shocks to the formal labor sector. While the analysis did consider the impacts on informal workers, the average split between indoor and outdoor workers was not found to be different than in the formal sector, therefore resulting in negligible differences. Under the ASP and ASP + climate action scenarios, the projected impacts on agriculture are expected to decrease by about half due to mechanization of the sector, though the cost of this transition are not considered in this analysis. Increased adoption of air conditioning in the industry and services sectors under the BAU + climate action and ASP + climate action scenarios would lessen their respective shocks, especially for the higher-intensity industry sector. However, in Tanzania today, access to air conditioners is minimal, and significant increases in coverage will be needed to reduce heat stress at the national scale. Fans and other cooling systems are cheaper than air conditioning, but also less effective in terms of reducing heat stress. Furthermore, while not considered in this analysis, increases in the number of workers in industry and service jobs over time and the migration of coastal populations inland towards Lake Victoria could further reduce the aggregate effect of heat stress on labor productivity. 3.1.2. Human health The most significant communicable diseases in Tanzania as of 2019 are lower respiratory infections, HIV/ AIDS, tuberculosis, malaria, and diarrheal diseases (IHME 2019). Notable noncommunicable diseases with high mortality burdens include stroke, heart disease, and liver cirrhosis. These leading health concerns are observable among inpatient hospital deaths from 2006–15, where mortalities were primarily attributable to malaria, respiratory diseases, HIV/AIDS, anemia, and heart diseases (Mboera et al. 2018). Tanzania also has one of the highest maternal and child mortality rates in the world, with an estimated 54 deaths per 1,000 people aged under five years (Odjidja, Gatasi and Duric 2019). Historically, Tanzania has experienced recurring urban dengue virus outbreaks in the peak season from May to June, most recently in 2019 (Kajeguka et al. 2023). Additionally, Tanzania’s ambient temperatures of greater than 30°C result in heat-related illnesses for both exposed workers and urban populations (Meshi et al. 2018; Adegun, Mbuya and Njavike 2022). Table 8 presents the historical death and incidence rates for the diseases evaluated in this analysis. Table 8: Historical death and incidence rates by disease Cause Death rate, 2019 (per Incidence rate, 2019 (per Percentage change in Percentage change in 100,000 people) 100,000 people) deaths, 2009–19 incidence, 2009–19 Total noninjury 584 568,338 -33.6% 1.2% Malaria 30 12,159 -32.6% -21.7% Dengue 0.004 413 12.5% -23.3% Heat-related 0.7 649 -36.2% -2.7% Waterborne 27.7 99,838 -39.7% 8.2% Note: All estimates are based on 2019 Institute for Health Metrics and Evaluation data. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  19 Overview of impact channel Climate change may impact human health through increased incidence of and deaths from vector-borne diseases such as malaria and dengue, heat-related diseases, and waterborne infectious diseases that cause acute diarrhea, which all influence the total labor supply. We apply different biophysical and statistical relationships between climate variables and the incidence of or transmissibility for each disease, with changes in disease incidence and death rates then used to estimate the number of hours of labor supply lost. A summary of the methodology used is presented below, with a more detailed description available in appendix B. When considering vector-borne diseases, the same general methodology is applied to both malaria and dengue. These are both mosquito-borne diseases whose spread depends on the right environmental conditions occurring for the mosquitoes to live, breed and increase in number. These conditions are approximated from three climate variables, namely mean monthly temperatures, cumulative annual precipitation, and minimum annual winter temperature, and are used to calculate a suitability index ranging from 0 (unsuitable) to 1 (suitable). Heat-related illnesses are modeled based on calculating excess mortality from daily maximum temperatures. The temperature–mortality relationship is assumed to be V-shaped, and the temperature value at which mortality is lowest is defined as the optimum temperature. For temperatures above the optimum threshold for a given location, excess heat-mortality is defined daily as a fraction of the average total noninjury-related deaths occurring that day. For waterborne diseases, we follow the modeling approach described by the World Health Organization (WHO 2014), which applies gridded estimates of average annual temperature anomalies to a statistical temperature–mortality risk relationship. The approach calculates the total estimated diarrheal deaths and cases in the future without climate change, and then estimates the climate change-attributable percent change. The expansion of water, sanitation and hygiene (WASH) services, which can reduce transmission of these diseases, is also considered in the analysis. We follow the methodology applied by Wolf et al. (2019), which is based on a statistical relationship between a fecal contamination composite index (FAECI) and the relative risk of diarrheal diseases. Finally, the changes in the incidence and death rates of the different types of diseases described above are then used to model the number of hours of labor supply lost. Additional deaths relative to the historical baseline and absenteeism from work due to people falling sick directly reduce the total labor available. Additionally, there is also an indirect effect of children getting sick and needing parental care for the duration of the disease. The total hours of labor lost for each disease are then calculated for the country for both historical periods as well as future projections. The specific data sources used to complete the analysis are detailed in appendix C. The total impacts to labor supply are estimated for a baseline, BAU, BAU + climate action, and ASP scenario. The baseline scenario assumes a continuation of the current death and incidence rates for malaria, dengue, waterborne, and heat-related diseases, with changes driven only by changing climatic conditions. However, the remaining scenarios consider reductions in mortality and morbidity rates for waterborne diseases based on improvements to the country’s WASH metrics. Table 9 summarizes the assumptions made for this impact channel, with the specific data sources detailed in appendix C. The results represent a percent shock to total labor supply, with the shock considering changes in the death and incidence rates of dengue, waterborne disease, and heat-related illness. 20  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Table 9: Development scenarios evaluated for the human health channel BAU BAU+ climate action ASP ASP+ climate action Assumes a continuation of current disease Assumes achievement of Assumes achievement of SDG Same assumptions death and incidence rates (per 100,000 National Five-Year Plan 6 targets for universal water, as ASP. people) by 2050 for malaria, dengue, and 2021–25 targets for: sanitation, and hygiene by 2050: heat-related illnesses. Rural safely managed drinking Sanitation: Basic 100%/safely Assumes a 10% reduction in waterborne water: 85% managed 100%/open defecation disease rates resulting from a continuation 0%/community coverage with basic of historical WASH coverage increase trends, Urban safely managed drinking 48% which consider and increase from: water: 95% Water: Basic 100%/safely managed Sanitation: Basic 31%/safely managed Rural improved sanitation: 100% 25%/open defecation 6% 75% Hygiene: Basic 100%/with soap Water: Basic 61%/safely managed 11% Resulting reductions in 100% waterborne disease rates of Hygiene: Basic 63%/with soap 29% 26%. Resulting reductions in waterborne disease rates of 77%. To: Sanitation: Basic 50%/safely managed 39%/open defecation 0% Water: Basic 100%/safely managed 20% Hygiene: Basic 59%/with soap 32% Results Figure 14 shows the breakdown of mortality and morbidity for the different diseases considered, for the historic period 2015 to 2019. Of the four disease categories evaluated, waterborne diseases were responsible for the majority of deaths and cases. Malaria comprised the second largest shares, while deaths and cases attributable to dengue and heat-related illness were substantially lower. Figure 14: Breakdown of disease mortality and morbidity for 2015–19 20% 2.3% 0.1% 15% 10% 5.0% 17.3% 0.0% 5% 5.0% 0.1% 0.1% 0% Deaths Incidence Malaria Dengue Water-borne Heat-related Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  21 Table 10 shows how the mortality and morbidity of the different diseases evaluated in this study are expected to change due to climate change in comparison to the historical baseline period. Cells colored red indicate an increase in mortality or morbidity, cells colored blue indicate a decrease in mortality or morbidity, and cells colored white indicate relatively no change. The values shown represent average annual changes in mortality and morbidity (per 100,000 people) in the period from 2041–50. Under the dry/hot mean scenario, we expect an increase in mortality for waterborne diseases (+3.7 deaths per 100,000 people) and heat-related diseases (+1.3 deaths per 100,00 people) in the period from 2041–50. We also expect an increase in the incidence of both diseases under the dry/hot mean scenario. These increases are caused by a positive relationship between incidence and temperatures. However, malaria deaths and cases are expected to decline (-0.7 deaths and -279 cases per 100,000 people) due to projected temperature and precipitation levels, which are less conducive to its transmission. In contrast to the above, malaria mortality and morbidity are also expected to increase under more moderate temperatures and higher precipitation in the wet/warm scenario. Table 10: Change in mortality and morbidity rates (per 100,000 people) by disease, 2041–50 Rate change under dry/hot mean Rate change under wet/warm mean Original rates climate scenario climate scenario Disease Deaths Cases Deaths Cases Deaths Cases Waterborne 31.7 97,546 3.7 11,259 2.5 7,692 Heat-related 0.8 676 1.3 1,186 0.8 721 Malaria 31.1 12,726 -0.7 -279 1.8 722 Dengue 0.0014 435 0.00012 37.0 0.00019 58.7 Figure 15 shows how the different diseases evaluated in this study are expected to contribute to the labor supply shock projected from the baseline scenario. Across all disease categories, waterborne and heat- related illnesses are expected to be responsible for most of the estimated negative labor supply shock. Dengue contributes close to 0 percent of shocks. Generally, additional disease impacts relative to current conditions are greater under the dry/hot mean scenario compared to the wet/warm mean scenario. Figure 15: Average labor supply shock by disease under the baseline, 2041–50 0.1% 0.0% −0.1% −0.2% −0.3% −0.4% −0.5% Water-borne Heat-related Malaria Dengue Dry/hot mean Wet/warm mean Dry/hot GCM range Wet/warm GCM range 22  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania While the largest proportion of annual waterborne disease deaths are concentrated in high-population areas like Dar es Salaam, the highest fraction of waterborne disease deaths attributable to temperature increases under climate change will be located in the central, southwest, and southern regions of the country due to spatial distribution of higher projected temperatures and historical mortalities (figure 16). Generally, the fractions of cases and deaths attributable to climate change across diseases are similar to the climate health vulnerability assessment results shared by the World Bank team. However, estimates for malaria are lower due to projected temperature and precipitation conditions falling outside optimal ranges for transmissibility under the dry/hot mean climate scenario, and only minor transmissibility improvements under the wet/warm mean scenario compared to historical conditions (figure 17). Notably, transmissibility under the dry/hot mean scenario is expected to decrease along the populous eastern coast as well as in the northeast and in the south, while transmissibility increases are anticipated in these regions under the wet/warm mean scenario. Figure 16: Fraction of waterborne diseases deaths attributable to temperature increase for 2041–50, relative to 1995–2020 a) Dry/hot b) Wet/warm 14% 12% 10% 8% 6% 4% 2% 0% Figure 17: Malaria transmissibility coefficient, 1995–2050 a) 1995–2020 b) Dry/hot (2041–50) c) Wet/warm (2041–50) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  23 Assuming constant mortality and morbidity rates, the overall labor supply in the country is expected to decrease over time as mid-century approaches, with the magnitude of the decrease across GCMs being relatively small, no more than -0.6 percent by 2050 (figure 18a ). Throughout the period we expect shocks to be most severe under the dry/hot mean scenario as compared to the wet/warm mean scenario. With the continuation of recent WASH metric trends through 2050, the BAU scenario brings average labor supply shocks by the 2040s down to approximately -0.1 percent for both mean climate scenarios through the resulting reductions in waterborne disease rates. Through the more significant waterborne disease reductions from targeted WASH improvements under both the BAU + climate action and ASP scenarios, negative labor supply shocks from climate change are not only negated but instead result in positive labor supply gains between +0.3 and +0.4 percent for BAU + climate action and about +1.8 percent for ASP. While the investments required to achieve these results are not costed, the impact of waterborne disease increases due to rising temperatures can be effectively mitigated through improvements in WASH infrastructure and access, creating labor supply gains in excess of those necessary to offset shocks from all modeled diseases. Figure 18: Labor supply shocks, all development scenarios, 3-year moving average a) Baseline d) BAU 2.00% 1.50% 1.00% 0.50% 0.00% -0.50% c) BAU with climate action d) ASP 2.00% 1.50% 1.00% 0.50% 0.00% -0.50% 2020 2030 2040 2050 2020 2030 2040 2050 Individual Dry/Hot GCMs Individual Wet/Warm GCMs Dry/Hot mean Wet/Warm mean Costs of adaptation Health adaptation costs are not considered as insufficient information is available to estimate the cost of achieving specific health targets. Discussion and key takeaways Overall, combined labor supply shocks from malaria, dengue, waterborne diseases, and heat-related illnesses are limited to no more than -0.5 percent by 2050 if current disease rates remain constant. The majority of these impacts stem from projected rises in temperature, with waterborne diseases experiencing the highest increase from current levels due to both temperature changes facilitating transmission and their 24  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania significant fraction of all cases in the country. Heat-related illnesses will also increase moderately through the same climatic drivers. The spread of malaria will vary with the temperature and precipitation conditions under each mean climate scenario, but with only minor shocks relative to its historic impacts. Improvements in access to safe water, sanitation, and hygiene in Tanzania are expected to notably reduce negative shocks under the BAU development scenario and produce labor supply gains under the BAU + climate action and ASP scenarios through resulting decreases in the mortality and morbidity of waterborne diseases. Overall, while macroeconomic labor supply impacts might be limited, household-level and systemwide healthcare effects might still be significant due to continuous increases in mortality and morbidity primarily caused by rising temperatures from climate change. 3.2. Natural capital Agriculture, land use, and forestry are all expected to experience a variety of impacts from climate change. Temperature increases are likely to reduce the suitability and productivity of crops, pastures, and livestock. Changes in precipitation patterns can result in reduced water resources available for agricultural users, impact erosion levels which in turn affect soil fertility and reservoir sedimentation, as well as influence reservoir water quality. The estimated impacts from climate change to Tanzania’s crop and livestock sectors, as well as erosion risk, are presented below. 3.2.1. Land use and land cover In addition to and in conjunction with impacts from climate change, the mix of land use and land cover (LULC) across Tanzania is expected to change differently depending on the development scenario. In general, the proportions of land allocated to cropland and forested are assumed to vary, driven by agriculture sector growth targets, as well as from deforestation and reforestation activities. Assumptions and resulting LULC changes are based on data provided by the World Bank team. Changes in LULC will impact the results of the crop production, soil erosion, and inland flooding channels, presented later in chapter 3. Figure 19 exhibits the national allocation of each land use type by development scenario. The BASE scenario represents the historical baseline allocation, while the other scenario categories correspond to BAU, ASP, BAU + climate action, and ASP + climate action allocations by 2050. Compared to historical allocations, all development scenarios see an increase in cropland, especially for the ASP and BAU scenarios; closed tree coverage increases across all scenarios, while open tree coverage increases for all scenarios except BAU; grassland increases under BAU and BAU + climate action but decreases under ASP and ASP + climate action; and shrubland decreases across all scenarios. Figure 19: Land use allocation (count, thousands of hectares) by development scenario 4e+06 3e+06 Count 2e+06 1e+06 0e+00 Shrubland Grassland Cropland Bare Built−up Water Shrub, flooded Tree, closed Tree, Open Land Use BASE BAU_2050 ASP_2050 BAU_Climate action ASP_Climate action Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  25 Figure 20 shows the assumed changes in LULC for cropland coverage under the BAU, ASP, BAU + climate action, and ASP + climate action development scenarios relative to the historical baseline period. Generally, cropland coverage in most regions is expected to increase across all scenarios, with the exceptions of Tanga, Njombe, Mtwara, Iringa, and Pwani, which are each expected to experience decreases with some consistency. Figure 20: Change in cropland coverage relative to 1995–2020, by region, all development scenarios a) BAU b) ASP c) BAU with climate action c) ASP with climate action Change in coverage (thousands ha) 1463 1250 1000 750 500 250 0 -250 -474 3.2.2. Crop production In Tanzania, smallholder farmers reliant on rainfall for irrigation make up the majority of the country’s agriculture sector (FAO, n.d.-a; International Trade Administration 2022a; FAO, n.d.-b). Less than 35 percent of available arable land is cultivated, and industry modernization is a challenge due to lack of technology, storage, and improved seed and fertilizer access. Key crops for food production include maize, beans, rice, cassava, groundnuts, sweet potatoes, bananas, sorghum, millet, and sugar cane (USDA Foreign Agriculture Service 2024; Arce and Caballero 2015). Maize is consumed as the primary crop for most households throughout the country. Export crops consist of cashews, coffee, cotton, tea, and tobacco. Unpredictable rainfall patterns, drought, and pests like locusts pose the greatest production risks to agricultural stakeholders, all of which are expected to be exacerbated by climate change. Reliance on rainfed agriculture creates particularly acute vulnerability to these weather-related challenges. In addition to climate risks, constrained private sector investments in agriculture due to low capacity and access to long-term capital limit development, and are increasingly problematic in combination with population growth and urbanization (USAID 2023). Overview of impact channel Under climate change, crop yields have the potential to be affected by changes in rainfall patterns/irrigation water availability, increasing evaporative demands, and extreme heat as temperatures rise. A summary of the modeling methodology used to estimate the impacts of climate change on crop production is presented below, with a more detailed description available in appendix B. First, representative crops were identified for the country based on local crop data and supplemented with FAO’s crop database (figure 21 shows the share of crops in Tanzania by area, production, and revenue). Data on harvested area, production, yield, and revenue statistics, were collected for each crop from available sources. When considering water availability, we apply the methods documented in FAO’s Irrigation and Drainage Paper 66, in which rainfed crop yields are estimated by applying crop-specific water sensitivity coefficients to the ratio of effective precipitation to potential crop evapotranspiration. For heat stress, the impacts to crop yields from extreme heat were modeled using AquaCrop’s approach, which considers a negative relationship between supra-optimal temperatures during the flowering stage of crop development. Finally, water availability and temperature shocks were then combined into a single shock by crop, with these effects aggregated nationally based on the spatial distribution of crop production. Crop-specific shocks were also aggregated into a total production shock using crop revenues as weights. 26  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 21: Share of rainfed crops by area, production, and revenue, 2018/2019 average 100% 2% 1% 1% 1% 0% 2% 2% 5% 9% 2% 1% 4% 2% 4% 4% 1% 2% 6% 80% 6% 8% 7% 2% 3% 7% 8% 8% 17% 8% 60% 7% 2% 9% 3% 8% 8% 11% 40% 8% 12% 27% 18% 20% 14% 8% 13% 14% 12% 0% 2% Area (ha) Production (ton) Revenue ($) Millet Sugarcane Sorghum Potato Sunflower Sweet potato Groundnut Beans Cassava Sesame Vegetables Rice Maize Tropical fruits Banana Note: Revenues consider production for both domestic consumption as well as exports. The projected changes in rainfed crop production were evaluated for the climate scenarios defined in chapter 2, as well as for different development scenarios explored below. Details on the data sources used for the analysis are provided in appendix C. Results For all crops considered in this analysis, the effects of climate change in Tanzania are expected to result in negative shocks to crop production under the dry/hot mean climate scenario and positive production gains under the wet/warm mean scenario. Figure 22 presents these shocks by crop for a baseline scenario that considers no agriculture investments or changes in cropland. Figure 22: Rainfed crop production shock under the baseline, 2041–50 20% 12% 9% 10% 10% 7% 7% 6% 6% 6% 7% 3% 4% 3% 1% 1% 0% −3% −3% −3% −4% −7% −7% −8% −10% −9% −9% −11% −10% −11% −14% −13% −14% −20% −25% −30% Vegetables Banana Beans Cassava Maize Millet Potato Rice Sesame Sorghum Sugarcane Sunflower Sweet potato Tropical fruits Groundnut Dry/hot mean Wet/warm mean Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  27 For the same scenario, figures 23 and 24 exhibit the regional shocks to crop production for bananas and tropical fruit, Tanzania’s two highest-value crops in terms of revenue which both have significant national production shocks. For both crops, the highest magnitude production shocks are concentrated in the eastern regions under the dry/hot mean scenario and in the central regions under the wet/warm mean scenario. As shown in Figure 25, most banana production is historically concentrated in northern regions such as Kagera, Kilimanjaro, Arusha, and Kigoma, where only minor to moderate shocks are present, while tropical fruit production is highest in Pwani, Kigoma, Mwanza, and Njombe, where impacts are less severe or more mixed between negative and positive shocks. Figure 23: Banana yield impact by region under the baseline, 2041–50 a) Dry/hot mean b) Wet/warm mean 27% 20% 10% 0% -10% -20% -26% Figure 24: Tropical fruit yield impact by region under the baseline, 2041–50 a) Dry/hot mean b) Wet/warm mean 12% 10% 5% 0% -5% -10% -15% -20% -25% -26% 28  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 25: Historical banana production Figure 26: Historical tropical fruit production distribution by region, 1995–20 distribution by region, 1995–20 41% 39% 40% 35% 35% 30% 30% 25% 25% 20% 20% 15% 15% 10% 10% 5% 0% 5% 0% Turning to our analysis on the impacts of climate change in conjunction with changes in average annual production, distribution, and irrigation, table 11 summarizes assumptions made for the development scenarios considered throughout the remainder of this impact channel. Table 11: Development scenarios evaluated for the crop production channel BAU ASP BAU+ climate action ASP+ climate action Assumes the following annual changes in Assumes the following annual changes in Same growth as Same as ASP but production on average between 2023 and production on average between 2023 and BAU but uses BAU + uses ASP + climate 2050: 2050: climate action LULC action LULC analysis results analysis results Rice = +0.1% Rice = +2.5% provided by the World provided by the Maize = +1.6% Maize = +4.6% Bank team. World Bank team. Cassava = +0.7% Cassava = +4.2% Irrigated areas grow from 289,381 Banana = +0.6% Banana = +2.7% hectares in 2020 to Bean = +0.8% Bean = +1.9% 3,057,594 hectares in 2050. All other crops = +1% All other crops = +1% Total crop acreage increases from 12.8 million Total crop acreage increases from 12.8 million hectares in 2020 to 21.1 million hectares hectares in 2020 to 19.3 million hectares in 2050. Initial crop production is based on in 2050. Initial crop production is based on local and FAO data for 2018/2019. Growth is local and FAO data for 2018/2019. Growth is distributed based on the BAU LULC analysis distributed based on the ASP LULC analysis results provided by the World Bank team. results provided by the World Bank team. Irrigated areas grow from 289,381 hectares in Irrigated areas grow from 289,381 hectares in 2020 to 1,400,000 hectares in 2050. 2020 to 3,057,594 hectares in 2050. By 2050 under the BAU scenario, climate change is anticipated to have negative production shocks on most rainfed crops under the dry/hot mean scenario and positive production shocks on most rainfed crops under the wet/warm mean scenario (figure 27). Only potato is expected to see negative effects under the wet/warm scenario and only rice is expected to see productivity gains under the dry/hot scenario, both of Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  29 which are relatively modest. Banana, potato, and vegetables are some of the most vulnerable rainfed crops, experiencing crop production shocks of between -18 and -19 percent under the dry/hot mean scenario. Rice and sugarcane are expected to see the biggest production gains under the wet/warm mean scenarios, at +15 and +16 percent, respectively. Figure 27: Rainfed crop production shock under BAU scenario, 2041–50 20% 15% 16% 10% 9% 9% 8% 6% 5% 4% 3% 4% 4% 3% 3% 1% 0% 0% −3% −2% −5% −5% −5% −10% −9% −8% −8% −10% −12% −14% −14% −20% −18% −19% −19% −30% Vegetables Sesame Banana Beans Cassava Maize Millet Potato Rice Sorghum Sugarcane Sunflower Sweet potato Tropical fruits Groundnut Dry/hot mean Wet/warm mean The production shocks shown in figure 27 are comprised of impacts due to increasing temperature and impacts due to changing water driven by altered precipitation patterns. Figure 28 separates the production shocks into each of these different effects. The heat effect due to increasing temperatures is negative for six crops under both climate means and negligible or zero for the remaining crops. Vegetables are expected to see the greatest effects due to heat. In contrast, the water effect is positive for all crops except potato under the wet/warm mean scenario and negative for all crops except rice under the dry/hot mean scenario. These differences are driven by the projected increases in rainfall under the wet/warm climate scenarios and projected decreases in rainfall under the dry/hot climate scenarios. Relative to the Baseline scenario, shocks to bananas and tropical fruit are expected to become more negative and more positive, respectively, under the assumed changes in irrigation, average production, and LULC for cropland in the BAU scenario. Negative water effects are partially curtailed for crops with existing irrigation—such as beans, cassava, maize, rice, sorghum, sugarcane, sweet potato, tropical fruits, and vegetables—as irrigated areas grow under the BAU scenario, which is assumed to reverse negative shocks for impacted hectares. When aggregated (as shown in figure 28), the significant water effect dominates the projected impacts for most crops under both climate scenarios, with the exception of vegetables, whose shocks are clearly driven by heat effects. Aggregating the projected production shocks across all the rainfed crops evaluated, figure 29 shows the overall impact of climate change on crop production under the BAU scenario. Overall, the dry/hot mean climate scenario is anticipated to result in negative production shocks during the period to 2050, whereas production shocks are generally positive under the wet/warm mean scenario, driven by projected increases in rainfall. In the 2040s, the average shock under the dry/hot mean scenario is expected to be around -7.5 percent, whereas the average shock under the wet/warm mean scenario is expected to be around +5.7 percent. 30  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 28: Rainfed crop production shock broken down into heat and water effect, 2041–50 a) Heat effect b) Water effect 20% 10% 0% −10% −20% Vegetables Vegetables Banana Beans Cassava Maize Millet Potato Rice Sesame Sorghum Sugarcane Sunflower Sweet potato Tropical fruits Banana Beans Cassava Maize Millet Potato Rice Sesame Sorghum Sugarcane Sunflower Sweet potato Tropical fruits Groundnut Groundnut Dry/hot mean Wet/warm mean Figure 29: Rainfed crop production shock, under a BAU scenario, 3-year moving average 20% 10% 0% -10% -20% 2020 2025 2030 2035 2040 2045 2050 Individual Dry/Hot GCMs Individual Wet/Warm GCMs Dry/Hot mean Wet/Warm mean The ASP, BAU + climate action, and ASP + climate action scenarios includes more widespread use of irrigation for beans, cassava, maize, rice, sorghum, sugarcane, sweet potato, tropical fruits, and vegetables, which are the crops that already have both irrigated and rainfed acreage. Because the yield gain from irrigation investments are not considered in this analysis, benefits are limited to reversing any negative impacts for the subset of crops receiving irrigation, even though further gains would be expected. An important assumption to note is that the expansion of irrigation in the country assumes sufficient water is available from surface water, groundwater, or other sources. This adaptation measure aims to lessen the water availability component of the production shock. Aggregating the projected production shocks across all the crops evaluated, figure 30 shows the overall impact of climate change on rainfed crop production under the BAU, ASP, BAU + climate action, and ASP + climate action scenarios. While production under the ASP, BAU + climate action, and ASP + climate action scenarios are expected to benefit from further irrigation development, the overall shocks under ASP and ASP + climate action are slightly more severe than under the BAU and BAU + climate action scenarios. This is driven by changes to cropland coverage under the LULC modeling, which alters the distribution and extent of cropland across scenarios. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  31 Figure 30: Rainfed crop production shock, under different scenarios, 3-year moving average a) BAU b) ASP 20% 10% 0% -10% -20% c) BAU with climate action d) ASP with climate action 20% 10% 0% -10% -20% 2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050 Individual Dry/Hot GCMs Individual Wet/Warm GCMs Dry/Hot mean Wet/Warm mean Costs of adaptation for rainfed crop production All modeled scenarios consider increased irrigation of rainfed crops. Irrigation is assumed to cost $1,500 per hectare, a metric provided by the World Bank team. To calculate total costs for this measure, a three percent interest rate, 15-year project length, and 2 percent operation and maintenance cost rate are used. Under the BAU scenario, total costs (capital and operating costs) are expected to reach around $3.1 billion from 2024 through to 2050, with an average annual cost of $153.1 million per year in the 2040s. Under the ASP, BAU + climate action, and ASP + climate action scenarios, total costs (capital and operating costs) are expected to reach around $6.6 billion from 2024 through to 2050, with an average annual cost of $366.4 million per year in the 2040s. Discussion and key takeaways Under climate change, the shocks to rainfed crop production will be driven by both heat effects due to higher temperatures as well as by water effects due to increased evapotranspiration and lower rainfall, with water deficits explaining a larger proportion of production shocks for all crops except vegetables. In general, climate change is expected to result in yield losses for most rainfed crops in the country under a dry/hot future and yield gains for most rainfed crops under a wet/warm future. For high-revenue crops, these yield impacts due to climate change are mainly concentrated in the eastern and central regions of Tanzania, such as Dar es Salaam, Zanzibar, Dodoma, Manyara, Singida, Morogoro, and Tanga. Despite assumptions 32  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania of greater irrigation and average production growth, the differing LULC cropland distributions under the ASP and ASP + climate action scenarios result in slightly higher climate vulnerability relative to the BAU and BAU + climate action scenarios. This is a result of the changes in the spatial distribution of harvested areas that are more vulnerable to climate change than the current mix. However, these results only consider climate- driven yield changes and exclude significant productivity gains that result from agricultural investments, such as irrigation. Including such productivity gains would result in a greater total output for the sector, regardless of the climate impacts. 3.2.3. Soil erosion Since the 1960s, sediment transport rates in Tanzania’s river systems have been on the rise due to agricultural land use change, deforestation, and overgrazing by livestock (Wynants et al. 2021). The population’s demand for agricultural services has continued to grow, despite loss of arable land and sector productivity due to soil resource degradation. An estimated 13 percent of all land in the country is degraded from soil erosion, poor farming practices, and related issues (WWF 2018). Precipitation variations and extreme events projected under climate change are likely to worsen that degradation. Regional climate models have suggested significant rainfall erosivity in East Africa under projected intensity increases, which will harm agricultural yields beyond the negative impacts from current high rates of soil erosion (Chapman et al. 2021). Mountainous regions in Tanzania are particularly vulnerable; terracing will likely decrease in effectiveness as impacts worsen, suggesting a need for increasingly extensive soil management and agroforestry practices. Overview of impact channel Erosion can be detrimental to landscapes, impacting plant and animal life, reducing the efficacy of reservoir storage and hydropower production through sedimentation, and causing declines in agricultural production by removing valuable nutrients from the topsoil, all of which can be made worse if climate change intensifies future rainfall intensity. A summary of the modeling methodology used to estimate the impacts of climate change on erosion is presented below, with a more detailed description available in appendix B. To determine erosion rates, we use the RUSLE, which requires five key inputs, namely rainfall-runoff erosivity, climate and land factors, as well as activity and farm-level management factors. Generally, areas that are impermeable—such as rocky surfaces or waterbodies—and areas with mean slopes that exceed 20 percent are excluded from the analysis because erosion on these surfaces tends to be low or highly uncertain with the RUSLE approach. Soil loss can reduce the nutrients available to crops, if not replenished by fertilizers, by eroding the topsoil. Although topsoil is generated naturally, natural generation is slow. To approximate the impact erosion has on the major crops grown in the country, we use a method developed by the FAO (Kassam et al. 1991). The approach is based on a tolerable loss rate over time and varies by levels of fertilizer inputs as well as the susceptibility of soils to productivity loss. We use raster data of fertilizer use (nitrogen and phosphorus) to determine the level of input in a country. Further details on the data sources used to complete the analysis are provided in appendix C. The projected changes in crop production due to erosion were evaluated for different development scenarios. Table 12 summarizes the assumptions made for each development scenario, while details on the data sources used to complete the analysis are provided in appendix C. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  33 Table 12: Development scenarios evaluated for the erosion channel BAU BAU+ climate action ASP ASP+climate action Assumes total acreage grows from Assumes the same crop area and yield Assumes total acreage Assumes the same crop 12,764,905 hectares in 2020 to assumptions as BAU. grows from 12,764,905 area and yield assumptions 21,111,381 hectares in 2050. hectares in 2020 to as ASP. Considers the same Distribution of growth was based Considers the following adaptation 19,333,551 hectares adaptation intervention from on LULC analysis results described interventions beginning in 2025: in 2050. Distribution of BAU + climate action. in section 3.2. Achieve 10% adoption of minimum growth was based on LULC tillage agriculture by 2050 analysis results described in Average annual yield increases, section 3.2. 2023–50: Leave crop residues on fields to serve as mulch Average annual yield Rice = 0.08% increases, 2023–50: Assume 10% adoption of this practice Maize = 1.6% by 2050 Rice = 2.5% Cassava = 0.72% These interventions are assumed to be Maize = 4.6% Banana = 0.63% cost-neutral to farmers since minimum Cassava = 4.2% Bean = 0.82% tillage sees a decline in labor and/or tractor but an increase in the need for Banana = 2.7% Other crops = 0.97% pesticides or other inputs. Bean = 1.9% Leaving crop residues as mulch is often Other crops = 1% used alongside conservation tillage practices. Results During the historical baseline period between 1995 and 2020, the country’s dry climate results in notable soil erosion Figure 31: Erosion during historical period, risk across most of the country, with highest concentrations 1995–20 of erosion risk in the southwest of the country, particularly in the Rufiji, Lake Nyasa, Ruvuma, and Lake Tanganyika basins (figure 31). Relative to these historical conditions, wet/warm scenarios are expected to result in erosion risk increases overall, while dry/hot scenarios are expected to result in mostly erosion risk decreases by the 2040s (figure 32). Under the wet/warm scenario, erosion risk increases throughout the entirety of the country, with the highest increases of 50+ tons per hectare per year aligning with historically high erosion risk areas. In contrast, erosion risk under the dry/hot scenario generally declines from historical baseline conditions, with some increases projected in the western and southwestern parts of the country. The wet/warm scenarios are generally associated with greater precipitation volumes and intensities than the dry/hot scenarios, resulting in more widespread increases in erosion risk. While erosion generally increases 0 4 8 17 44 3007 > over time, the rate of erosion is dependent on changes in Soil Erosion Risk (tons/ha/year) precipitation, resulting in erosion rate decreases under drier conditions. 34  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 32: Erosion risk by 2040 under two futures, BAU scenario a. Dry/hot GCM GFDL-ESM4 SSP2–4.5 b. Wet/warm GCM EC-EARTH3 SSP2–4.5 > > -50 -10 -2 0 2 10 50 Change in Erosion Risk (tons/ha/year) Under the BAU and ASP development scenarios, this erosion is expected to have important implications for crop production within the country due to the loss of fertile topsoil. The wet/warm mean climate scenario is expected to result in negative production shocks for every crop evaluated, while the dry/hot mean scenarios are expected to result in nearly negligible shocks for all crops except banana, sugarcane, and beans, which each have shocks nearer to -1 percent (figure 33). Shocks are greater under the wet/warm mean scenario than under the dry/hot mean scenario due to more intense projected precipitation throughout the country, resulting in greater erosion and related crop production impacts. However, under the BAU + climate action and ASP + climate action development scenarios, the achievement of conservation agriculture practices result in fully negligible or marginally positive crop production shocks between 0 and +1 percent. Across all development scenarios evaluated, potato and tropical fruit are most vulnerable to severe production impacts from changes in erosion, representing the most positive or negative shocks under the development scenarios without and with conservation agriculture adoption, respectively, across all crops. Crops with more severe impacts relative to others are more susceptible to erosion risk due to their respective cropland distributions’ topography, soil characteristics, current rainfall dynamics, and expected changes in precipitation. Aggregating the projected production shocks across all the crops evaluated, figure 34 shows the overall impact of erosion on crop production across all development scenarios. Under the BAU and ASP development scenarios, negative shocks grow in the decades leading to 2050 under both the wet/warm mean and dry/ hot mean scenarios, reaching values of approximately –1.3 and -0.4 percent respectively in the period from by 2041–50. Under the NDC scenarios, with the implementation of conservation tillage and mulching to reduce erosion impacts despite more intense precipitation events, production shocks are projected to be either less severe or positive for each climate scenario by mid-century, reaching approximately -0.1 percent by 2041–50 under the wet/warm mean scenario and +0.3 percent under the dry/hot mean scenario by 2041–50. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  35 Figure 33: Crop production shock due to erosion, 2041–50 a) BAU b) ASP 2% 0% -2% -4% c) BAU with climate action d) ASP with climate action 2% 0% -2% -4% Vegetables Vegetables Banana Beans Cassava Maize Potato Rice Sesame Sorghum Sugarcane Sunflower Tropical fruits Banana Beans Cassava Maize Potato Rice Sesame Sweet potato Sorghum Sugarcane Sunflower Sweet potato Tropical fruits Groundnut Millet Groundnut Millet Dry/hot mean Wet/warm mean Dry/hot GCM range Wet/warm GCM range Figure 34: Crop production shock due to erosion, under different development scenarios, 3-year moving average a) BAU b) ASP 1% 0% -1% -2% c) BAU with climate action d) ASP with climate action 1% 0% -1% -2% 2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050 Dry/hot mean Wet/warm mean Individual dry GCMs Individual wet GCMs 36  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Costs of adaptation The BAU + climate action and ASP NDC development scenarios consider the introduction of conservation agriculture practices to reduce erosion and topsoil loss, including minimum tillage and crop residue as mulch. Conservation tillage is assumed to be cost neutral to farmers since labor and/or tractor use decline, while this intervention may increase the need for pesticides or other inputs. The use of crop residue as mulch is also assumed to be cost neutral and is often used alongside conservation tillage practices. Discussion and key takeaways Erosion can be detrimental to landscapes, impacting plant and animal life, reducing the efficacy of reservoir storage and hydropower production through sedimentation, and causing declines in agricultural production by removing valuable nutrients from the topsoil, all of which can be made worse if climate change increases future rainfall intensity. Due to the elevated frequency and intensity of precipitation events from climate change combined with continued land use change, overgrazing, and deforestation, projected erosion risk in Tanzania is expected to result in crop production shocks of between -0.4 and -1.3 percent by the 2040s, under the dry/hot and wet/warm climate scenarios, respectively. Relatively low levels of adoption of conservation agriculture practices are sufficient to reverse these negative shocks under anticipated climate conditions, resulting in negligible or positive crop production impacts. This suggests conservation tillage and crop residue mulching are strong pathways for adaptation against climate-related precipitation effects in Tanzania. 3.2.4. Livestock production Tanzania’s livestock population is the third largest in Africa, following Ethiopia as the second largest (FAO, n.d.-b). Approximately 27 percent of agriculture’s contribution to GDP is attributable to the livestock sector; beef, dairy, and other products, such as eggs, hides, and skin, account for about 40, 30, and 30 percent of this share, respectively. Figure 35 presents the main livestock products in Tanzania by revenue. Leading products include cattle meat, cattle milk, goat meat, and eggs, which make up 87 percent of revenue. In 2020, Tanzania had about 34 million cattle, 24.5 million goats, 8.5 million sheep, and 87.7 million poultry. Approximately half of all households in Tanzania owned livestock, with about 15 percent of annual rural household income on average obtained from the sale of animal products (de Glanville et al. 2020). Despite its prevalence, the livestock sector has low growth and poor productivity due to issues like high mortality, high rates of disease, and low reproduction (United Republic of Tanzania, Ministry of Livestock and Fisheries 2017). Government policymakers have developed plans to pursue livestock investment interventions as a result, an effort continued by the Livestock Sector Transformation Plan for 2023–27 (United Republic of Tanzania, Ministry of Livestock and Fisheries 2022). Precipitation variability and droughts expected from climate change are likely to have a significant impact on stress conditions and grazing land availability for livestock throughout Tanzania, worsening existing challenges like the water shortages already threatening the sector (Ires 2021). Overview of impact channel Climate change poses risks to livestock production from both direct and indirect effects. Increasing heat stress on animals (direct effect) causes reductions in productivity, while climate change can also cause potential reductions in the availability of feed sources (indirect effect), resulting in lower energy intakes and reduced yields. Our analysis focused on the main species globally (cattle, chicken, swine, sheep, and goats) for the three most important products (milk, meat, and eggs). A summary of the modeling methodology used to estimate the impacts of climate change on livestock is presented below, with a more detailed description available in appendix B. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  37 Figure 35: Share of livestock revenue, by product, 2017–21 average 2% 25% 5% 3% 1% 9% 3% 48% 3% Cattle Goats Chickens Sheep Swine Other Meat Milk Eggs Wool Other Note: Revenues consider production for both domestic consumption as well as exports. We estimated the direct effects using a combination of animal-specific equations for milk and eggs that relate a daily temperature-humidity index as an indicator of heat stress, with productivity losses based on heat tolerance thresholds. The temperature-humidity index calculation varies by species and requires data on air temperature, wet-bulb temperature, and relative humidity. Estimation of indirect effects relies on changes in grassland productivity, impacting the fraction of livestock that feeds from grazing pastures. We model pasture productivity following the FAO’s Irrigation and Drainage Paper 66 for water availability, and consider a heat stress effect on grasses using the AquaCrop approach, which considers a negative relationship between supra-optimal temperatures during crop development. Reductions in feed availability are converted into changes in metabolizable energy, which causes metabolic reductions that are utilized as direct meat production losses. These individual effects are aggregated nationally based on livestock headcounts by region and revenues by product. The projected changes in livestock production and livestock emissions were evaluated for the climate scenarios defined in section 2.3.2, as well as for different development scenarios. Table 13 summarizes the assumptions made for each development scenario, while details on the data sources used to complete the analysis are provided in appendix C. Table 13: Development scenarios evaluated for the livestock channel BAU ASP BAU+climate action ASP+climate action Assumes the following annual changes Assumes the following annual changes Assumes livestock Assumes livestock in production on average between in production on average between production and headcount production and headcount 2023 and 2050: 2023 and 2050: growth are the same as BAU. growth are the same as ASP. Chicken meat = +4.8% Chicken meat = +10% Additionally, assumes Additionally, assumes Chicken headcount = +3.8% Chicken headcount = +7.5% 90% of pasture losses are 90% of pasture losses are Cattle meat = +5.5% Cattle meat = +7.5% substituted with imported substituted with imported feed (impacting meat feed (impacting meat Cattle headcount = +5.5% Cattle headcount = +5.5% production) and heat production) and heat Other meat = +5.1% Other meat = +8.7% abatement measures are abatement measures installed for dairy cattle are installed for dairy Other headcount = + 4.6% Other headcount = +6.5% (impacting dairy production). cattle (impacting dairy Total livestock headcount increases Total livestock headcount increases production). from 126.7 million in 2020 to 395.7 from 126.7 million in 2020 to 874.2 million in 2050. Initial livestock million in 2050. Initial livestock production is based on the FAO’s production is based on the FAO’s reported average for 2017–21. reported average for 2017–21. 38  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Results As noted, livestock production is affected by climate change in two ways: meat is impacted by feed availability and milk is impacted by heat effects. Within Tanzania’s livestock sector, meat from cattle, goats, and sheep are the most important sources of meat. As shown in figure 36, the mid-century shock to meat production from cattle, goats, and sheep ranges between approximately -3 and -4 percent under the wet/warm mean scenario and between -10 to -13 percent under the dry/hot mean scenario under BAU conditions. Swine meat is relatively insignificant in terms of revenue and is anticipated to have near- negligible shocks because it primarily relies on feed sources other than grazing pastures, which are not modeled. Shocks to cattle and goat milk are modeled through heat effects and anticipated to be relatively mild, ranging from approximately -1 to -3 percent for cattle milk under the wet/warm and dry/hot mean scenarios, respectively, and roughly negligible for goat milk under both mean climate scenarios. Shocks to chicken eggs, which are also measured through heat effects, are comparable to those for cattle milk. Shocks to these products are limited because the projected changes in temperature are expected to be less impactful than the projected changes in precipitation for the livestock sector. This suggests that the impacts of reduced feed availability pose a greater threat to the livestock sector in Tanzania than the effects of increased heat. These national shocks to meat production are driven by regions where that production is concentrated, despite climate effects that could impact feed availability extending beyond those productive areas. Figure 36: Livestock production shock under a BAU scenario by product, 2041–50 Cattle meat Cattle milk Chickens eggs Goats meat Goats milk Sheep meat Swine meat 0% -5% -10% -15 Dry/hot mean Wet/warm mean Figure 37 exhibits product-specific shocks across all development scenarios considered. As expected, relative to BAU, modeled increases in average production and headcount under the ASP scenario do not alter the expected shocks attributable to climate change. However, under the BAU + climate action and ASP + climate action scenarios, the 90 percent substitution of pasture losses with imported feed significantly reduces negative shocks for meat products by up to three-quarters, and heat abatement slightly reduces shocks for milk and egg products as well. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  39 Figure 37: Livestock production shock under all development scenarios, 2041–50 a) BAU b) ASP 0% -5% -10% -15% c) BAU with climate action d) ASP with climate action 0% -5% -10% -15% Cattle Cattle Chickens Goats Goats Sheep Swine Cattle Cattle Chickens Goats Goats Sheep Swine meat milk eggs meat milk meat meat meat milk eggs meat milk meat meat Dry/hot mean Wet/warm mean Figure 38 shows the revenue-weighted shock to total livestock productivity due to climate change. Under all development scenarios, climate change is expected to cause livestock production losses through mid- century, with greater losses under the dry/hot scenarios than under the wet/warm scenarios for the majority of the analyzed period. Average shocks are projected to reach approximately -2.8 and -2.6 percent under the wet/warm mean scenario and -7.3 and -6.8 percent dry/hot mean scenarios for the BAU and ASP development scenarios, respectively, by the period from 2041–50. Projected shocks earlier in the century are significant as well, with average shocks from 2024–41 anticipated at -3.1 to -5.4 percent under BAU and -2.6 to -5.2 percent under ASP, depending on the climate scenario. Difference in the BAU and ASP scenarios are due to yields and headcount growing at different rates across products. Under the BAU + climate action and BAU+ASP scenarios, shocks are reduced substantially reduced to approximately -0.4 to -0.5 percent and -2.3 percent for the wet/warm and dry/hot mean climate scenarios, respectively, by the 2040s. While shocks under adaptation measures from the two NDC scenarios are still negative, they are stable and significantly lower than the BAU and ASP development scenarios approaching mid-century. Costs of adaptation The BAU + climate action and ASP + climate action scenarios consider two resilience-enhancing measures for Tanzania’s livestock sector: importing feed to substitute up to 90 percent of the pasture losses to reduce meat impacts on ruminants, and heat abatement investments for commercial dairy farms. Imported feed requirements are calculated as the difference between mean feed intake requirements by herd and reduction in grassland productivity for the fraction of livestock that relies on pastures. The model assumes that grazing pastures are being exploited at capacity, therefore any reductions in yield will result in insufficient feed. Imported feed is estimated based on the current consumption and import prices of nonpasture feed sources for grains, crop residues, and haylage from the World Bank’s World Integrated Trade Solution database and weighted based on current import proportions obtained from the FAO’s Supply Utilization Accounts. Heat abatement investments consider fans and forced ventilation for commercial dairy farms, with benefits and costs obtained from St-Pierre, Cobanov and Schnitkey (2003). These investments can reduce the 40  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 38: Livestock production shock under all development scenarios, 3-year moving average a) BAU b) ASP 0% -5% -10% -15% c) BAU with climate action d) ASP with climate action 0% -5% -10% -15% 2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050 Individual dry/hot GCMs Individual wet/warm GCMs Dry/hot mean Wet/warm mean temperature-humidity index that drives milk losses, and its costs are estimated at $250 for a cooling system that can cool up to 3,800 kilograms of equivalent livestock units (about 15 250-kilogram cows) plus energy consumption costs. These costs only include cooling systems, and it is assumed that cattle in the adaptation group are already housed. Capital costs are annualized, assuming a 15-year lifetime and 3% discount rate. Variable costs consider the cost of energy consumption. In total, under the BAU + climate action and ASP + climate action scenarios, investment requirements are approximately $282 million and $330 million annually, on average, from 2041 to 2050. From 2024 to 2050, these costs amount to approximately $3.7 billion and $4.2 billion by scenario, respectively. The elevated production and headcount assumptions under the ASP + climate action scenarios require greater amounts of feed import and heat abatement, resulting in higher adaptation costs. Discussion and key takeaways Cattle, goat, and sheep meat are projected to be Tanzania’s most vulnerable livestock products to the effects of climate change, facing shocks as high as -10 to -13 percent by 2050. Milk and egg production in Tanzania is projected to be less vulnerable to the impacts of climate change than meat production, suggesting reductions in feed availability are a greater threat to the sector than heat effects. The projected impacts to cattle meat, goat meat, and sheep meat production can be significantly reduced by replacing lost feed with feed from other sources, as modeled under the NDC scenarios, although residual shocks of no more than Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  41 approximately -2.3 percent remain under certain dry/hot climate scenarios even after feed substitution. These residual shocks will have especially far-reaching effects on households that rely on their own herd for food or income. Feed substitution does not provide added resilience for milk or egg production as it remains vulnerable to increasing temperatures. While additional climate adaptation investments such as the installation of heat abatement measures such as fans and sprinklers can reduce the production shocks experienced, the residual shocks remain similarly mild relative to BAU, and comparable to residual meat production shocks after feed substitution. 3.3. Physical capital Climate change is likely to impact physical capital, and the services provided by it in a variety of ways including by increasing the frequency and magnitude of extreme events that result in damage to assets, as well as by increasing deterioration caused by heat and precipitation levels. The estimated impacts to Tanzania’s roads and bridges are presented below, as well as an evaluation of inland flooding damage due to climate change. 3.3.1. Inland flooding Tanzania is considered the East African country most impacted by flooding (USAID 2021). As greater proportions of precipitation in Tanzania consist of heavy rainfall events from climate change, vulnerability to flooding will increase. Rivers like the Ruvuma are already prone to flooding and significant seasonal flow variability, which will worsen as extreme precipitation events increase. The highest riverine flood risks exist in the north near Lake Victoria and the Mara River, central regions along Lake Eyasi and the Kisigo River near Dodoma, in the west along Malagarasi River, Lake Tanganyika, and the Ugalla River, in the east near the Pangani and Wami Rivers, and in the south along the Great Ruaha River, Kilombero Valley Floodplain, Rufiji River, and the Mbwemkuru River. In May 2020, heavy cumulative rainfall resulted in a water level of 13.45 meters at Lake Victoria, creating widespread fluvial flooding that displaced nearby communities (C40 Cities Finance Facility 2020). Flooding at Lake Tanganyika in Western Tanzania during the same month caused similar damage. Populous areas in the country impacted by flooding experience substantial damage, such as a severe flood in Dar es Salaam in 2018, which inflicted $107–227 million in economic losses. Overview of impact channel Climate change may exacerbate flooding by increasing the frequency, intensity, and duration of storm events. This analysis relies on projected changes in the return interval of precipitation events from the World Bank’s Climate Knowledge Portal. Flood hazard maps are developed to determine areas with a certain probability of flooding for a given baseline and climate change projected return period. The outputs of the flood hazard mapping include the extent and depth of flood inundation, which are then used to estimate damage to infrastructure. The analysis is done for each time period of interest, climate scenario, and recurrence interval at a spatial resolution determined by the available hydrology and infrastructure asset distribution data. The resulting outputs are aggregated to a national scale, and correspond to the expected share of assets damaged relative to a historic baseline (1995 to 2020). A more detailed description of the modeling methodology used is available in appendix B, with appendix C offering details on the data sources utilized in the analysis. For this analysis, we estimate shocks to capital across different development scenarios, as summarized in table 14. To determine the absorption capacity of soils, we use LULC analysis results provided by the World Bank team and described at the beginning of chapter 3. LULC scenarios are mapped to flood curve numbers, which represent the ability of the surface to absorb rainfall. 42  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Table 14: Development scenarios evaluated for inland flooding channel BAU ASP BAU+Climate action ASP+Climate action Assumes current design Assumes current design Assumes half of all existing Assumes half of all existing standards (i.e., the 10-year standards (i.e., the 10-year capital within the floodplain area capital within the floodplain area flood) do not change and that no flood) do not change and that no are protected up to the first meter are protected up to the first meter floodproofing of infrastructure is floodproofing of infrastructure is of the structure by 2050. of the structure by 2050. conducted. conducted. Assumes the BAU + climate Assumes the ASP + climate Assumes the BAU LULC scenario Assumes the ASP LULC scenario action LULC scenario from section action LULC scenario from section from section 3.2 by 2050. from section 3.2 by 2050. 3.2 by 2050. 3.2 by 2050. Results Inland flooding exposure and damage in the historic baseline Infrastructure is considered exposed to inland flooding if it is located within Figure 39: Percent of capital exposed to flooding, 1995–2020 an area that is flood prone. Figure 39 illustrates the percentage of capital by district that is exposed to inland 16% flooding during the historical baseline 15% (1995–2020). At the national level, 12% around 10.6 percent of capital in the 10% country is estimated to be exposed to floods, while certain northwestern and Zanzibar 8% southeastern regions (Pwani, Kigoma, Dodoma Dar es 5% Morogoro, Tanga) are particularly at Salaam 2% risk, with between 13 to 16 percent of capital in these regions exposed 0% to flooding. This suggests that in the historical baseline, the country’s capital faces substantial exposure to inland flooding, without considering any incremental effects caused by a changing climate. While it might be expected that greater proportions of capital are exposed to inland flooding around Lake Victoria due to changes in water levels, this analysis only considers riverine flooding. When infrastructure in an exposed area experiences a flood event, flood damage could occur depending on a variety of factors, including the magnitude of the flood and resulting flood depths. Figure 40 presents the percentage of local capital damaged due to inland flooding by return period in the historical baseline. Flood events of higher return periods—that is, events of a higher magnitude and a lower probability of occurrence, such as a 100-year event—result in higher flood depths, hence a higher percent of capital is damaged as compared to more common, smaller-magnitude events such as a 10-year flood). For a single magnitude of event, different parts of the country experience different degrees of damage. For instance, for the 100-year event in figure 40, northern and southern regions such as Dar es Salaam, Mwanza, Kaskazini Unguja, Kilimanjaro, and Shinyanga experience significant damage, with 4.7 to 8.5 percent of capital in these areas impacted historically. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  43 Figure 40: Inland flooding impacts expressed as percent of capital stock, 1995–2020, by region a) 20-year event b) 50-year event c) 100-year event 9.6% 8.0% 6.0% 4.0% 2.0% 0.0% In total, 0.8% of national capital In total, 1.9% of national capital In total, 2.6% of national capital damaged damaged damaged While the results presented above focus on the percent of national capital that is damaged by a single 20- or 50-year flood event, a standard metric used to describe flood damage is annual expected damage. An estimate of the annual expected damage for a country is the sum of all magnitudes of flood events times their annual probability of occurring (a 10-year flood times 1/10, plus the 11-year event times 1/11, plus the 12-year event times 1/12, and so on). Expected damage impacts are often seemingly very small due to the impact of multiplying the damage by their probability of occurring in a given year. For Tanzania, the expected annual damage for the historical baseline period are estimated at 0.1 percent of the national capital. Given the nature of flood events, this is often experienced as a series of years with next to no damage, followed by a single large flood event that causes significant damage in the year it occurs. Future inland flooding exposure and damage considering the impacts of climate change Having estimated the flood exposure and flood damage that occurs in the historic baseline from 1995–2020, this section now looks at the projected incremental changes for 2021–50 as a result of climate change. As shown in figure 40, under historical conditions, a single 20-year event is estimated to result in an impact to capital of around -0.8 percent, while a single 50-year and 100-year event result in impacts of around -1.9 and -2.6 percent, respectively. Looking at future projected shocks from inland flooding, Figure 41 summarizes the projected future flooding impacts, including impacts from climate change, under BAU. For the SSP2–4.5 climate scenario, losses are expected to increase to around -1.7, -2.5 and -3.2 percent of capital by the 2040s for the 25-year, 50-year and 100-year events, respectively. These represent increases of as much as 53 percent for a 25-year event and 31 and 22 percent for the 50-year and 100-year events, respectively. The percentage change in capital damage is projected to grow less extreme when considering larger, less frequent events such as the 50 and 100-year flooding events. Figure 42 takes the national-level damage associated with 100-year flood events in the future (as shown in Figure 41) and disaggregates it spatially for SSP3–7.0. Overall, the change in capital damage is expected to be highest for northern, northeastern, and certain southern regions, such as Mara, Tanga, Njombe, and Dar es Salaam, with increases in capital damage due to inland flooding reaching between 60 and more than 76 percent relative to the historical baseline. As a result of land use and land cover changes across development scenarios, anticipated regional changes in capital damage from flooding relative to the historical baseline are varied. As shown in figure 43, under the ASP, BAU + climate action, and ASP + climate action development scenarios, the southern and eastern regions of Mtwara, Njombe, Ruvuma, Lindi, Pwani, and Iringa, as well as Zanzibar, are consistently more severely impacted by inland flooding under climate change than in the BAU scenarios. However, under ASP and ASP + climate action, certain central 44  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure 41: Mean annual percent of total capital damaged by return period, under a BAU scenario a) 25-year flood b) 50-year flood c) 100-year flood 3.5% 3.0% 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Historical 2021-30 2031-40 2041-50 Historical 2021-30 2031-40 2041-50 Historical 2021-30 2031-40 2041-50 Historical SSP1-1.9 mean SSP2-4.5 mean SSP3-7.0 mean and northern regions such as Singida, Dodoma, and Kagera are slightly less impacted by flooding than under the BAU scenario. These changes are consistent with declines in forest and especially crop coverage in these development scenarios relative to BAU (as presented in section 3.2.1), resulting in reduced rainfall absorption and worsened flooding. Figure 42: Change in capital damage for 100-year flood in each development scenario relative to 1995–2020, by region, with difference caused by climate change under SSP3–7.0 ) BAU a) ) ASP b) ) BAU with climate action c) ) ASP with climate action d) 76% 70% 60% 50% 40% 30% 20% 14% Figure 43: Change in capital damage for 100-year flood under SSP3–7.0 relative to BAU, by region a) ASP b) BAU with climate action c) ASP with climate action 28% 25% 20% 15% 10% 5% 0% -5% -6% Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  45 Shifting once again from looking at damage associated with a single flood event to considering expected annual damage, figure 44 shows the change in the expected annual damage from inland flooding under all development and climate scenarios. As shown, the expected annual damage from inland flooding under the BAU and ASP scenarios by up to between 0.06 and 0.07 percent on average by the 2040s. The bars in panels c and d indicate the effect of floodproofing the first meter of structures under the BAU + climate action and ASP + climate action scenarios. As expected, this is projected to have positive effects, with the change in the expected annual inland flooding damage decreasing relative to the no adaptation BAU and ASP scenarios for all climate scenarios considered. Figure 44: Annual average shock to capital by development and climate scenario under all development scenarios, 2041–50 a) BAU b) ASP c) BAU with climate action d) ASP with climate action 0.02% 0.00% -0.02% -0.05% SSP1-1.9 mean SSP2-4.5 mean SSP3-7.0 mean Costs of adaptation The BAU + climate action and ASP + climate action adaptation scenarios consider implementation of floodproofing half of existing exposed structures within the floodplain the first meter. Floodproofing costs are based on Miyamoto International Inc. (2021) and capital costs are assumed to be 10 percent of the building’s capital value. Under the BAU + climate action and ASP + climate action scenarios, capital costs are expected to reach approximately $89.3 million annually on average from 2041–50, or $2.4 billion total from 2024–50. Discussion and key takeaways Overall, our results suggest that in the historical baseline, the country experiences expected annual damage from inland flooding of 0.1 percent of capital, which are significant in monetary terms. Under climate change, expected annual damage is projected to increase by an additional 0.06 to 0.07 percent on average in the period from 2041–50 under the SSP2–4.5 mean climate scenario. Additionally, the change in inland flooding capital damage for individual floods of a specified magnitude are projected to be significant in the future. For example, the change in capital damage associated with the 100-year flood event is projected to increase by as much as 76 percent by 2050 in certain regions of the country, under SSP3–7.0. For this reason, we evaluate the implementation of infrastructure floodproofing and find that these measures can effectively reduce these impacts to capital by mid-century, reducing additional annual capital damage by up to 97 percent under SSP2–4.5 and fully negating damage into a reduction in expected annual damage under SSP1–1.9. Changes in land use and land cover under each development scenario notably affect regional capital damage, with declines in crop and forest cover resulting in more severe flooding. 46  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania 3.3.2. Bridges Along with road infrastructure, bridges in Tanzania are likely to experience impacts related to climate change, particularly through increased precipitation and flooding. Past flooding events have severely affected the country’s bridges. The Tanga flood in 2019 resulted in significant damage to bridges and other critical infrastructure (IMF 2023). In May of this year, flash floods due to a tropical cyclone incapacitated four of Dar es Salaam’s bridges, causing shutdowns along a major urban highway (HICGI News Agency 2024). Bridges in the middle Msimbazi river basin have also been identified as key hotspots for flood mitigation, such as the Kawawa Road bridge (C40 Cities Finance Facility 2021). While Tanzania’s government is progressing towards enhancing bridge infrastructure to better withstand flooding (IMF 2023), additional measures may be necessary to prevent increased costs, hindered access to social services, and mobility disruptions (Stephen and Makyara 2022). Overview of impact channel Climate change may impact bridge infrastructure in a variety of ways. This includes damage to—and increased repair and maintenance of—bridges from changes in the recurrence of peak precipitation events. The need for increased repair and maintenance can be due to increased moisture, washaways, and overtopping from riverine flooding events. Repair and maintenance costs for road infrastructure are estimated based on unit cost assumptions. Bridge exposure is measured by quantifying the number of bridges over waterways, and their vulnerability based on the characteristics of those rivers in terms of road traffic levels and their surface material. Flood damage is determined using modeled river flows and velocities per river segment, which are used to determine areas with a certain probability of flood damage for a given baseline and climate change projected return period. In addition to the repair and maintenance costs estimated for bridges, a disruption cost analysis was also conducted to estimate the delay costs of damaged bridges requiring repair and maintenance, which in turn result in labor supply effects. The disruption analysis evaluates the time that each bridge in the network is estimated to be fully under repair or fully out of service as a result of climate change. A more detailed description of the modeling methodology used is available in appendix B, with appendix C offering details on the data sources utilized in the analysis. The total impacts in terms of maintenance and repair costs, as well as delays are estimated for a BAU scenario as well as a BAU + climate action scenario. Table 15 presents the assumptions for each development scenario, with appendix C offering details on the data sources utilized in the analysis. Table 15: Adaptation scenarios evaluated for the bridges channel BAU BAU + Climate action ASP ASP + Climate action Bridge damage is incurred assuming Assumes bridges are upgraded to withstand the Same as BAU. Same as BAU + the current bridge stock with no next-higher design event (e.g., a bridge designed climate action. proactive adaptation measures for the 25-year flow is improved to the 50-year undertaken. design event). It is assumed that 2% of the bridge inventory is upgraded annually between 2021 and 2050. Results Looking at future projected shocks from inland flooding on bridges, figure 45 summarizes the projected future impacts under the BAU scenario. Under historical conditions, a single 25-year event is estimated to result in modest impacts ($5 million). When considering impacts from climate change, we anticipate a Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  47 sizable increase in damage reaching nearly $100 million under the SSP2–4.5 ensemble mean in the period from 2041–50. Across all return periods shown below, projected impacts are expected to increase under climate change relative to historical conditions, though increases across more infrequent events (the 50-year and 100-year events) are anticipated to be smaller. Increases in bridge damage across return periods and climate scenarios align with capital damage trends presented in the inland flooding channel—for example, damage for both channels is the highest for floods under SSP2–4.5. This similarity reflects the relationship between projected flood impacts and damage to bridge infrastructure. Figure 45: Average annual bridge damage cost, by return period, under BAU a) 25-year flood b) 50-year flood c) 100-year flood 250 200 Bridge damage costs (mil USD) 150 100 50 0 Historical 2021-30 2031-40 2041-50 Historical 2021-30 2031-40 2041-50 Historical 2021-30 2031-40 2041-50 Historical SSP1-1.9 mean SSP2-4.5 mean SSP3-7.0 mean Shifting from looking at damage associated with a single flood event to considering expected annual damage, figure 46a shows the change in the expected annual damage to bridges for the BAU scenario across selected climate scenarios. As shown, the expected annual damage from inland flooding on bridge infrastructure is anticipated to result in incremental damage of approximately $4 million by 2050 under SSP2–4.5. Figure 46b looks at the impact of improving building design standards under the BAU + climate action scenario. As expected, this is projected to have positive effects, with the change in the expected annual inland flooding damage reduced by nearly $1.5 million for all climate scenarios considered. Again, expected annual damage across climate scenarios aligns closely with trends in expected capital loss under the inland flooding channel. Figure 46: Expected annual incremental bridge damage by development scenario a) BAU b) ASP 4 Incremental cost (mill USD) 3 2 1 0 2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050 SSP1−1.9 mean SSP2−4.5 mean SSP3−7.0 mean 48  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Next, considering labor supply impacts, figure 47a shows the expected annual change in labor supply due to bridge infrastructure damage for the BAU scenario across selected climate scenarios. As shown, the expected annual labor supply shock is expected to reach no more than -0.0024 percent by 2050 under SSP2–4.5. Figure 47b considers the impact of improving bridge design standards as outlined under the BAU + climate action scenario. As expected, this is projected to have positive effects, with the labor supply shock reduced for all climate scenarios considered. Figure 47: Labor supply shock due to bridge delays, by adaptation scenario a) BAU b) BAU with climate action 0.000% -0.001% -0.002% -0.003% 2020 2025 2030 2035 2040 2045 2050 2020 2025 2030 2035 2040 2045 2050 SSP1−1.9 mean SSP2−4.5 mean SSP3−7.0 mean Costs of adaptation Under the BAU + climate action development scenario, bridge upgrades to withstand riverine flooding events of the next highest return period are expected to cost nearly $185 million in 2024–50. Over the analysis period, these costs are equivalent to approximately $7 million per year on average to upgrade 2 percent of Tanzania’s bridge inventory annually until 2050. Discussion and key takeaways Under projected conditions in Tanzania, climate change related impacts to riverine flooding events are anticipated to increase damage to bridge infrastructure relative to historical baseline conditions, aligning with expectations of increased capital damage established by the inland flooding channel. Specifically, by 2050, the expected annual damage to bridge infrastructure is anticipated to result in incremental damage of approximately $4 million under SSP2–4.5. Delay impacts are expected to create only minor labor supply impacts, suggesting that higher peak precipitation events under climate change are more likely to be realized in terms of infrastructure damage requiring additional repairs and maintenance. However, improving the design standards of bridges, as explored under the BAU + climate action scenario, is expected to reduce incremental bridge damage across the climate scenarios considered. This adaptation is also likely to slightly reduce labor supply impacts from delays. In total, interventions explored under the BAU + climate action scenario are estimated to cost nearly $7 million in the period from 2021–50, suggesting that annual upgrade costs likely exceed the monetized benefits of avoiding the most severe expected annual incremental damage. However, while costs consider the full investment amount, the stream of benefit only consider those realized until 2050. 3.3.3. Roads Tanzania has 86,472 kilometers of roads, consisting of 12,786 kilometers that are considered trunk roads and 21,105 kilometers of regional roads (International Trade Administration 2022b; Logistics Cluster 2020). District, urban, and feeder roads make up the remaining 52,581 kilometers. The road network is relied on by 90 percent of passenger travel and 75 percent of freight traffic, making it the country’s primary Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  49 means of transport. Trunk and regional roads are managed by the Tanzania National Roads Agency, while the Prime Minister’s Office Regional Administration and Local Government is responsible for the remaining roads. Limited rehabilitation and maintenance funds due to high construction costs and lack of institutional coordination result in high accident rates from poor road conditions. In 2018, about 67 percent of trunk roads and 6 percent of regional roads were paved (TANROADS 2018). Transport infrastructure in Tanzania is historically susceptible to extreme weather. Flooding events in 2009–10 caused severe damage to roads and bridges in the Morogo and Dodoma regions, and heavy precipitation experienced by Morogo in 2011 destroyed six bridges and harmed multiple roads (United Republic of Tanzania, Vice President’s Office 2012). Increased extreme precipitation events due to climate change therefore exacerbate possible damage to roads in Tanzania. Overview of impact channel Climate change may impact road infrastructure in a variety of ways. These include increased temperatures, precipitation, and flooding that cause roads to deteriorate faster, which in turn influences infrastructure repair and maintenance costs and causes delays for passengers. Our analysis relies on the Infrastructure Planning Support System, which models the impact of daily temperatures, precipitation, and flooding stressors on paved, gravel, and unpaved roads. Generally, temperature impacts only paved roads, precipitation impacts both paved and unpaved roads, and flooding impacts all kinds of roads. For roads, the analysis examines the impacts of climate change in the form of increased repair and maintenance costs incurred as a result of climate damage. The need for increased repair and maintenance can be due to accelerated aging of binder, rutting of asphalt, bleeding of seals due to temperature, reduced carrying capacity due to increased moisture from precipitation, and washaways and overtopping from flooding. Repair and maintenance costs for road infrastructure are based on unit cost assumptions. A more detailed description of the modeling methodology used is available in appendix B, with appendix C offering details on the data sources utilized in the analysis. The total impacts in terms of maintenance and repair costs as well as delays are estimated under a no adaptation BAU scenario, as well as three adaptation development scenarios described in table 16. Table 16: Adaptation scenarios evaluated for the roads channel BAU BAU + climate action ASP ASP + climate action Damages due to climate Assumes 750 km of tertiary Assumes 750 km of tertiary roads Assumes 1,168 km of tertiary roads change are incurred roads are paved and 250 km of are paved and 250 km of paved are paved and 374 km of paved assuming no changes to the paved highways are upgraded highways are upgraded by 2050. highways are upgraded by 2050. current road stock. Assumes by 2050. Assumes current current design standards of design standards of road Assumes proactive climate Assumes proactive climate road infrastructure remain infrastructure remain constant. resilience measures are resilience measures are constant. implemented throughout the entire implemented throughout the entire road network, excluding those road network, excluding those receiving the upgrades described receiving the upgrades described above. Measures are implemented above. Measures are implemented once existing infrastructure reaches once existing infrastructure reaches its end of life or needs rehabilitation its end of life or needs rehabilitation after damage has occurred. Climate after damage has occurred. Climate resilience measures improve resilience measures improve the ability of roads to withstand the ability of roads to withstand higher levels of temperature higher levels of temperature and precipitation, as well as the and precipitation, as well as the magnitude of a future 50-year magnitude of a future 50-year flooding event. flooding event. 50  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Results As shown in the left panel of figure 48, from 2041 to 2050, the annual incremental road costs and delays due to climate change under the BAU scenario are expected to worsen under both GCM ensemble means with respect to the historical baseline. Under the SSP2–4.5 and SSP3–7.0 GCMs on average, annual costs in the 2040s increase by approximately $68 million and $64 million relative to the historical baseline, respectively. This damage is driven predominantly by flooding impacts (accounting for approximately 64 and 63 percent of total damage under the SSP2–4.5 and SSP3–7.0 ensemble means, respectively), followed by precipitation- and temperature-related damage. This is expected given the country’s high proportion of secondary gravel roads which are most vulnerable to the impacts of flooding and precipitation and aligns with the notable effect of flooding on infrastructure projected in the inland flooding channel. During the period from 2041–50, we also anticipate an increase in road delays of about 27 million hours under both the SSP2–4.5 and SSP3–7.0 climate means, or a labor supply shock of approximately -0.07 percent on average (figure 48b). Figure 48: Annual incremental road costs and delays under BAU for 2041–50 a) Cost ($, millions) b) Delay (hours, millions) 100 40 80 30 Annual incremental changes 60 20 40 10 20 0 0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 mean SSP3-7.0 mean SSP2-4.5 25th-75th percentile SSP3-7.0 25th-75th percentile Figure 49 explores which parts of the country are expected to see the greatest additional road damage under the BAU scenario compared to the historical period. Under both ensemble means, notable incremental damage is expected across most of the country, except for the central-eastern regions of Manyara, Singida, and Dodoma, where impacts are limited to $500–1,500 per kilometer in the period from 2041–50, regardless of climate scenario. However, for each of the ensemble means, we expect road impacts to be most severe in separate parts of the country due primarily to differences in the distribution of flooding and precipitation impacts for each climate outcome. For the SSP2–4.5 ensemble mean the highest damage is expected in southeastern regions of Morogoro and Iringa, whereas the SSP3–7.0 scenario mean sees comparably high damage in the northern regions of Geita and Mara and the southeastern most region of Mtwara instead. This peak damage reaches $3,000–3,100 by the 2040s. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  51 Figure 49: Additional annual damage in 2041–50 relative to 1995–2020, by region, under BAU a) SSP2-4.5 mean b) SSP3-7.0 mean Damage ($1,000s) per km 3.058511 3.0 2.5 2.0 1.5 1.0 0.5 0.000000 As described above for the BAU scenario alone, under both BAU and BAU + climate action, average annual costs due to increased road repair and maintenance between 2041 and 2050 are projected to total between $63 and 68 million under the SSP3–7.0 and SSP2–4.5 means, respectively (figure 50a and b). The additional upgrades applied to a subset of roads under the BAU + climate action scenario places incremental costs on the lower end of this range relative to BAU. With investment in adaptation measures to withstand higher precipitation intensities, floods of greater magnitude, and rising temperatures (figure 50c and d), incremental annual costs for the SSP2–4.5 and SSP3–7.0 means are reduced by nearly 90 percent under both SSPs to approximately $8 million. Across both ensembles, these reductions in total road costs are driven by investments in more resilient road infrastructure, making them more resistant to the significant flooding impacts expected in the future under climate change. Figure 50: Projected annual incremental road costs by development scenario, 2041–50 a) BAU b) BAU with climate action c) ASP d) ASP with climate action 100 Incremental annual cost ($, millions) 80 60 40 20 0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 mean SSP3-7.0 mean SSP2-4.5 25th-75th percentile SSP3-7.0 25th-75th percentile 52  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania The impact of investing in adaptation has a similar effect on commuting delays projected for 2041–50 (figure 51a). Under the ASP and ASP+NDC development scenarios, relative to the approximately 27 and 26 million incremental annual delay hours under BAU and BAU + climate action, additional annual delays by the 2040s are estimated at between 7 and 8 million, signifying a reduction in delay hours of between approximately 69 and 74 percent. This reduction corresponds to a labor supply shock of approximately -0.02 percent under ASP and ASP + climate action for both climate ensembles. Figure 51: Projected annual incremental delays by adaptation scenario, 2041–50 a) BAU b) BAU with climate action c) ASP d) ASP with climate action 40 Incremental annual delays (hours, millions) 30 20 10 0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 SSP3-7.0 SSP2-4.5 mean SSP3-7.0 mean SSP2-4.5 25th-75th percentile SSP3-7.0 25th-75th percentile Costs of Adaptation Under the ASP and ASP + climate action development scenarios, proactive climate resilience measures are implemented to improve road infrastructure at the end of its lifespan or during rehabilitation. These measures provide additional protection against increases in temperature, precipitation, and flooding extremes due to climate change. Under the ASP + climate action scenario, 418 additional kilometers of tertiary roads are paved. For ASP, adaptation measures are estimated to cost $29–30 million per year on average from 2041–50, or $929–959 million total from 2024–50, under the SSP3–7.0 and SSP2–4.5 ensemble means, respectively. For ASP + climate action, these measures instead cost $28–29 million per year on average from 2041–50, or $908–938 million from 2024–50, under the SSP3–7.0 and SSP2–4.5 ensemble means, respectively. The difference in estimated costs between the ASP and ASP + climate action scenarios results from a greater proportion of paved roads under the ASP + climate action scenarios. Unpaved road infrastructure is more costly to improve than paved road infrastructure; the greater proportion of unpaved roads under the ASP scenario is therefore more expensive to improve for climate resilience. However, paving roads is not costed under any of the BAU + climate action, ASP and ASP + climate action scenarios. With consideration of costs for the greater proportion of paving and road upgrades under the ASP + climate action scenario, total costs would likely exceed costs under the ASP scenario. Discussion and key takeaways Under projected climate conditions, flooding is expected to be the primary driver of road damage in Tanzania, which further demonstrates the anticipated capital damage discussed under the inland flooding channel. Without adaptation, increased flooding, precipitation events, and temperatures may increase annual damage to roads by approximately $68 million per year under the SSP2–4.5 ensemble mean by the period Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  53 from 2041–50. Building roads with proactive adaptation measures could lead to a reduction in this damage by close to 90 percent, lowering costs to $8 million annually on average by mid-century. With adaptation costs ranging between approximately $28 and 30 million annually by the 2040s depending on the extent of proactive climate resilience measures implemented, the estimated reduction in incremental costs would provide net benefits for investment in these road infrastructure updates. Similarly to bridge infrastructure impacts, the effects of delays from climate events are limited in terms of labor supply, but the significant reduction in incremental hours of annual delays from adaptation could provide additional social and logistical benefits that are not quantified in this analysis. 54  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania 4. Conclusion The objective of this report is to outline the process and present the results of estimating the economic damage of climate change for Tanzania. This was done by first selecting a representative set of climate scenarios, used to assess the macroeconomic effects under a range of future climate conditions. Macroeconomic shocks arising from relevant “channels of impact” were then produced under a BAU scenario that assumed growth and development trends in Tanzania continue on a similar trajectory as recent years, an aspirational development scenario that considered structural transformation of the country’s economy in line with socioeconomic development objectives, and corresponding NDC scenarios where the country’s development takes into account climate change adaptation. The shocks estimated through each impact channel then served as input for a country-specific macroeconomic model. Across the channels considered, climate change is expected to have varying degrees of impacts over time and across development scenarios, as summarized in table 17.1 Table 17: Summary of estimated shocks by channel by the 2040s BAU+ climate ASP + climate BAU ASP action action Human health, 2041–50 Wet/warm mean -0.06% 1.84% 0.39% Labor supply Dry/hot mean -0.14% 1.83% 0.33% Heat stress on labor productivity, 2041–50 Wet/warm mean -2.5% -0.8% -2.5% -0.8% Agriculture Dry/hot mean -4.1% -2.2% -4.1% -2.2% Wet/warm mean -2.5% -2.5% -1.4% -1.4% Industry Dry/hot mean -3.4% -3.4% -2.1% -2.1% Wet/warm mean -1.1% -1.1% -0.9% -0.9% Services Dry/hot mean -1.8% -1.8% -1.5% -1.5% Investment costs ($, millions, 5.85 5.85 industry/services only) Natural capital channels, 2041–50 Wet/warm mean 5.7% 2.4% 5.7% 1.7% Rainfed crops Dry/hot mean -7.5% -9.6% -7.3% -10.2% Investment costs ($, millions) 153.1 366.4 153.1 366.4 Wet/warm mean -1.3% -1.3% -0.1% -0.1% Soil erosion Dry/hot mean -0.4% -0.4% 0.3% 0.3% 1 Due to differences in how individual channels are modeled and shocks are quantified, the absolute value of the shocks estimated for individual channels are not directly comparable. Across all channels, negative shocks indicate worsening, while positive shocks indicate improvement relative to the baseline. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  55 BAU+ climate ASP + climate BAU ASP action action Wet/warm mean 4.3% 0.8% 5.4% 1.4% Rainfed + erosion Dry/hot mean -8.6% -10.9% -8.0% -10.8% Wet/warm mean -2.8% -2.6% -0.4% -0.5% Livestock Dry/hot mean -7.3% -6.8% -2.3% -2.3% Investment costs ($, millions) 282.4 329.9 Inland flooding, 2041–50 SSP2–4.5 mean -0.064% -0.066% -0.007% -0.007% Expected capital stock damage SSP3–7.0 mean -0.060% -0.061% -0.003% -0.004% Investment costs ($, millions) 89.3 89.3 Bridges shocks, 2041–50 SSP2–4.5 mean -0.002% -0.001% Labor supply (expected) SSP3–7.0 mean -0.002% -0.001% SSP2–4.5 mean 3.5 2.5 Expenses (expected) SSP3–7.0 mean 2.9 2.0 Investment costs ($, millions) 6.8 Roads shocks, 2041–50 SSP2–4.5 mean -0.074% -0.025% -0.071% -0.024% Labor supply (%) SSP3–7.0 mean -0.074% -0.022% -0.071% -0.021% SSP2–4.5 mean 67.6 8.3 65.9 7.9 Expenses ($, millions) SSP3–7.0 mean 64.3 8.5 62.8 8.0 Investment costs ($, millions) 29.3 28.3 Increasing temperatures are expected to cause significant heat stress across sectors in Tanzania: • Significant heat stress on labor productivity is anticipated by mid-century, particularly in agriculture. While comparatively lower impacts for industry and services are expected, the concentration of workers in the hot and humid Dar es Salaam region will cause more severe heat stress productivity shocks relative to similar countries. Regarding human health, significant increases in waterborne and heat-related illnesses are also driven by increasing heat. Temperature changes also affect malaria transmission but with minor incidence reductions, resulting in benefits under a drier, hotter climate. • Potential migration of these coastal populations inland towards Lake Victoria, as well as reductions in the fraction of workers in agriculture, are expected to reduce the overall exposure to heat stress in Tanzania’s labor force. However, better conditions for malaria transmissibility in the Lake Victoria region, both currently and in the future, could reduce some of these gains. • Heat effects on agriculture can affect crop, cattle milk, and chicken egg productivity. Vegetables are the most affected crop, with shocks reaching up to -18 percent under a dry/hot climate scenario. 56  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Livestock products are less vulnerable to high temperatures, resulting in impacts reaching about 5 percent under a hot and dry future. Overall, agriculture is more susceptible to changes in precipitation, driving uncertain net climate effects on the sector that fluctuate from negative to positive by mid-century: • For rainfed crops, changes in rainfall and subsequent water deficit produce substantial shocks to yields. Across most crops, precipitation changes result in uncertain effects by mid-century, with expected productivity declines under a dry and hot future and gains under a wet and warm future. While less significant, further yield impacts would result from increased topsoil erosion under wetter conditions. Still, considering both effects, a warm and wet future is expected to produce yield gains on rainfed crops. • Changes in the location of harvested areas derived from land use and land cover changes, in turn driven by sectoral and macroeconomic factors, can have consequential impacts on the overall vulnerability of rainfed crop yields. Under an aspirational development future, yields are expected to be more vulnerable to overall climate effects than under a BAU future. However, while yields are more vulnerable, total production is expected to increase under an ASP scenario due to the higher investments in irrigation and agricultural productivity, compensating for higher climate vulnerability regardless of the climatic conditions. • In the livestock sector, changes in rainfall can negatively affect pasture availability for grazing animals, leading to a reduction in meat productivity regardless of the climate projections. Cattle, goat, and sheep meat, which comprise 60 percent of the total revenues in the livestock sector, can experience negative shocks as high as -13 percent by mid-century. • In the agriculture sector overall, an aspirational development scenario is expected to produce the largest gains in labor productivity, and moderate improvements in livestock yields relative to a BAU development scenario. However, crop yield shocks worsen relative to a BAU due to spatial reallocation of crops into more affected areas. Flooding impacts are likely to increase under climate change, although incremental impacts are generally lower than expected flood damage under current conditions: • Inland flooding damage under the current climate can be significant, reaching over 2.5 percent of the national capital under a 100-year event. Overall, the change in capital damage is expected to be highest for northern, northeastern, and certain southern regions, reaching above a 60 percent increase relative to the current climate. Despite differences in land use land cover projections, including changes in forest cover, the Mara, Tanga, Njombe, and Dar es Salaam regions show consistent damage increases. While localized and event-specific flooding impacts by mid-century can be substantial, annualized national-level damage due to climate change represents a small fraction of the total capital stock. • Flooding events are the primary driver of significant incremental damage expected for bridges and roads. Combined expenses can reach over $60 million per year in additional repair and maintenance costs, in addition to the capital losses from inland flooding. These disruptions will also result in a significant number of hours lost in traffic. However, these hours represent a minor fraction of the total labor force. Still, downstream effects on business interruptions caused by road and bridge disruptions, not quantified in this analysis, could further increase the economic impacts of climate change. • Floodproofing buildings and upgrading road and bridge infrastructure are effective climate adaptation strategies that can significantly reduce climate change effects by mid-century. However, capital investments can be large, therefore, careful planning is needed depending on infrastructure value, lifespan, maintenance, and specific site conditions. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  57 References Adegun, O, Mbuya, E C and Njavike, E. 2022. “Responses to Heat Stress Within an Unplanned Settlement in Dar Es Salaam, Tanzania.” Frontiers in Built Environment 8. https://doi.org/10.3389/fbuil.2022.874751. Arce, C E and Caballero, J. 2015. Tanzania: Agricultural Sector Risk Assessment. Washington DC: World Bank Group. https://documents1.worldbank.org/curated/en/248961468001158010/pdf/94883-REVISED-Box393201B-PUBLIC- Tanzania-Ag-Risk-Assessment-web-6-29-15-JK-jtc.pdf. BLS. 2022. The Occupational Requirements Survey (ORS). Washington, DC: U.S. Bureau of Labor Statistics (BLS). https://www.bls.gov/ors/. C40 Cities Finance Facility. 2020. Flooding in East Africa: The Impacts on and Implications for Cities in East Africa, with a Case Study of Dar Es Salaam. https://www.thegpsc.org/sites/gpsc/files/cff_report_design_-_dar_es_salaam- final_2020_-_spread-compressed.pdf. C40 Cities Finance Facility. 2021. Flood Management Feasibility Study for the Msimbazi Middle Catchment Area: Vulnerability Analysis. https://cff-prod.s3.amazonaws.com/storage/files/q3G45F8w7Vcas31GdF7NqdDukxLjBcwjSw3U1Bxe.pdf. Chapman, S, Birch, C E, Galdos, M V, Pope, E, Davie, J, Bradshaw, C, Eze, S and Marsham, J H. 2021. “Assessing the Impact of Climate Change on Soil Erosion in East Africa Using a Convection-Permitting Climate Model.” Environmental Research Letters 16 (8). https://doi.org/10.1088/1748-9326/ac10e1. de Glanville, W A, Davis, A, Allan, K J, Buza, J, Claxton, J R, Crump, J A, Halliday, J E B, Johnson, P C D, Kibona, T J, Mmbaga, B T, Swai, E S, Uzzell, C B, Yoder, J, Sharp, J and Cleaveland, S. 2020. “Classification and Characterisation of Livestock Production Systems in Northern Tanzania.” PLoS One 15(12). https://doi.org/10.1371/journal.pone.0229478. FAO. n.d.-a. Family Farming Knowledge Platform: Tanzania, accessed March 25, 2024. https://www.fao.org/family- farming/countries/tza/en/. FAO. n.d.-b. Tanzania at a Glance, accessed March 25, 2024. https://www.fao.org/tanzania/fao-in-tanzania/tanzania- at-a-glance/en/. HICGI News Agency. 2024. “Tanzania Shuts down a Key Highway Due to Bridge Damage Caused by Floods,” May 7. https://hicginewsagency.com/2024/05/07/tanzania-shuts-down-a-key-highway-due-to-bridge-damage-caused-by- floods/. IHME. 2019. United Republic of Tanzania. Health Research by Location. 2019. https://www.healthdata.org/research- analysis/health-by-location/profiles/tanzania. ILOSTAT. 2022a. Employment in Agriculture (% of Total Employment) (Modeled ILO Estimate). The World Bank DataBank. 2022. https://data.worldbank.org/indicator/SL.AGR.EMPL.ZS. ILOSTAT. 2022b. Employment in Industry (% of Total Employment) (Modeled ILO Estimate). The World Bank DataBank. 2022. https://data.worldbank.org/indicator/SL.IND.EMPL.ZS. ILOSTAT. 2022c. Employment in Services (% of Total Employment) (Modeled ILO Estimate). The World Bank DataBank. 2022. https://data.worldbank.org/indicator/SL.SRV.EMPL.ZS. IMF. 2023. United Republic of Tanzania: Building Resilience to Climate Change.” IMF Country Report 23/154. International Trade Administration. 2022a. Tanzania Country Commercial Guide: Agriculture and Agricultural Processing. December 14. https://www.trade.gov/country-commercial-guides/tanzania-agriculture-and-agricultural-processing. International Trade Administration. 2022b. Tanzania Country Commercial Guide: Construction. December 14. https://www.trade.gov/country-commercial-guides/tanzania-construction#:~:text=The%20road%20network%20 in%20Tanzania,district%2C%20urban%20and%20feeder%20roads. International Trade Administration. 2022c. Tanzania Country Commercial Guide: Market Overview. December 14. https://www.trade.gov/country-commercial-guides/tanzania-market-overview. 58  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Ires, I. 2021. “Intensive Agriculture as Climate Change Adaptation? Economic and Environmental Tradeoffs in Securing Rural Livelihoods in Tanzanian River Basins.” Frontiers in Environmental Science 9. https://doi.org/10.3389/ fenvs.2021.674363. Kajeguka, D C, Mponela, F M, Mkumbo, E, Kaaya, A N, Lasway, D, Kaaya, R D, Alifrangis, M, Elanga-Ndille, E, Mmbaga, B T and Kavishe, R. 2023. “Prevalence and Associated Factors of Dengue Virus Circulation in the Rural Community, Handeni District in Tanga, Tanzania.” Journal of Tropical Medicine. https://doi.org/10.1155/2023/5576300. Kassam, A H, van Velthuizen, H T, Sloane, P H, Fischer, G W and Shah, M M. 1991. Agro-Ecological Land Resources Assessment for Agricultural Development Planning: A Case Study of Kenya. Resources Data Base and Land Productivity. Rome: Food and Agriculture Organization of the United Nations (FAO) and International Institute for Applied Systems Analysis (IIASA). Mboera, L E, Rumisha, S F, Lyimo, E P, Mangu, C D, Mremi, I R, Kumalija, C J, Joachim, C, Kishamawe, C, Massawe, I S, Matemba, L E, Kimario, E, Bwana, V M and Mkwashapi, D M. 2018. “Cause-Specific Mortality Patterns among Hospital Deaths in Tanzania, 2006-2015.” PLoS One 13(10). https://doi.org/10.1371/journal.pone.0205833. Meshi, E B, Kishinhi, S S, Mamuya, S H and Rusibamayila, M G. 2018. “Thermal Exposure and Heat Illness Symptoms among Workers in Mara Gold Mine, Tanzania.” Annals of Global Health 84(3): 360–68. https://doi.org/10.29024/ aogh.2318. Odjidja, E N, Gatasi, G and Duric, P. 2019. “Delivery of Integrated Infectious Disease Control Services under the New Antenatal Care Guidelines: A Service Availability and Readiness Assessment of Health Facilities in Tanzania.” BMC Health Services Research 19 : 153. https://doi.org/10.1186/s12913-019-3990-8. Rigaud, K K, de Sherbinin, A, Jones, B, Bergmann, J, Clement, V, Ober, K, Jacob Schewe, Adamo, S, McCusker, B, Heuser, S and Midgley, A. 2018. Groundswell: Preparing for Internal Climate Migration. Washington DC: World Bank. https:// doi.org/10.1596/29461. Stephen, G and Makyara, J. 2022. “Climate Change: Evidence and Policy Implications on Road Infrastructure.” Presented at the Workshop on Adaptation of Impacts of Climate Change in Management of Road Assets: Institute of Bank of Tanzania - Mwanza, July. https://www.roadsfund.go.tz/uploads/publications/sw-1658755232-Adaptation%20to%20 Impact%20of%20Climate%20Change%20on%20Road%20Infrastructure.pdf. St-Pierre, N R, Cobanov, B and Schnitkey, G. 2003. “Economic Losses from Heat Stress by US Livestock Industries.” Journal of Dairy Science 86: E52–77. https://doi.org/10.3168/jds.S0022-0302(03)74040-5. TANROADS. 2018. Trunk and Regional Roads Network. https://www.tanroads.go.tz/uploads/files/Trunk_and_Regional_ Road_Map.pdf. UNDP. 2023. “Tanzania Economic Transformation: Key Developments.” November 3. https://www.undp.org/tanzania/ blog/tanzania-economic-transformation-key-developments. United Republic of Tanzania, Ministry of Livestock and Fisheries. 2017. Tanzania Livestock Master Plan (2017/2018- 2021/2022). https://www.mifugouvuvi.go.tz/uploads/projects/1553601793-TANZANIA%20LIVESTOCK%20MASTER% 20PLAN.pdf. United Republic of Tanzania, Ministry of Livestock and Fisheries. 2022. Livestock Sector Transformation Plan (LSTP): 2022/23–2026/27. https://www.mifugouvuvi.go.tz/uploads/publications/sw1675840376-LIVESTOCK%20 SECTOR%20TRANSFORMATION%20PLAN%20(LSTP)%20202223%20-%20202627.pdf. United Republic of Tanzania, Vice President’s Office. 2012. National Climate Change Strategy. https://faolex.fao.org/ docs/pdf/tan191137.pdf. USAID. 2021. Tanzania Water Resources Profile Overview. Water Resources Profile Series. https://winrock.org/wp- content/uploads/2021/08/Tanzania_Country_Profile-Final.pdf. USAID. 2023. Agriculture: Tanzania. Feed the Future. https://www.usaid.gov/sites/default/files/2023-09/Tanzania%20 Agriculture%20Factsheet_1.pdf. USAID. n.d. Economic Growth and Trade, accessed March 25, 2024. https://www.usaid.gov/tanzania/economic-growth- and-trade. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  59 USDA Foreign Agriculture Service. 2024. Country Summary: Tanzania. International Production Assessment Division, accessed March 8, 2024. https://ipad.fas.usda.gov/countrysummary/default.aspx?id=TZ. WHO. 2014. Quantitative Risk Assessment of the Effects of Climate Change on Selected Causes of Death, 2030s and 2050s. World Health Organization. https://apps.who.int/iris/handle/10665/134014. Wolf, J, Johnston, R, Hunter, P R, Gordon, B, Medlicott, K and Prüss-Ustün, A. 2019. “A Faecal Contamination Index for Interpreting Heterogeneous Diarrhoea Impacts of Water, Sanitation and Hygiene Interventions and Overall, Regional and Country Estimates of Community Sanitation Coverage with a Focus on Low- and Middle-Income Countries.” International Journal of Hygiene and Environmental Health 222(2): 270–82. https://doi.org/10.1016/j.ijheh.2018.11.005. World Bank. 2021. Climate Change Knowledge Portal. https://climateknowledgeportal.worldbank.org/. World Bank. 2022a. Global Scenarios for CCDR Analyses. World Bank. 2022b. Macro Poverty Outlook for Tanzania. https://documents.worldbank.org/en/publication/documents- reports/documentdetail/099455304262241826/idu06c49cdf10d039042fd0981d097317cae17ca. WWF. 2018. “Tanzania Announces Historic Move to Restore 5.2 Million Hectares of Degraded Land,” August 13. https://wwf.panda.org/wwf_news/?332953/Tanzania-announces-historic-move-to-restore-52-million-hectares-of- degraded-land#:~:text=A%20recent%20assessment%20report%20by,pollution%2C%20and%20loss%20of%20 biodiversity. Wynants, M, Patrick, A, Munishi, L, Mtei, K, Bodé, S, Taylor, A, Millward, G, Roberts, N, Gilvear, D, Ndakidemi, P, Boeckx, P and Blake, W H. 2021. “Soil Erosion and Sediment Transport in Tanzania: Part II – Sedimentological Evidence of Phased Land Degradation.” Earth Surface Processes and Landforms 46(15): 3112–26. https://doi.org/10.1002/esp.5218. 60  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Appendix A: Climate Scenario Selection The evaluation of climate change impacts conducted in this study required climate information as an input to the modeling process. This appendix details the process of generating and processing the necessary climate scenarios. A1. Climate data and projections used To develop climate scenarios and future projections, historical and future projected datasets were obtained, with precipitation and temperature the main variables of interest for this study. Historical monthly climate data was obtained for 1950 to 2020 from the Climatic Research Unit gridded Time Series of the University of East Anglia (CRU TS 4.05) (Harris et al. 2020). These data are available from 1959 to 2020 at a spatial resolution of 0.5 x 0.5-degree grids and monthly temporal resolution for various variables including mean, maximum and minimum temperature, and total precipitation. Future climate projections were obtained from the World Bank’s Climate Change Knowledge Portal for 29 General Circulation Models (GCMs) from the Coupled Model Intercomparison Project 6 (CMIP6) suite of Intergovernmental Panel on Climate Change model outputs (World Bank 2021). This large suite of GCMs was run for a set of emissions scenarios, as shown in figure A1. Figure A1: CMIP6 SSPs 140.0 120.0 100.0 80.0 Gigatonnes CO2 60.0 40.0 20.0 0.0 -20.0 1980 2000 2020 2040 2060 2080 2100 Historical SSP1-1.9 SSP1-2.6 SSP4-3.4 SSP5-3.40s SSP2-4.5 SSP4-6.0 SSP3-7.0 SSP5-8.5 Notes: Shaded area shows range of no-policy baseline scenarios. Interactive version with more data available at Carbon Brief: https://www.carbonbrief.org/cmip6- the-next-generation-of-climate-models-explained Table A1 shows a comprehensive list of the available data utilized for this project organized by GCM and historical data/SSP. Each GCM has up to five combinations of SSP and RCP emissions scenario runs, including SSP1–RCP 1.9, 1–2.6, 2–4.5, 3–7.0, and 5–8.5. For each GCM-SSP combination, the Climate Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  61 Change Knowledge Portal provides a modeled history from 1995 to 2014 and projections from 2015 to 2100. For the selection of climate scenarios, we employed 31 GCMs, originally provided at a spatial resolution of 1 x 1-degree grids for the globe and monthly temporal resolution for years 1995 to 2100 and monthly mean daily temperature and monthly total precipitation. Table A1: List of GCMs and the associated SSPs available for this study Model Hist SSP1-1.9 SSP1-2.6 SSP2-4.5 SSP3-7.0 SSP5-8.5 access-cm2      access-esm1-5      awi-cm-1-1-mr     bcc-csm2-mr      cams-csm1-0       canesm5       cesm2  cmcc-cm2-sr5      cmcc-esm2      cnrm-cm6-1      cnrm-esm2-1       ec-earth3       ec-earth3-veg       fgoals-g3       gfdl-esm4       hadgem3-gc31-ll     inm-cm4-8      inm-cm5-0      ipsl-cm6a-lr       kace-1-0-g      kiost-esm     miroc-es2l       miroc6       mpi-esm1-2-hr      mpi-esm1-2-lr      mri-esm2-0       nesm3     noresm2-lm      noresm2-mm      taiesm1      ukesm1-0-ll       Count 30 12 30 30 27 30 62  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania A2. Processing of climate information Given that GCMs are biased relative to observed climate conditions, we applied the bias correction and spatial disaggregation technique to disaggregate the projections to 0.5 x 0.5-degree grid cells, and to then bias correct these projections using the observed historical dataset from 1995 to 2000 from the CRU TS 4.05 dataset. For each grid cell, the bias correction procedure sets up “quantile maps” for each month to statistically compare the GCM hindcast to the CRU observations, and then uses those maps to bias correct all projections. This approach was previously applied in the World Bank’s Enhancing the Climate Resilience of Africa’s Infrastructure study (Cervigni et al. 2015), using the CMIP5 ensemble. We use standard spatial downscaling practice to reduce the spatial resolution from 1 x 1-degree grids to 0.5 x 0.5-degree grids and properly line the data up with the historical data, both reduced to the 67,420 CRUs of the CRU TS 4.05 data covering the land area of the globe. This includes the following steps for each file: • Obtain the raw data (resulting spatial dimensions are 181 latitude x 360 longitude).2 • Perform Inverse Distance Weighting with a multiplier of 2, power of 2, and radius of 2 to increase the resolution to 0.5 x 0.5 degree (resulting spatial dimensions are 362 latitude x 720 longitude). • Remove the 1st and 362nd latitude rows, because centroids of the original 181 one-degree latitude bands range from -90 to +90, which means that latitude bands at -90 and +90 are each only 0.5, rather than 1, degree. As a result, removing the 1st and 362nd reduces the size of these bands to 0.5 degree (resulting spatial dimensions are 360 latitude x 720 longitude). • Reduce and vectorize the data to the 67,420 CRUs covering the land areas of the glove (resulting spatial dimensions are 67,420 CRUs). Figure A2 presents a visual representation of a few of these steps for a (randomly selected) scenario, ACCESS-CM2-SSP2–4.5. 2 The raw data from the World Bank have a spatial resolution of 181 latitude x 361 longitude. Based on communication with the Climate Change Knowledge Portal team, the 361st longitude column was removed. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  63 Figure A2: Spatial downscaling methodology for ACCESS-CM2-SSP2–4.5 a. Result of Step 1 NetCDF to MAT (181 X 361) Annual pr (mm) NetCDF to MAT (181 X 361) Annual tas (ºC) 2500 40 20 20 30 40 2000 40 20 60 60 1500 80 80 10 100 100 0 1000 120 120 -10 140 140 500 160 160 -20 180 0 180 -30 50 100 150 200 250 300 350 50 100 150 200 250 300 350 b. Result of Step 3 Spatially Disaggregated IDW (360 X 720) Annual tas (ºC) Spatially Disaggregated IDW (360 X 720) Annual pr (mm) 2500 40 50 50 30 100 2000 100 20 150 1500 150 10 200 200 0 1000 250 250 -10 300 500 300 -20 350 0 350 -30 100 200 300 400 500 600 700 100 200 300 400 500 600 700 c. Result of Step 5 Vectorized (67420 X 1) Annual pr (mm) Vectorized (67420 X 1) Annual tas (ºC) 2500 40 50 50 30 2000 100 100 20 150 1500 150 10 200 200 0 1000 250 250 -10 500 300 300 -20 350 350 0 -30 100 200 300 400 500 600 700 100 200 300 400 500 600 700 64  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania A3. Selection of climate scenarios For this study, we analyzed the available set of GCM/SSP permutations previously shown in figure A3 to obtain a subset of scenarios that represent an appropriate range of possible future climate conditions. In particular, we consider two different sets of climate futures: one to assess the impact of uncertain global mitigation efforts and one to assess local climate risks and overall model uncertainty.3 While scenarios that capture model uncertainty are relevant for any impact channel, certain inputs may be limited to SSP aggregates only. In those cases, the analysis relies on scenarios of global mitigation efforts alone. The first set of scenarios (i.e., scenarios selected to allow for comparisons across emissions trajectories, referred to as mitigation scenarios) are selected in accordance with World Bank guidance which recommends selecting an optimistic and a pessimistic scenario of GHG concentrations that are driven by global GHG emissions trajectories and mitigation policies (World Bank 2022). For these, we use the SSP3–7.0 ensemble mean as a pessimistic case and the SSP1–1.9 ensemble mean as an optimistic case. SSP1–1.9 represents reductions in GHG emissions in line with 1.5°C warming by 2100. SSP3–7.0 is a scenario in which warming reaches 4°C by 2100, due to lax climate policies or a reduction in ecosystems and oceans’ ability to capture carbon. For the second set of scenarios (i.e., scenarios selected to assess overall model uncertainty), we select a subset of extreme GCM runs that represent a “dry and hot” and a “wet and warm” future for the country under analysis, for the study period between 2020 and 2050. This process is made up of the following steps: 1. Calculate country-scale changes in mean annual temperature and mean total precipitation between 2031 and 2050 versus the historical baseline of 1995 to 2020. 2. Consider GCMs within the SSP2 and SSP3 ensembles (about 50 total), as potential candidates for extreme conditions. This eliminates from consideration the aggressive mitigation pathway (SSP1) and the aggressive emissions pathway (SSP5). 3. Select three dry/hot scenarios around the 10th percentile of change in mean precipitation (i.e., dry) and 90th percentile change in mean temperature (i.e., hot), across all GCMs (within SSP2–4.5 and 3–7.0). Compute a 4th scenario as the mean across the three selected GCM/SSP runs. 4. Select three wet/warm scenarios around the 90th percentile of change in mean precipitation (i.e., wet) and 10th percentile change in mean temperature (i.e., warm), as above, and a 4th scenario as the mean. Figure A3 shows a visual representation of these steps for an illustrative country. 3 Climate model uncertainty: Diverse GCMs have been developed drawing on the best available science. These continue to evolve, and while sophisticated, they remain imperfect tools. Each model is unique and generates slightly different projections, even when run using identical GHG emissions scenarios. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  65 Figure A3: Climate scenario selection process a. Results from Step 1 1.6 1.4 Change in Temperature (°C) 1.2 1 0.8 0.6 0.4 -4 -2 0 2 4 6 8 10 Percent Change in Precipitation (%) ssp245 ssp370 b. Results from Steps 3 and 4 66  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania A4. Daily interpolation of monthly data Climate data from the Climate Change Knowledge Portal is available at a monthly timestep. However, many of the biophysical models used in this study rely on daily temperatures and precipitation. For the selected climate projections as well as historical baseline, we interpolate the already downscaled and bias-corrected monthly data to a daily timestep using a historical hindcast from 1948 to 2008 at a 0.5 x 0.5-degree gridded resolution from the Terrestrial Hydrology Research Group from Princeton University (Li, Sheffield and Wood 2010). The process considers the following steps: • For each 0.5 x 0.5 grid cell, we group the daily hindcast by month, for each of the twelve months. • Then, we sort each month’s data into quantiles based on the corresponding month’s total precipitation or average temperature. • For each month in the monthly data from the Climate Change Knowledge Portal, we find the quantile that month falls into and randomly select a daily value to use based on that month’s daily variability. A5. Results of the climate scenario selection process Figure A4 shows the results of the climate scenario selection process. The scatterplot shows the distribution of SSP/GCM combinations based on changes in temperature and precipitation, highlighting those that were selected (including the ensemble mean scenarios for reference). Type # SSP GCM Dry/hot future 1 SSP2–4.5 CNRM-ESM2-1 2 SSP3–7.0 EC-EARTH3-VEG 3 SSP2–4.5 GFDL-ESM4 Wet/warm future 4 SSP2–4.5 EC-EARTH3 5 SSP2–4.5 INM-CM4-8 6 SSP2–4.5 CMCC-ESM2 Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  67 Figure A4: GCM selection results for dry/hot and wet/warm futures a. 2031-2050 average projection versus 1995-2020 average historical 1.6 1.4 1 1.2 Change in Temperature (°C) dry/hot 2 ssp370 1 3 4 0.8 6 ssp119 wet/warm 5 0.6 0.4 -5 0 5 10 15 GCM selections associated with dry/hot and wet/warm futures Dry-hot Wet-warm SSP2–4.5 CNRM-ESM2-1 SSP2–4.5 EC-EARTH3 40 20 20 10 0 0 -20 -10 2030 2040 2050 2060 2030 2040 2050 2060 SSP3–7.0 EC-EARTH3-VEG SSP2–4.5 INM-CM4-8 20 20 10 0 0 -20 -10 2030 2040 2050 2060 2030 2040 2050 2060 SSP2–4.5 GFDL-ESM4 SSP2–4.5 CMCC-ESM2 40 20 20 0 0 -20 2030 2040 2050 2060 2030 2040 2050 2060 Mean Mean 10 20 0 10 -10 0 -20 2030 2040 2050 2060 2030 2040 2050 2060 Percent change in precipitation (%) ssp245 ssp370 Extreme ssp370 Targets Candidate Selected Means 68  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Trajectories of changes in temperature and precipitation for the selected scenarios are presented in figure A5 for each specific SSP/GCM combinations. Maps that show spatial changes in temperature and precipitation are presented in figure A6. Figure A5: GCM temperature (°C) and precipitation (%) change trajectories a. Temperature change trajectories GCM temperature (°C) 30% Percent Change in Precipitation (%) 20% 10% 0% -10% -20% 2020 2030 2040 2050 b. Precipitation change trajectories Precipitation (%) 2.0 1.5 Change in Temperature (°C) 1.0 0.5 0.0 -0.5 2020 2030 2040 2050 SSP2-4.5 CNRM-ESM2-1 SSP2-4.5 GRCL-ESM4 SSP2-4.5 INM-CM4-8 SSP3-7.0 EC-EARTH3-VEG SSP2-4.5 EC-EARTH3 SSP2-4.5 CMCC-ESM2 Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  69 Figure A6: GCM temperature (°C) and precipitation (%) change by 0.5-degree grid cell, 2041–50 vs. 1995–2020 a. Temperature SSP2-4.5 CNRM-ESM2-1 SSP3-7.0 EC-EARTH3-VEG SSP2-4.5 GFDL-ESM4 °C 1.6 1.4 1.2 1.0 SSP2-4.5 EC-EARTH3 SSP2-4.5 INM-CM4-8 SSP2-4.5 CMCC-ESM2 0.8 0.6 0.4 0.2 0.0 b. Precipitation SSP2-4.5 CNRM-ESM2-1 SSP3-7.0 EC-EARTH3-VEG SSP2-4.5 GFDL-ESM4 %dif 40 20 SSP2-4.5 EC-EARTH3 SSP2-4.5 INM-CM4-8 SSP2-4.5 CMCC-ESM2 0 -20 70  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania A6. References Cervigni, R, Liden, R, Neumann, J E and Strzepek, K M. 2015. Enhancing the Climate Resilience of Africa’s Infrastructure: The Power and Water Sectors. Africa Development Forum. Washington DC: World Bank. https://doi.org/10.1596/978- 1-4648-0466-3. Harris, I, Osborn, T J, Jones, P and Lister, D. 2020. “Version 4 of the CRU TS Monthly High-Resolution Gridded Multivariate Climate Dataset.” Scientific Data 7(1): 109. https://doi.org/10.1038/s41597-020-0453-3. Li, H, Sheffield, J and Wood, E F. 2010. “Bias Correction of Monthly Precipitation and Temperature Fields from Intergovernmental Panel on Climate Change AR4 Models Using Equidistant Quantile Matching.” Journal of Geophysical Research: Atmospheres 115 (D10). https://doi.org/10.1029/2009JD012882. World Bank. 2021. Climate Change Knowledge Portal. https://climateknowledgeportal.worldbank.org/. World Bank. 2022. Global Scenarios for CCDR Analyses. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  71 Appendix B: Impact Channel Methods Climate change impacts on the country’s economy are modeled through impact channels. The impact channels considered in this study are summarized in chapter 2 of the main report. For each channel, shocks to macroeconomic inputs are generated for future climate projections under multiple scenarios, also presented in chapter 2. Chapter 3 briefly describes how each impact channel was modeled and presents results for each channel. This appendix presents further technical detail on the analytical approach used to model each channel, including a description of the limitations of the methodology. Appendix C presents an overview of the specific data sources used to evaluate each impact channel. B1. Human capital Climate change may reduce human capital through increases in extreme temperatures that result in excess mortality and reduced labor capacity, as well as by facilitating the spread of infectious diseases as larger areas experience favorable climatic conditions, which in turn cause excess mortality and increased disease incidence on the population. We estimate these effects through the following channels: • Heat and labor productivity, which models changes in the ability of labor to perform work as workday temperatures increase in the future. • Human health, which models changes in the incidence and mortality of vector-borne (malaria and dengue), waterborne (i.e., cholera, dysentery, etc.), and heat-related diseases as local temperatures and precipitation levels and patterns change in the future. B1.1. Labor supply model All the channels mentioned above impact labor. Heat stress reduces the productivity of labor by reducing the effective number of hours a person can perform work, while human health and the associated policy channels impact total labor supply by way of changes in death and incidence of diseases. In order to calculate these effects, we model the total labor hours in the country, which we then shock from each effect. First, we obtain annual population estimates for both history and projected future by sex and age, considering population ages 15 to 64 as the working age population (OECD 2023). When available, we use both historical and projected future labor force statistics. If such statistics are not available, we apply available labor participation rates by sex to estimates of the working age population to compute the total labor force. The labor force is multiplied by available data on mean weekly hours worked for both males and females to produce the total annual hours of labor supply. B1.2. Heat and labor productivity Temperature directly affects the productivity of labor, where the effect intensifies for labor types that are outdoors and are conducting more intense physical work. Labor productivity impacts follow the methodology applied by the International Labor Organization (ILO 2019b), which has been applied globally and used in other similar studies, such as in Kjellstrom et al. (2018). The approach is based on workday wet bulb globe temperatures as an indicator of heat stress, which refers to the exposure of individuals to extreme heat or hot environments that lead to the body’s inability to regulate internal temperature (CDC 2020). A measure of heat stress is used to quantify the percentage of a typical working hour that a person can work. The analysis is done at a 0.5 x 0.5-degree spatial resolution for the relevant sectors of the economy, resulting in annual shocks to labor productivity for every scenario being evaluated. 72  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Methodology Workers are exposed to different temperatures during typical workday hours. The functional relationship between work ability and heat stress is quantified using wet bulb globe temperatures (measured in Celsius degrees). Wet bulb temperature is a measure of air temperature in relation to moisture content, while dry bulb temperature is a measure of ambient temperature. The globe temperature is a measure of the thermal radiation that would be absorbed by someone’s skin. We utilize different approach for estimating indoor and outdoor wet bulb globe temperatures. For outdoor conditions, we use the environmental stress index formulation by Moran et al. (2003) plus a linear bias correction factor adding 1–3°C, based on solar radiation, to compensate for underestimation under high solar radiation conditions that the environmental stress index does not capture. This correction factor is calculated based on a comparison of the environmental stress index against the Liljegren et al. (2008) formulation approach done by Kong and Huber (2022). For indoor conditions, we utilize the Kjellstrom et al. (2018) application of the Bernard and Pourmoghani (1999) formulation of wet bulb globe temperature. Both equations require data on air temperature and relative humidity. Outdoor wet bulb globe temperature also requires solar radiation, which we calculate based on latitude and longitude data of the globe and mean cloud cover. To approximate workday temperatures from daily minimum and maximum temperatures, we use the “4+4+4” method used by the International Labour Organization (ILO 2019b). This method assumes that in a typical 12-hour daylight day, four hours per day are close to minimum daily temperatures, four are close to the maximum, and four are close to the midpoint between the two. As done by the International Labour Organization, we assume these temperatures apply to indoor workers as air temperature is measured in the shade. Labor productivity effects are then estimated based on the percentage of hours that an acclimatized worker can be engaged in work based on their level of heat stress. Occupations with lower physical activity can tolerate higher levels of heat stress. Labor productivity loss curves from wet bulb globe temperatures for three levels of physical activity (measured in Watts), presented in figure B1, are derived from the ISO 7243:1989 standard (ISO 1989) and (Kjellstrom et al. 2018) and validated through additional epidemiological studies, as presented in International Labour Organization (ILO 2019a). Generally, 200 Watts represents clerical or light physical work, 300 Watts moderate physical work in industry, and 400 Watts heavy physical work in agriculture or construction. Work intensities are then matched to labor hours by sector and occupation, from available reported data. We apply available assumptions on outdoor exposure by occupation to data on employment by sector in order to split the share of hours worked indoors versus outdoors. Then, we estimate the impacts of heat stress on labor productivity for each category of workers (by sector and occupation), and for both indoor and outdoor workers. For indoor workers, we exclude the share of those who worked in spaces with air conditioning and assume they do not experience any heat stress. Adoption of air conditioning within a country (by 0.5-degree grids) is estimated using regional percent coverage of curves from Davis et al. (2021), based on mean household income levels and cooling degree days by grid cell.4 Generally, regions with higher mean household incomes and/or cooling-degree days have a higher air conditioning adoption rate. Mean household incomes by country are obtained from United Nations and World Bank data (World Bank, n.d.; United Nations 2022). When available, data on mean percent adoption of air conditioning at a national scale is used to calibrate the estimates obtained from Davis et al. (2021). 4 Cooling degree days assume that when the outside temperature is 65°F, people do not need cooling to be comfortable. Cooling degree days are the difference between the daily mean temperature minus 65°F. https://www.weather.gov/key/climate_heat_cool. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  73 Figure B1: Work capacity as a percentage from wet bulb globe temperatures 100% 75% Work ability (%) 50% 25% 0% 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 WBGT (°C) 200W 300W 400W Source: IEc analysis For the final step, monthly labor productivity impacts by 0.5-degree grid cell are aggregated nationally by macroeconomic sectors (agriculture, industry, and services) and on an annual scale for all the completed time series. For agriculture, grid cell level impacts are aggregated using the share of cropland as a proxy of the spatial distribution of agricultural workers, using data from the Global Agriculture Lands dataset (Ramankutty et al. 2010). For industry, we assume a distribution of workers using gridded gross domestic product data from Wang and Sun (2022). For services, we aggregate using gridded population data from WorldPop of the University of Southampton (Bondarenko et al. 2020b, 2020a). Limitations • Work intensities and the outdoor exposure of each occupation is matched to labor statistics based on the best available information. • All occupations are assumed to be performed evenly throughout the day and year. No seasonality of occupations is considered. • Wet bulb globe temperatures are approximated using monthly mean relative humidity and solar radiation, which is kept constant for future years. While there are more sophisticated methods to estimate wet bulb globe temperatures, such data are not available for this study. • Data regarding adoption of air conditioning is sparse. We utilize regional curves for household coverage as a proxy for workplace coverage. B1.3. Human health Climate change may impact the total labor supply through increased incidence of and death rate from various diseases, which results in time away from work due to absenteeism as well as from an increased number of deaths. Climate change could result in increased health effects of vector-borne diseases such as malaria and dengue, waterborne infectious diseases that cause acute diarrhea, and heat-related diseases (WHO 2014; Romanello et al. 2021). The approach utilizes different biophysical and statistical relationships between climate variables and the incidence of or transmissibility for each disease. 74  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Changes in incidence and death rates are calculated to model the number of hours of labor supply lost. On the one hand, excess deaths relative to the baseline will reduce the total labor. This effect is calculated independently for every year and does not consider population fertility and death rate dynamics. On the other hand, absenteeism from work due to people falling sick will further reduce the total available labor. This effect is divided in two, following the methodology applied by John et al. (2021). First, there is a direct effect from working age population getting sick and not able to work, and second, an indirect effect from children getting sick and needing parental care for the duration of the disease. Total hours of labor lost for each disease are then calculated for the country for both historical periods as well as future projections, for each disease and effect. Lastly, a percent shock to available labor supply relative to the baseline conditions is then calculated for the country total. • Labor hours lost due to death (hours/year): • Labor hours lost due to absenteeism for the disease (hours/year): = deaths = f(climate) C = cases = prevalence x total population = f(climate) = 15-64 fraction on total population (%) = 0-15 fraction on total population (%) = labour participation 15-64 (%) = average recovery time = week/year (f(disease)) x h/week = average hours worked per year per person Methodology: vector-borne diseases Vector-borne diseases are illnesses that are transmitted to humans through the bites of infected arthropods, typically mosquitoes. Malaria and dengue are two major vector-borne diseases that cause public health concerns around the globe. The modeling approach is based on the modeling of the conditions for stable transmission of the disease (i.e., suitable areas), based on the methodology applied by Ebi et al. (2005). The analysis is done for both malaria and dengue, typical mosquito-borne diseases whose spread depends on the right environmental conditions occurring for the mosquitoes to live, breed and increase in number, at a 0.5-degree resolution. These conditions are approximated from three climate variables: mean monthly temperatures, cumulative annual precipitation, and minimum annual winter temperature. Those variables are normalized using fuzzy functions following Craig, Snow and le Sueur (1999), determining a suitability index ranging 0 to 1. Fuzzy functions based on the equation below are applied to each climate variable. The S and U factors are the upper and bottom thresholds of the climate variables respectively, x is the observed climate variable, y is the resulting fuzzy variable (the fuzzy variable being y for the decreasing curve, (1-y) for the increasing curve): 1- When the fuzzy variable is 0 (i.e., nonsuitable), transmission is very unstable, with the disease either absent or with rare epidemics; when the fuzzy variable is 1 (i.e., suitable), disease transmission is most likely Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  75 stable; values between zero and one (0.1–0.9) represent a gradient from unstable to increasingly stable transmission. We select the minimum value between temperature and precipitation’s fuzzy variables. Then, the mean value of each year is compared to the minimum winter temperature fuzzy variable. For malaria, the thresholds are U=0, S=80 for precipitation; U=18, S=22 for the mean monthly temperature increasing part of the curve, S=32, U=40 for the decreasing part of the curve; and U=4, S=6 for annual winter temperature (i.e., minimum value of coldest month) (Ebi et al. 2005). Figure B2 illustrates the shape of the fuzzy functions for malaria. The same general equations are applied for dengue transmissibility. The corresponding threshold values are U=450, S=800 for precipitation; U=15, S=20 for the mean temperature (increasing), S=25, U=30 (decreasing); and U=-1, S=3 for winter temperature (Caminade et al. 2012). Figure B3 illustrates the shape of the fuzzy functions for dengue. Figure B2: Fuzzy functions for malaria a) Precipitation b) Mean temperature c) Winter temperature 1.00 1.00 1.00 0.75 0.75 0.75 Fuzzy variable Fuzzy variable Fuzzy variable 0.50 0.50 0.50 0.25 0.25 0.25 0.00 0.00 0.00 0 20 40 60 80 100 120 140 15 20 25 30 35 40 45 2 4 6 8 Monthly P (mm) Mean T (°C) Min T (°C) Figure B3: Fuzzy functions for dengue a) Precipitation b) Mean temperature c) Winter temperature 1.00 1.00 1.00 0.75 0.75 0.75 Fuzzy variable Fuzzy variable Fuzzy variable 0.50 0.50 0.50 0.25 0.25 0.25 0.00 0.00 0.00 0 200 400 600 800 1000 10 15 20 25 30 35 40 -5.0 -2.5 0.0 2.5 5.0 Annual P (mm) Mean T (°C) Min T (°C) Annual fuzzy variables are then used to calculate the exposed population by grid cell, as a probability of the occurrence of the disease in the area (LeSueur et al. 1998). Consequent deaths and cases due to the disease are then normalized and calibrated using reported deaths and prevalence rates from official sources. While malaria generally exhibits a stable transmission rate over time (with increasing or decreasing rates), dengue is more episodic and prone to random outbreaks that are difficult to attribute to clear factors (Chen et al. 2015). For this analysis, we do not model outbreaks or other epidemiological dynamics, and calculate changes in death and incidence rates for both diseases relative to the average reported rates of the latest five years, or the best available data. 76  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Methodology: heat-related illness Heat is a risk factor that can lead to a group of conditions that occur when the temperature of the human body is exposed to high temperature and humidity. Heat-related illnesses can include severe conditions such as exhaustion or stroke (CDC 2020). Heat effects on mortality and morbidity are modeled based on the 2021 Report of the Lancet Countdown on Health and Climate Change (Romanello et al. 2021). The methodology is based on calculating excess mortality from daily maximum temperatures, following the study by Honda et al. (2014). The temperature–mortality relationship is assumed to be V-shaped (figure B4) and the temperature value at which mortality is lowest is defined as the optimum temperature (OT). For temperatures above the optimum threshold for a given location, excess heat-mortality burden is defined daily as a fraction of the average total noninjury-related deaths occurring that day. Figure B4: V-shaped function of excess mortality due to high temperatures Excess mortality Mortality OT Daily maximum temperature Source: WHO 2014. An equivalent approach to vector diseases is applied to compute heat-related morbidity and is quantified by the following equation: =0×× E = heat-related excess mortality/morbidity in one day y0 = noninjury mortality rate on that day (yearly rate divided by 365) AF = attributable fraction on that day, calculated as =(−1)/=1−−(−) t = daily maximum temperature β = exposure-response factor (Honda et al. 2014) OT = optimum temperature (Honda et al. 2014) Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  77 Excess mortality is calculated for each grid cell and calibrated to nationally reported statistics of noninjury heat-related mortality. Methodology: waterborne diseases Waterborne diseases are illnesses caused by contact with water that is contaminated with infectious microorganisms that cause diarrhea, vomiting, and fever. Waterborne pathogens can spread through drinking water, irrigation waters, or recreational water bodies. Climate change is likely to increase the spread of waterborne pathogens such as cholera, typhoid, dysentery, or leptospirosis, resulting in increased mortality and morbidity from diarrheal diseases (Levy, Smith and Carlton 2018; Nichols, Lake and Heaviside 2018). Our analysis is based on the modeling approach applied in WHO (2014). While most studies highlight a positive association between the incidence of diarrhea and temperature, results of the association between mean monthly rainfall and E. Coli diarrheal cases are more varied, and the effect is much smaller (Philipsborn et al. 2016). For this reason, the WHO model applies gridded estimates only of average annual temperature anomalies to a statistical temperature–mortality risk relationship. The approach considers combining total estimated diarrheal deaths and cases in the future without climate change and estimating the climate change-attributable percent change. The following general equation is applied at a yearly time-step at a 0.5-degree grid cell resolution: n = number of climate change-attributable average annual diarrhea deaths or cases N = total number of average annual diarrheal deaths or cases in a future without climate change, obtained as the product between population and a baseline rate from available official statistics ΔT = yearly temperature anomaly β = log-linear increase in diarrheal deaths per degree of temperature increase with β = log (1 + α), where α is the linear increase in diarrheal deaths per degree of temperature increase. We assume α as a midestimate from WHO (2014). The annual number of diarrheal cases and deaths is then calculated as n+N for each grid cell and then aggregated at the country level. Figure B5 illustrates the percent change in the incidence of cases and deaths from temperature changes. Limitations • We considered those diseases that are more widely cited and modeled in the literature. However, this is not an exhaustive list of all the health effects that can be caused by climate change. Other direct and indirect causes of illness that can be linked to climate variables, such as schistosomiasis, outdoor air pollution, occupational hazards, or malnutrition are not considered in this analysis. • We utilize mean baseline mortality and incidence rates to measure the mean effects of climate change. Outbreaks and more complex epidemiological dynamics are not modeled. • Population dynamics over time that account for factors such as births, deaths, or aging of the population and result in later effects in labor supply are not considered in this study. 78  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Figure B5: Change in waterborne disease incidence from changes in temperature 40% 35% 30% 25% 20% 15% % change 10% 5% 0% −5% −10% −15% −20% −25% -2 -1 0 1 2 3 4 5 T change (°C) Water, sanitation, and hygiene The human health effects of climate change documented earlier in this section focus on assessing potential impacts on labor supply due to increase mortality, including the spread of waterborne diarrheal diseases. Yet, development policy initiatives in WASH can indirectly influence the severity of potential climate change impacts on human capital, as the quality of infrastructure can help to reduce diarrhea cases and related mortality (World Bank 2018b). This impact channel evaluates the benefits of enhanced investments in WASH, by comparing a baseline scenario where current trends of coverage and quality of infrastructure continue over time, relative to a scenario where additional investments in WASH reduce the incidence of waterborne diseases. These investments are presented as part of the policy scenarios discussed in chapter 3 of the main report. The approach follows the methodology applied by Wolf et al. (2019), which is based on a statistical relationship between a fecal contamination composite index (FAECI) and the relative risk of diarrheal diseases. Methodology A population-attributable fraction of the total diarrheal deaths and cases is computed to assess the part of those deaths and cases caused by inadequate WASH. This is based on available official statistics on WASH coverage. The relative risk associated with such WASH access is computed and the change in relative risk due to different policy scenarios is reflected through changes in the population-attributable fraction. The relative risk assessment is estimated through the FAECI (Wolf et al. 2019), which utilizes a rubric to assign a 0, 1, or 2 value to eight indicators related to water, sanitation and hygiene access (figure B6). The FAECI corresponds to the sum of these indicators, ranging from 0 to 16. Figure B7 presents the relationship between the FAECI and corresponding diarrhea relative risk. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  79 Figure B6: Fecal Contamination composite index indicators WASH categories FAECI indicators Construct Basic drinking water services (W1) Safely managed drinking water Water services (W2) Open defecation/unsafe child faeces disposal (S1) Basic sanitation services (S2) Sanitation Fecel community Safely managed sanitation contamination services (S3) Community coverage with basic sanitation services (S4) basic handwashing facilities (H1) Hygiene handwashing with soap after potential fecal contact (H2) Source: Wolf et al. 2019. Figure B7: Diarrheal disease relative risk 1.2 Humphrey Clasen Null 1 Briceno Patil Reese Khush Pickering Relative risk (diarrhoeal disease) 0.8 AZIZ Klasen MESSOU WALKER 0.6 LUBY Godfrey Pradhan 0.4 MORAES 0.2 0 16 15 14 13 12 11 10 9 8 7 6 5 4 3 FAECI Source: Wolf et al. 2019. Resulting relative risks are calibrated for the country using reported data on mortality and morbidity linked to inadequate WASH infrastructure. Then, a percentage change in diarrheal mortality and morbidity from improvements in WASH infrastructure relative to base conditions scenarios is computed for each scenario at a country scale. 80  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Limitations • We utilize mean baseline mortality and incidence rates to measure the mean effects of climate change. Outbreaks and more complex epidemiological dynamics are not modeled. • Population dynamics over time that account for factors such as births, deaths, or aging of the population and result in later effects in labor supply are not considered in this study. In particular, child stunting impacts from inadequate WASH and associated population effects (e.g., the later effect on labor supply of changing death rates in children) are not considered. B2. Natural capital Agriculture, land use, and forestry are all expected to experience a variety of impacts from climate change. Temperature increases are likely to reduce the suitability and productivity of crops, pastures, and livestock. Changes in precipitation patterns can result in reduced water resources available for agricultural users, impact erosion levels which in turn affect soil fertility and reservoir sedimentation, as well as influence reservoir water quality. We estimate these effects through the following channels: • Crop production, which models changes in rainfed crop productivity as a function of the availability of rainfall, as well as heat stress effects from increasing temperatures. Impacts to irrigated crop production from climate change are derived from prior studies. • Livestock production, which models changes in productivity by animal and product type as a result of extreme heat and changing feed availability effects through animal-specific impact curves. • Erosion impacts, which models changes in soil conditions and topsoil erosion from altered precipitation, which in turn results in changes in crop productivity. B2.1. Crop production Under climate change, crop yields may be affected by changes in rainfall patterns, increasing evaporative demands, and extreme heat as temperatures rise. The analysis is conducted for selected crops at a 0.5 x 0.5-degree spatial resolution. The approach relies on the Food and Agriculture Organization’s (FAO) Irrigation and Drainage Paper 66, Crop Yield Response to Water (Steduto et al. 2012), in which rainfed crop yields are estimated by applying crop- specific water sensitivity coefficients to the ratio of effective precipitation to potential crop evapotranspiration. This approach is the basis for the FAO’s monthly crop model CropWat, which is the predecessor to their daily biophysical crop model, AquaCrop.5 The water availability approach is then supplemented with impacts to crop yields from extreme heat during reproductive stages of development, when crops are more sensitive, causing a reduction in seed numbers (Prasad, Staggenborg and Ristic 2015; Roberts 1988). Heat stress impacts are modeled daily following AquaCrop’s approach (Salman et al. 2021), which considers a negative relationship between supra-optimal temperatures during the flowering stage of crop development. Methodology First, representative crops are selected for the country. In general, crops are selected so as to represent at least 80 percent of the total production revenues as well as harvested area in the country. Additional crops of national relevance may also be added to the list. For each country, crop calendars are obtained from the FAO (2022a) and supplemented with local sources when available. Crop calendars (i.e., the time of the year 5 https://www.fao.org/aquacrop/overview/whatisaquacrop/en/. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  81 when crops are sown, grown, and harvested) are allocated to 0.5-degree grid cells based on the production zones in the FAO’s calendar, which is based on administrative districts and/or agroclimatic regions. We gathered harvested area, production, yield, and revenue statistics from available local or global sources. We also gathered irrigation statistics from the best available sources, distinguishing rainfed from irrigated production. Potential evapotranspiration (ET0) is used as a reference to estimate crop water requirements by adjusting it based on the crop and growth stage. For each crop, water demand is calculated by multiplying monthly potential evapotranspiration (ET0) by monthly crop water demand coefficients (Kc), which produce annual crop evapotranspiration requirements (ETc). Monthly potential evapotranspiration was calculated using the modified Hargreaves method (Droogers and Allen 2001), which requires data on extraterrestrial radiation, monthly precipitation, and minimum and maximum temperatures. Crop water demand coefficients for each month of the growing season were obtained from the FAO’s Irrigation and Drainage Papers 33 and 56 (Allen et al. 1998; Doorenbos and Kassam 1979). Rainfed crop water supply is effective precipitation (Pe), which is monthly precipitation adjusted for drainage qualities of the soil and then capped at ETc levels. Pe is calculated from monthly precipitation data following the methodology from the FAO’s Irrigation Water Management Training Manual no. 3 (Brouwer and Heibloem 1986). Next, we calculate the annual ratio of effective precipitation (Pe) and crop water need (ETc) by grid cell for each crop and then multiply these results by the corresponding annual yield response coefficient (Ky), as presented in the equation below, following the approach from the FAO’s Irrigation and Drainage Paper 66 (Steduto et al. 2012). Maximum crop evaporation assumes no water constraints, whereas actual evapotranspiration is reduced based on available rainfall, which allows for the calculation of actual yields (Ya), as a deficit from nonwater-constrained yields (Yx). For Ky values below one, crop yields fall below the water deficit, whereas for Ky values greater than one, yield losses are relatively greater than the water deficit. Figure B8 illustrates the relationship between mean monthly precipitation and resulting Figure B8: Relationship between precipitation and crop annual yield response following the approach yield response described above, for six different levels of 1.00 Mean mean monthly potential evapotranspiration monthly (PET). For a given potential evapotranspiration PET (mm) level, higher precipitation volumes result 0.75 250 in higher yield responses. When potential Yield response evapotranspiration is low, less precipitation is 200 required to reach the maximum yield response 0.50 (i.e., 1) 150 Next, a yield impact from temperature (Ks) 0.25 100 is calculated based on daily maximum temperatures, determined by crop-specific 50 optimum temperatures and tolerance 0.00 thresholds. The modeling starts by 0 50 100 150 200 250 300 gathering optimum and maximum tolerance Mean monthly precip (mm) temperatures by crop from the FAO’s Crop Ecological Requirements (ECOCROP) database (FAO 2015), which will determine at which temperature a crop will start experiencing damage until it suffers full loss. Crops are typically more vulnerable to heat during reproductive stages than vegetative stages. We 82  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania consider the months during the flowering stage of crop development to determine the maximum temperature a crop is exposed to. These months are identified following the methods outlined in the FAO’s Irrigation and Drainage Papers 33 and 56 (Allen et al. 1998; Doorenbos and Kassam 1979). Temperature yield responses (Ks) are estimated based on a logistic relationship between temperature and maximum attainable yields, as indicated in the equation below and illustrated in figure B9 for an illustrative threshold of 35°. Topt is the optimum temperature above which crop yields start decreasing; T is the maximum daily temperature, and B and v are factors that are calibrated for each crop based on its tolerance thresholds. Results range from 1 to 0, where 1 represents no stress and zero represents total crop failure. Temperature effects are typically experienced after consecutive days of exposure. For each day during Figure B9: Illustrative relationship between the flowering period, we consider the lowest effect temperature and yield response between the daily effect on t, t-1, and t-2 from the 1.00 equation above. Major food crops, such as wheat, sorghum, maize, and oil crops start experiencing the effects of heat after three consecutive days of 0.75 exposure (Wahid et al. 2007; Nuttall et al. 2018; Yield response Hatfield and Prueger 2015; Gourdji et al. 2013). We utilize the average 3-day maximum temperature of 0.50 2-week periods across the entire flowering period to estimate the annual temperature effect on yields for all crops. 0.25 Temperature impacts the potential yield of crops during the flowering stage, while precipitation 0.00 impacts the resulting production, therefore producing 25 30 35 40 45 50 multiplicative effects. Hence, water availability and T (ºC) temperature shocks are then combined into a single shock by crop. Grid cell level shocks are aggregated nationally based on the spatial distribution of crop production from available sources. Crop-specific shocks are also aggregated into a total production shock using crop revenues as weights. The impacts of climate change on irrigated crops are not modeled; rather, they are evaluated based on a review of the existing literature. Limitations • We consider a subset of all crops grown in the country, which necessarily excludes some crops. • Additional effects from agricultural practices, soil characteristics and conditions, fertilizer use, and other climate variables (e.g., wind speed or radiation) are not considered in the model. • Both water and temperature coefficients are taken from existing literature for representative varieties of each crop. Whether local varieties have different levels of tolerance is not considered in this study. • We consider heat exposure for crops during the hottest 3-day period during the flowering stage, which is the time at which crops are most sensitive to heat. However, shorter episodes of extreme heat as well as longer periods of consistently high temperatures can result in additional yield losses Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  83 or potential crop failure. In addition, crop biomass could suffer from heat exposure in other stages of development, however, these effects are not considered. • The start date and length of the crop’s growing seasons are assumed to be static over time. However, farmers may adjust growing patterns based on short-term weather forecasts. • An assessment of the implications of deteriorating water quality and increasingly saline soils on water demands in future years is not included. The decrease in water quality could likely either further reduce water reuse practices or cause productivity impacts. • Reservoir volumes are assumed to remain constant at reported levels, with limited effects from sedimentation considered. This assumption may overestimate water storage availability over the next 40 years. B2.2. Livestock production Climate change poses risks to livestock production from increasing heat stress on animals (direct effect), which in turn causes reductions in productivity (St-Pierre et al. 2003; Salama et al. 2014; Bohmanova, Misztal and Cole 2007). Heat stress refers to the exposure of animals to extreme heat or hot environments that lead to the body’s inability to regulate internal temperature. Climate change can also cause potential reductions in the availability of feed sources (indirect effect) that result in lower energy intakes, hence reducing yields (Seo et al. 2021; Mahgoub, Lu and Early 2000; Noblet et al. 1994). The analysis focuses on the main species globally (cattle, chicken, swine, sheep, and goats) for the three most important products (milk, meat, and eggs) at a 0.5 x 0.5-degree spatial resolution. The direct effects are estimated using a combination of animal- and product-specific equations that relate a daily temperature-humidity index (THI) as an indicator of heat stress, with productivity losses based on tolerance thresholds. The THI calculation varies by species and requires data on air temperature, wet-bulb temperature, and relative humidity. Estimation of indirect effects relies on the best available information on livestock feed sources (i.e., pastures, fodder crops, crop residues) by species, which is typically sparse. For pastures, we model the productivity of grasses following the same approach, using appropriate rainfall and temperature coefficients. We do not model changes in the availability of imported feed nor in other sources such as concentrates or other byproducts. Reductions in feed availability are converted into changes in dry matter intake, which causes metabolic reductions that are utilized as direct meat production losses. Grid cell level shocks are aggregated nationally for each product based on the best available data on livestock headcounts by region and revenues by product. Methodology The relationship between livestock productivity and temperatures is modeled using the THI, which relates air temperatures in degrees Celsius and relative humidity (as a percentage). The THI measures the perceived temperature by the animal as widely used in livestock research. This is used as an indicator for assessing the level of heat stress on animals caused by weather conditions. The THI calculation varies by species and requires data on air temperature, wet-bulb temperature, and relative humidity. Effects are calculated daily, and range from 1 up to the optimum THI, from where values start decreasing up to the maximum THI value, with no production assumed on days that experience the maximum THI. Our quantification of milk production losses from cattle follows Mauger et al. (2015), who estimate milk production losses in the United States from Holstein dairy cows following the THI equation (equation (1) below) based on maximum daily temperature (Tmax) and relative humidity (RH). Milk production loss begins 84  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania when the daily maximum THI exceeds an identified threshold as cows move beyond their ideal thermoneutral zone. Figure B10 illustrates the relationship between temperature and THI for cattle for different levels of relative humidity (RH) and an illustrative threshold (dashed line). Milk losses (LOSS) are calculated at a daily time step in kilograms per cow (following equation (2)), with D as the fraction of the day that the maximum THI is above the threshold (0≤ D ≤ 1). (1) THI = (1.8 Tmax + 32) – (0.55 – 0.55 * RH)(1.8 * Tmax – 26) (2) LOSS = α (THI max – THI threshold )2 x D Milk yield from goats is quantified using methods outlined in Salama et al. (2014). Here yield is modeled as a function of daily THI. We calculated the THI following equation (3). Figure B10 illustrates THI for goats, which are more resistant to heat stress than cattle but, for the same temperature and relative humidity, experience a higher THI. The biological response function in equation (4) is used to calculate milk yield per head in L/day. (3) THI = (Tdb - (0.31 - 0.0031 * RH) * (Tdb – 14.4) (4) Yield = (-0.0336 * THI) + 5.3539 Meat production losses from cattle, pigs, and chickens are quantified following the approach by St-Pierre et al. (2003), where production losses due to heat stress result from higher death rates. As mentioned by St-Pierre et al., while there is some evidence that higher THI can result in reduced dry matter intake by animals, this relationship is not fully understood and is considered negligible. Therefore, this analysis considers production losses from a change in the monthly death rate due to heat stress, following equation (5). THIload is calculated as the area below the maximum THI and above the THI threshold. Monthly THI for meat-producing cattle is calculated using equation (1) above. THI values for pigs and chickens are calculated as a function of wet bulb (Twb) and dry bulb (Tdb) temperatures following equations (6) and (7), respectively. An illustration of THI for chickens is presented in figure B10. (5) PDeath = 0.0004275 * EXP(0.00981 * THIloadm) (6) THI = (0.25 * Twb) + (0.75 * Tdb) (7) THI = (0.6 * Tdb) + (0.4 * Twb) The monthly change in the death rate is summed across years by global climate model and applied to a baseline headcount of meat-producing cows, pigs, and chickens, obtained from the FAO for the year 2017–21 for the country. Each month, we assume that the livestock population regenerates to its original headcount. We then quantify meat losses by using an estimate of meat production per livestock head value to estimate total meat production losses resulting from heat-related deaths. Lastly, for heat stress effects, egg production losses are estimated following an impact equation developed by St-Pierre et al. (2003) (see equation (8)). LOSS is calculated as a function of the daily THIload, to quantify losses in egg production in kilograms per hen per day. This value is multiplied by the average headcount of laying hens in the country from 2017 – 2021, obtained from the FAO. (8) LOSS = 0.048 – ((0.8 – ((0.00034 X THIload)) * (0.06 – (0.0000123 x THIload))) We calibrate the THI thresholds for the equations above using data from Rahimi et al. (2021) and St-Pierre et al. (2003), depending on the region of the country. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  85 Figure B10: Relationship between temperature and THI Cattle Goats Chickens 40 40 35 35 40 30 30 THI THI THI 30 25 25 20 20 20 15 15 15 20 25 30 35 40 15 20 25 30 35 40 15 20 25 30 35 40 T (°C) T (°C) T (°C) RH 0.00 0.25 0.50 0.75 1.00 To estimate the effects of changes in the availability of feed, the first step in the methodology is gathering (or estimating) the volume of feed that is consumed by each species, for each type of feed. Feed sources include (1) grazing pastures, which is typically the main source for ruminants (i.e., cattle, goats, and sheep); (2) fodder crops grown for livestock such as cereals, grains, seeds (typically for chicken), or fruits and vegetables (typically for pigs); (3) crop residues from the production of major cereals (e.g., maize, wheat, rice) and legumes; (4) by-products of crop processing such as cereal gluten, sugarcane molasses, or oilseed cakes; and other industrial products like concentrates. From these, we exclude the portion of production that is imported i.e., we only consider domestic production of feed. When available, we consider differences in average feed intake by herds, distinguishing between dairy and meat ruminants and broiler/ layer chickens. Furthermore, we also incorporate variations by extensive and intensive producing systems. Extensive systems allow for animals to move freely and graze in open pastures or ranges (for ruminants) or backyard spaces (for chickens and pigs), while intensive systems keep animals in confined spaces with controlled diets and additional inputs to maximize productivity. For pastures and all feed sources derived from local crop production, we follow the approach from the FAO’s Irrigation and Drainage Paper 66, Crop Yield Response to Water (Steduto et al. 2012), in which rainfed yields are estimated by applying grass-or crop-specific water sensitivity coefficients to the ratio of effective precipitation to potential crop evapotranspiration. Then, a heat stress impact is modeled daily following AquaCrop’s approach (Salman et al. 2021), which considers a negative response relationship between supra- optimal temperatures during the flowering stage (see detailed description of this approach in section B.2.1). The analysis assumes that pastures are currently being grazed at their maximum carrying capacity or above and, as a result, any reduction in productivity from a changing climate will translate into a direct productivity loss. We consider the main types of grasses in the country and, if no other information is available, we assume that these are equally distributed across space. Live body weight gain in livestock depends on the total feed intake, which depends on animal-specific maintenance needs and varies by age and preproduction stages, and on the efficiency of the feeding system. Animal feed efficiency is typically measured using a Feed Conversion Efficiency (FCE) parameter, a measure of how efficiently animals convert feed into body weight gains or other useful outputs, such as milk or eggs, and is calculated as the ratio of feed intake to live weight gain or the weight of the useful output (Berry and Pryce 2014). Live body weight can increase from higher intakes of dry matter or from higher FCE, which 86  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania can be attributed to differences across breeds and feed sources for any type of livestock (Lima et al. 2017; Mackle et al. 1996; Khan et al. 2007; Wen et al. 2018). For this analysis, we assume constant FCE values and assess changes in live body weight, resulting in changes in total meat production, from changes in total feed intake. Baseline FCE values are obtained from the FAO’s Global Livestock Environmental Assessment Model (FAO 2022b). Reductions in feed sources (in kilograms) are then converted into shocks to dry matter intake for each animal, herd, and production system, and in turn, translated into meat production losses. Limitations • The relationship between changes in metabolizable energy intakes and production of milk or eggs is not fully understood, as these vary greatly based on the particular breed, age of the animal, farming system, level of input, and climatic region, and depend on how farmers chose to cope with the scarcity of feed (i.e., by reducing animal intake per head, shrinking the herd, purchasing alternative feed, etc.). We consider the effects of reduced feed intake on meat production, as there is more available literature applicable to the scope of this study. As such, the effects on milk or egg production from lower energy intake are not considered. • Changes in feed availability only consider climate changes effects on domestic production. Regional or international effects, which can be a result of effects beyond climate change, are not considered. • The equations utilized in this analysis have been developed under particular controlled conditions, for certain breeds and practices. We calibrate the curves to the best extent possible, but local differences in responses to heat and energy intake may alter the results. • The analysis does not consider potential reductions in water availability for livestock watering, which can further decrease the productivity of animals. • Only meat, milk, and egg production are considered in this analysis. Other products such as wool, hides, offal, honey, beeswax, game meat, and meat from animals such as camels, horses, turkeys, ducks, rabbits, or any other animal are not modeled. However, while those products may be relevant at a local scale, they tend to represent a small share of the total country livestock production and revenues. • The analysis does not consider the different growth stages of animals, nor differentiate effects on particular ages (i.e., calves or a population dynamics model that includes reproduction and death rates). • Pastures are assumed to currently be grazed at their maximum carrying capacity or above and, as a result, any changes in the productivity of grasses will result in a lack of feed. • No transhumance or movement of livestock across the country over the course of a year or season is considered in the analysis. Pastoralism, to the extent that is present in the country, assumes that animals do not migrate out of the modeled grid cells. B2.3. Soil erosion Soil erosion is a major concern in many countries. Erosion can be detrimental to landscapes, impacting plant and animal life, reducing the efficacy of reservoir storage and hydropower production through sedimentation, and causing declines in agricultural production by removing valuable nutrients from the topsoil, which may be made worse if climate change intensifies future rainfall intensity. In addition, Nambajimana et al. (2020) find a correlation between higher poverty levels and estimated erosion rates. To determine erosion rates, we use the Revised Universal Soil Loss Equation (RUSLE) developed by the United States Department of Agriculture (Wischmeier and Smith 1978) and revised by Renard et al. (1997). Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  87 This equation is the most widely used approach to estimate erosion and soil loss rates and has been used in Rwanda (Nambajimana et al. 2020), Uganda (Karamage et al. 2017), and many other nations worldwide (Panagos et al. 2017). For information on the uncertainties of the RUSLE approach, see Alewell et al. (2019) and Benavidez et al. (2018). Soil erosion estimates are converted into crop yield losses following an approach developed by the FAO (Kassam et al. 1991). Methodology The RUSLE calculation requires five key inputs, which are shown below. A is the soil loss and R is rainfall- runoff erosivity (i.e., the potential of rainfall to cause erosion by generating runoff). K and LS are static climate and land factors, while C and P are activity and farm-level management factors. The following outlines the data sources and approaches used to determine each of these parameters. Estimating the revised rainfall-runoff erosivity (R) requires highly temporally detailed (30 min) rainfall records for a variety of storm events. However, many methods have been developed to approximate R. Two datasets are used to determine the rainfall-runoff erosivity: we use a historical dataset of R factors from Panagos et al. (2017) and adjust for future climate scenarios using Lo et al. (1985). The soil erodibility factor (K) correlates with soil properties (i.e., fraction of sand, silt, clay, and organic carbon). The K-factor was estimated using the relationship between soil properties and K developed by Williams (1995). The slope and slope length (LS) factor is a product of slope length (L-factor) and slope steepness (S-factor). The L-factor was computed following the method proposed by Desmet and Govers (1996) while the algorithm recommended by McCool et al. (1989) was used to calculate the S-factor. The C-factor determines the impact of land cover and management practices on the magnitude of soil erosion. The equation proposed by Durigon et al. (2014) is used here to approximate the C-factor. This requires land cover data, which was sourced from the biweekly mean MODIS (or Moderate Resolution Imaging Spectroradiometer) normalized difference vegetation index (NDVI) provided by the National Aeronautics and Space Administration (NASA 2022). Lastly, support practices (P) reflect erosion-reducing practices employed by farmers and vary by conservation support practices such as contouring, strip-cropping, and terracing. Although all three of these practices can be implemented, a generic P-factor is used that was recommended by Wischmeier and Smith (1978), which varies with slope. Generally, areas that are impermeable (e.g., rocky surfaces or waterbodies) and areas with mean slopes that exceed 20 percent are excluded from the analysis because erosion on these surfaces tends to be low or highly uncertain with the RUSLE approach. As noted, soil loss can reduce the nutrients available to crops, if not replenished by fertilizers, by eroding the topsoil. Although topsoil is generated naturally, natural generation is slow, usually less than 1 mm/year, or roughly 12t/ha, depending on soil density (Hammer 1981; Hudson 1981). To approximate the impact this has on the major crops, we use a method developed by the FAO (Kassam et al. 1991). The approach is based on a tolerable loss rate over time and varies by levels of inputs (high, intermediate, and low) as well as the susceptibility of soils to productivity loss. We use global raster data of fertilizer use (nitrogen and phosphorus) from the Global Agricultural Inputs dataset to determine the level of input in a country (Potter et al. 2010). 88  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Limitations • While this approach is appropriate for a national-level assessment, the methods rely on satellite- derived datasets. For farm-level analysis, more detailed data on farm practices, soils, rainfall erodibility, slope, and other locally derived information will provide a more accurate estimate of farm-level erosion. • Although outputs from state-of-the-art global climate models are used in this analysis, these models are not able to accurately estimate changes in extreme events such as heavy rainfall that can cause significant soil loss in a short period of time. The RUSLE approximates soil loss on short time scales (half-hour), but changes in precipitation on these time scales are beyond the scope of global climate models. • Erosion causes the depth of the soil layer to deplete over time. The depth to bedrock or other semi- impervious ground content (such as heavy clay) may significantly reduce the rooting depth that crops can achieve during the growing season. Shallower rooting depths reduce the soil water available to crops, which then reduces crop yields and eventually renders crop fields unusable. This effect was not evaluated in this study. • Production decreases in response to topsoil loss from this approach are not crop-specific. B3. Physical capital Climate change is likely to impact physical capital, and the services provided by it in a variety of ways including by increasing the frequency and magnitude of extreme events that result in damage to assets, as well as by increasing deterioration caused by heat and precipitation levels. We model these effects through the following channels: • Roads and bridges, which models increases in the repair and maintenance to road infrastructure due to higher temperatures, precipitation, and flooding recurrence, as well as road disruption effects on labor. • Inland flooding, which models damage to capital across the country from changes in the magnitude and frequency of riverine (fluvial) flooding events. B3.1. Infrastructure and capital stock model Various biophysical models will be used to estimate the hazard and damage on each individual infrastructure channel. The exposure and vulnerability of assets are determined by three key variables: the geospatial location, the type or sector of the asset, and the total value of the asset. We developed a common asset location and value layer that served as the basis for calculating damage and impacts. Asset values are determined at a national scale based on data or estimates of total capital value from gross domestic product. Total capital stock value is estimated by dividing gross domestic product data by the capital-output ratio of the country or region (i.e., the ratio of the total economic output generated for each unit of invested capital). Capital-output ratios are used as an indicator of the efficiency of an economy, with lower values indicating high productivity and lower capital requirements to achieve additional growth. If local information is not available, national capital stocks and output ratios are obtained from the International Monetary Fund’s Investment and Capital Stock Dataset (IMF 2021). Then, sectoral as well as spatial breakdowns are modeled using the best available information from local sources, which typically is available by province, district, or similar administrative boundaries. In cases where these data are unavailable, we estimate the spatial distribution using proxy high-resolution data such as land use land cover data from the Copernicus Fractional Land Cover dataset (Buchhorn et al. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  89 2020), gridded population data from WorldPop of the University of Southampton (Bondarenko et al. 2020b, 2020a), gridded gross domestic product data from Wang and Sun (2022), value of crop production from the FAO’s Global Agro-Ecological Zones project (GAEZv4) (Fischer et al. 2021), and shapefiles of specific infrastructure types (e.g., residential, industry, transport) from the Humanitarian OpenStreetMap Team geospatial data (HOTOSM 2020). Inland flooding Flooding events disrupt daily life and damage infrastructure and physical capital. Climate change may exacerbate flooding by increasing the frequency, intensity and duration of storm events. This analysis relies on projected changes in the return interval of peak precipitation events from the World Bank’s Climate Knowledge Portal. Flood hazard maps are developed to determine areas with a certain probability of flooding for a given baseline and climate change projected return period. The outputs of flood hazard mapping include the extent and depth of flood inundation, which are then used to estimate damage to infrastructure. The analysis is done for the available eras, recurrence intervals, climate scenarios in the Climate Change Knowledge Portal for changes in the annual exceedance probability of the largest single-day precipitation relative to history, at a spatial resolution determined by the available hydrology and asset distribution data. An era refers to a period of time with distinct characteristics or patterns, for example, a decade or a 30-year period. The eras considered are 2010–39 (centered in 2025), 2035–64 (center 2050), and 2060–89 (center 2075). The modeling considers a combination of the probability of occurrence of 5, 10, 20, 25, 50 and 100-year return period events. The resulting outputs are aggregated to a national scale and correspond to the expected share of assets damaged relative to a historic baseline (1995–2020). Methodology The first step in the process is to model runoff from historical daily precipitation data for a range of return periods using the TR-20 approach, which relies on curve numbers to estimate the amount of runoff. Curve numbers are an empirical parameter developed by the United States Department of Agriculture Soil Conservation Service that represent the ability of the surface to absorb rainfall before rainfall occurs, depending on the land use, soil type, and moisture conditions. For this process, curve numbers are determined for each catchment, then, estimating runoff as the excess between rainfall and soil infiltration. Figure B11 illustrates the relationship between precipitation (P) and runoff (Q) for different curve numbers. Next, peak flows in cubic meters per second (m3/s) are estimated for each catchment and used to generate a hydrograph. Projected future runoff is calculated by applying the change in the annual exceedance probability for each scenario from the Climate Change Knowledge Portal to historical precipitation, and recalculating runoff. Figure B11: Curve number model: relationship between precipitation and runoff 20 0 95 10 = 90 N 18 (P-0.2S)2 C Q= P+0.8S 85 16 80 14 75 70 12 65 Q (cm) 10 60 55 8 50 6 45 4 40 2 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 P (cm) 90  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania For the next step in this process, we utilize the HydroRivers geospatial dataset (Lehner and Grill 2013), which contains a vectorized river network for the globe, to estimate floodplain boundaries as buffers around river centerlines. We model river flows from the modeled runoff data using the Muskingum-Cunge method of flood routing (Ponce 2014), which considers routing parameters based on hydraulics to simulate accumulating flows as water moves from upstream to downstream over time. Brunner and Gorbrecht (1991) present an application of this method. Figure B12 presents a schematic of this process. The HydroBasin data is used to develop a stream network from the pour point at the ocean all the way upstream, which could extend beyond the country borders. Peak discharge time increases and peak flows decrease as flow is routed downstream (see panel b for nodes 45, 48, and 49). Figure B12: Routing of hydrographs schematic a. River node calculation order b. Inflow and outflow for river nodes 45, 48 and 49 54 35 36 3500 -32.8 58 55 37 27 26 56 28 3000 53 39 33 -33.0 57 38 34 29 52 31 25 3 1 30 2500 2 51 50 32 -33.2 49 24 5 4 45 48 47 22 21 8 7 2000 Latitude 40 41 23 MAT 44 46 20 19 -33.4 43 6 1500 42 18 17 9 1000 -33.6 10 16 15 500 -33.8 14 12 13 0 11 0 20 40 60 80 100 18.2 18.4 18.6 18.8 19.0 19.2 45 inflow 45 outflow 48 inflow Longitude 48 outflow 49 inflow 49 outflow Next, we identify those floodplain areas within which assets are subject to damage. We calculate a bankfull river width (i.e., the surface width at a bankfull river stage) for each river link using the equation developed by Allen, Arnold ad Byars (1994). For calculating bankfull widths, we consider the streamflows that are exceeded by 10 percent of the observed records (i.e., q10 flows). Then, we calibrate floodplain extent based on floodplain-to-width of stream ratios developed by Bhowmik (1984), which consider stream order. The stream order value indicates the river ordering from sink to source, with downstream rivers having lower floodplain-to-width ratios. For each climate scenario, flood widths are next converted into flood depths within the identified floodplain. Depths are calculated for each stream reach using a triangular arithmetic approach based on widths obtained from the routing model, a rectangular cross-section, and Manning’s equation which relates flow rate, velocity, and depth of water in a river as shown below: Where Q represents the flow rate (m3/s), n the Manning’s roughness coefficient (dimensionless), A a cross- sectional area of flow (m2), R the hydraulic radius (m), and S the channel slope (m/m). Flood hazard maps are finally produced for the various scenarios and used as the basis for quantifying capital losses. Depths for each section of the floodplain are then combined with depth-damage functions to estimate the total share of infrastructure that is damaged in a particular flood event. Damage functions describe the relationship between the level of asset damage and the flooding depth. The maximum flood damage to an object can be computed as a certain damage factor multiplied by the total dollar value of the object. Global flood depth-damage functions are available by region and state defined for water depths Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  91 between 0 and 6 meters (Huizinga, de Moel and Szewczyk 2017). Figure B13 shows an illustration of available depth-damage functions from Huizinga, de Moel and Szewczyk (2017) for agricultural, commercial, and residential capital, for three different regions of the world. We calibrate these depth-damage functions by assuming that, for each catchment, infrastructure is built to withstand the historical 10-year event, hence experiencing no damage at the corresponding flood depths. Figure B13: Depth-damage functions a) Agriculture b) Commercial c) Residential 1.00 0.75 Damage (%) 0.50 0.25 0.00 0 1 2 3 4 5 6 0 1 2 3 4 5 6 0 1 2 3 4 5 6 Flood depth (m) Africa Asia South/Central America Lastly, we estimate the resulting damage from rainfall events of the return periods available in the climate data (i.e., 5, 10, 20, 25, 50, and 100-year events) for every catchment by overlaying capital value data by the percent damage from the damage function according to the estimated flood depth. The overlap between the floodplain and capital represents the total exposed assets, while the multiplication of its values and the percent loss from the damage function returns the absolute loss. Total expected damage (i.e., the damage times the probability) is also quantified by summing the total area under the damage-exceedance probability curves presented above and multiplying by the total exposed assets within each basin. Final results are aggregated nationally, weighted by the share of assets within each basin, representing the total damage to the country. Limitations • The analysis considers the flooding impacts from single-day extreme rain events within the region. Flooding may be caused by longer periods of continuous rain. These effects are not considered in this study. • Detailed modeling and high-resolution terrain data would be required to estimate more accurate depths at a given location. • We are not able to calibrate and verify the model output fully, in some cases, due to a lack of available records collected during major flood events. Project-scale analysis and modeling should be used to evaluate and verify these results when available. • More complex numerical hydrologic models are available but require extensive input and field measurements to ensure accuracy. • Asset values are proxied using the best available information on localized gross domestic product and capital-output ratios, and damage is quantified using generic depth-damage curves. Localized damage to a particular building may differ from these estimates depending on actual infrastructure 92  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania conditions, flood depth and infiltration, or location of critical inventory such as mechanical and electrical equipment. • Capital value is estimated using high-level (e.g., for a whole country or sector) capital-output ratios, which assume the same levels of productivity and conditions of capital. Local values may differ from this estimate due to differences in the type of the physicality of capital (i.e., different types of infrastructure), age, depreciation levels, sectors, and overall conditions and context. Bridges Climate change may impact bridge infrastructure in a variety of ways. This includes damage to and increased maintenance of bridges from changes in the recurrence of peak precipitation events. In turn, this is also associated with delays for passengers and labor supply impacts. Methodology For the bridges channel, we first gather local data on bridge location and characteristics. For this analysis we construct an artificial bridges-over-waterways dataset by intersecting a shapefile of roads from the Global Roads Inventory Project (GRIP) dataset with a shapefile of rivers from the HydroRivers geospatial dataset including only rivers with a Strahler Stream order of two or higher. We then consider unique intersections of roads and rivers to identify the locations of bridges. In addition to the location of roads, the GRIP dataset also contains information on road composition, with these characteristics used to make assumptions about bridge design standards. These assumptions are presented below: • If the road from the GRIP dataset is a highway, trunk, or primary road, we assume that the resulting bridge is designed to withstand the current 50-year flood. • If the road from the GRIP dataset is paved but not a highway or primary road, we assume that the resulting bridge is designed to withstand the current 25-year flood. • If the road from the GRIP dataset is unpaved or gravel, we assume that the resulting bridge is designed to withstand the current 20-year flood. Next, we determine damage to bridges from riverine flooding events. Flood hazard is determined by considering the extent and depth of inundation, as well as the magnitude of flood flows. These inputs are overlayed on the bridge infrastructure dataset developed as described above, which together are then used to estimate flow velocity, which ultimately drives the extent of damage to bridge infrastructure. Bridge repair costs vary greatly depending on the characteristics of the bridge, the type and magnitude of the damage, and the bridge component that is damaged (e.g., surface, superstructure, bearing). Generally, unit costs for major maintenance and reconstruction works after severe damage range from a low end of US$350–550 per square meter to a high end of $1,100–1,500 per square meter (JICA 2017; World Bank 2018a). For this analysis, we consider a central value of $1,000 per square meter. Projected flood depths for each section of the floodplain are then combined with depth-damage functions to estimate the total share of bridge infrastructure that is damaged in a particular flood event. In this analysis we assume a coefficient of 0 indicates no damage to a bridge while a coefficient of 0.75 indicates that the bridge is completely washed away from the flooding event. Based on this proportion, we estimate the cost of the damage using a cost per square meter from 2023 bridge repair costs in South Africa (Wade Rain Irrigation Systems, n.d.). We then multiply the damaged proportion by the area of the bridge and the unit repair costs to estimate the cost of damage to a particular bridge from a certain flood event. In addition, we also quantify bridge delays resulting from riverine flooding events. This is accomplished by assuming a relationship between the area of the bridge that is damaged and resulting delay hours. This relationship is presented below: Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  93 • If less than one-fifth of the bridge is damaged, we estimate 1 hour of delay per day for 3 months. • If less than half of the bridge is damaged, we estimate 2 hours of delay per day for 3 months. • If less than three-quarters of the bridge is damaged, 3 hours of delay per day for 6 months. • If the bridge is entirely washed away, we estimate 4 hours of delay for 6 months. This assumes that a temporary prefabricated replacement bridge or new route is opened 6 months after the bridge collapse. We then use information on traffic patterns from the Japan International Cooperation Agency to estimate traffic along the damaged bridge (JICA, n.d.). Lastly, we multiply average daily traffic (assuming two people per vehicle) by the delay impact outlined above, and the labor force participation rate to estimate labor supply impacts. Limitations • The analysis relies on an estimated number of bridges over waterways as in most cases a bridge inventory is unavailable. • Neither detailed topographic characteristics nor specific bridge designs that can lead to increased or reduced resilience are considered. • The analysis assumes that infrastructure is repaired as needed, resulting in increased expenses. This simplification does not reflect situations in which infrastructure is left unattended and potentially unusable, which may result in other effects such as further delays, loss of business, or road accidents and injuries. • Delay costs are approximated using general redundancy and traffic level standards by type of bridge. No behavioral changes in transport modes, travel times, or route selection are considered, as such analysis will require much more granular data and observations as well as detailed modeling beyond the scope of this analysis. Roads Climate change may impact road infrastructure due to increased temperatures, precipitation, and flooding that cause the infrastructure to deteriorate faster, which influences infrastructure repair and maintenance costs and causes delays for passengers.6 The analysis relies on the Infrastructure Planning Support System,7 which has previously been applied to the Enhancing the Climate Resilience of Africa’s Infrastructure study (Cervigni et al. 2017). The Infrastructure Planning Support System is a decision support system that performs engineering analysis within a broader resiliency perspective. It models infrastructure vulnerability to future climate and weather conditions, considers specific adaptation scenarios, and provides a cost- benefit based risk analysis. Methodology This impact channel covers impacts from daily temperatures, precipitation, and flooding stressors to paved, gravel, and unpaved roads. Generally, temperature impacts only paved roads, precipitation impacts both paved and unpaved roads, and flooding impacts all kinds of roads. For roads, the analysis first examines the impacts of climate change in the form of increased repair and maintenance costs incurred as a result of climate damage. The need for increased repair and maintenance can be due to accelerated aging of binder, rutting of asphalt, bleeding of seals due to temperature, reduced 6 Note the potential double-counting of flooding impacts with the inland flooding impact channel. 7 https://resilient-analytics.com/ipss. 94  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania carrying capacity due to increased moisture from precipitation, and washaways and overtopping from flooding. The analysis uses construction standards for each type of road infrastructure, as well as the typical lifespans and maintenance requirements under current conditions. Repair and maintenance costs for road infrastructure are based on unit cost assumptions and stressor response rules. The study then models the road network assuming investment in adaptation actions takes place. Adaptation investment is only considered when the road reaches the end of life and is evaluated on an annual basis using an equally distributed vintaging process. It is assumed that (1/lifespan) percent of the roads in a grid reach the end of life annually. Table B1 summarizes the effects of climate stressors on each type of road, the additional repair and maintenance costs incurred due to climate damage, as well as the different adaptation measures considered in the analysis. Table B1: Possible climate effects on roads and adaptation measures Climate stressor Effect Repair and maintenance costs incurred Possible adaptation measure due to climate damage Paved roads Temperature Increased temperature leads to Additional sealing required on a more Construct dense seals (e.g., Sand Seal, Otta accelerated aging of binder. frequent basis such as a five-year Seal, Cape Seal). Typically, Cape Seals are schedule instead of a seven-year used on heavily trafficked roads. schedule, due to faster degradation of road quality. Increased temperature leads to Additional patching required each year Adoption of base bitumen binders with rutting (of asphalt) and bleeding to fill cracks resulting from pavement higher softening points (including polymer and flushing (of seals). weakening. modification) for surface seals and asphalt. Precipitation Increased precipitation leads Increase the focus on annual patching Add wider paved shoulders to improve to increased average moisture to minimize exposed cracking resulting surface drainage. content in subgrade layers and from seasonal surface failure. reduced load-carrying capacity. Precipitation Increased precipitation leads Increase the focus on annual patching Add wider paved shoulders to improve to increased average moisture to minimize exposed cracking resulting surface drainage. content in subgrade layers and from seasonal surface failure. reduced load-carrying capacity. Fill sub-base where erosion has Increase base strength (thickness and/ occurred due to water infiltration. or quality) from the typical 150 mm to Follow with additional patching. 225–300 mm depending on precipitation levels, to increase protection of subgrade layers. Flooding (in Washaways and overtopping of Repair of localized washouts including Increase flood design return period excess of design road. cleaning culverts, replacing culverts, by increasing the size of culverts to flood) replacing subbase, and replacing accommodate new 1 in 50-year flood level asphalt surface. (in most cases will require raising the road to allow larger culvert to fit). Gravel roads Temperature Not applicable Precipitation Increased precipitation leads Regrade road localized to precipitation, Increase gravel wearing course thickness to to increased average moisture fill sub-base and reapply gravel top increase cover and protect subgrade layers. content in subgrade layers, and layer. reduced load-carrying capacity. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  95 Climate stressor Effect Repair and maintenance costs incurred Possible adaptation measure due to climate damage Flooding Washaways and overtopping of Same as for paved except application of Increase flood design return period by (in excess of road. gravel top layer rather than application increasing the size of culverts (in most design flood) of asphalt layer. cases will require raising the road to allow larger culvert to fit). Unpaved roads Temperature Not applicable Precipitation Increased precipitation leads Regrade road localized to precipitation, Upgrade to gravel road and increase gravel to increased average moisture fill sub-base and reapply earth top layer. wearing course thickness to increase cover content in subgrade layers, and and protect subgrade layers. reduced load carrying capacity. Flooding Washaways and overtopping of Same as for gravel except application of Increase flood design return period (in excess of road. earth top layer rather than application by increasing the size of culverts to design flood) of gravel. accommodate new 1 in 50-year flood level (in most cases will require raising the road to allow larger culvert to fit). In addition to the repair and maintenance costs estimated for roads, a disruption cost analysis was also conducted to estimate the delay costs of damaged roads requiring repair and maintenance, which in turn result in labor productivity effects. The disruption analysis evaluates the time that each road in the network is estimated to be fully under repair or fully out of service as a result of climate change. It is assumed that a trip lost during a road closure is equivalent to a half-day lost for each of the two passengers (we assume two passengers per vehicle). For the road analysis, disruption is quantified as hours of passenger delay due to road maintenance and rehabilitation. Specifically, road disruption is calculated by finding the amount of time that each vehicle/passenger will be delayed based on road work and closures caused by climate related stressors. If a road is being repaired, it is assumed that an eighth of a day is lost per passenger. The redundancy of the network is quantified by a redundancy factor which is a function of road length in each grid. When evaluating possible adaptation actions for roads, the study considers the costs associated with achieving climate-resilient repair and maintenance standards. New road infrastructure is constructed to resist high levels of temperature and precipitation, as well as the magnitude of a future 50-year flooding event, once existing infrastructure reaches its end of life or needs rehabilitation after climate thresholds are exceeded. Limitations • The Infrastructure Planning Support System model assumes that standard maintenance and repair practices are performed in the country regularly. While absolute climate change impacts may be larger in cases when maintenance and repair are inadequate, we assume that relative effects are an appropriate estimate. • Delay costs are approximated using general redundancy and traffic level standards by type of road. No behavioral changes in transport modes, travel times, or route selection are considered, as such analysis will require much more granular data and observations as well as detailed modeling beyond the scope of this analysis. • The analysis assumes that infrastructure is repaired as needed, resulting in increased expenses. This simplification does not reflect situations in which infrastructure is left unattended and potentially unusable, which may result in other effects such as further delays, loss of business, or road accidents and injuries. 96  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania B4. References Alewell, Christine, Pasquale Borrelli, Katrin Meusburger, and Panos Panagos. 2019. “Using the USLE: Chances, Challenges and Limitations of Soil Erosion Modelling.” International Soil and Water Conservation Research 7 (3): 203–25. https://doi.org/10.1016/j.iswcr.2019.05.004. Allen, Peter M., Jeffrey C. Arnold, and Bruce W. Byars. 1994. “Downstream Channel Geometry for Use in Planning-Level Models.” Journal of the American Water Resources Association 30 (4): 663–71. https://doi. org/10.1111/j.1752-1688.1994.tb03321.x. Allen, Richard G., Luis S. Pereira, Dirk Raes, and Martin Smith. 1998. “Crop Evapotranspiration - Guidelines for Computing Crop Water Requirements.” FAO Irrigation and Drainage Paper 56. Rome: FAO - Food and Agriculture Organization of the United Nations. https://www.fao.org/3/X0490E/x0490e00.htm#Contents. Benavidez, Rubianca, Bethanna Jackson, Deborah Maxwell, and Kevin Norton. 2018. “A Review of the (Revised) Universal Soil Loss Equation ((R)USLE): With a View to Increasing Its Global Applicability and Improving Soil Loss Estimates.” Hydrology and Earth System Sciences 22 (11): 6059–86. https://doi.org/10.5194/hess-22-6059-2018. Bernard, T. E., and M. Pourmoghani. 1999. “Prediction of Workplace Wet Bulb Global Temperature.” Applied Occupational and Environmental Hygiene 14 (2): 126–34. https://doi.org/10.1080/104732299303296. Berry, D. P., and J. E. Pryce. 2014. “Feed Efficiency in Growing and Mature Animals.” In Proceedings of the World Congress on Genetics Applied to Livestock Production. Vancouver, Canada. Bhowmik, Nani G. 1984. “Hydraulic Geometry of Floodplains.” Journal of Hydrology 68:369–401. https://doi. org/10.1016/0022-1694(84)90221-X. Bohmanova, J., I. Misztal, and J. B. Cole. 2007. “Temperature-Humidity Indices as Indicators of Milk Production Losses Due to Heat Stress.” Journal of Dairy Science 90 (4): 1947–56. https://doi.org/10.3168/jds.2006-513. Bondarenko, Maksym, David Kerr, Alessandro Sorichetta, and Andrew Tatem. 2020a. “Census/Projection-Disaggregated Gridded Population Datasets, Adjusted to Match the Corresponding UNPD 2020 Estimates, for 183 Countries in 2020 Using Built-Settlement Growth Model (BSGM) Outputs.” University of Southampton. https://doi.org/10.5258/SOTON/ WP00685. Bondarenko, Maksym, David Kerr, Alessandro Sorichetta, and Andrew Tatem. 2020b. “Census/Projection-Disaggregated Gridded Population Datasets for 51 Countries across Sub-Saharan Africa in 2020 Using Building Footprints.” University of Southampton. https://doi.org/10.5258/SOTON/WP00683. Brouwer, C., and M. Heibloem. 1986. “Irrigation Water Needs.” Irrigation Water Management Training Manual 3. https://www.fao.org/3/s2022e/s2022e00.htm#Contents. Brunner, Gary W., and Jurgen Gorbrecht. 1991. “A Muskingum-Cunge Channel Flow Routing Method for Drainage Networks.” Davis, CA: U.S. Army Corps of Engineers. Buchhorn, Marcel, B. Smets, L. Bertels, M. Lesiv, N-E. Tsendbazar, D. Masiliunas, L. Linlin, M. Herold, and S. Fritz. 2020. “Copernicus Global Land Service: Land Cover 100m: Globe (Version V3.0.1).” [Data set]. 10.5281/zenodo.3939050. Caminade, Cyril, Jolyon M. Medlock, Els Ducheyne, K. Marie McIntyre, Steve Leach, Matthew Baylis, and Andrew P. Morse. 2012. “Suitability of European Climate for the Asian Tiger Mosquito Aedes Albopictus: Recent Trends and Future Scenarios.” Journal of The Royal Society Interface 9 (75): 2708–17. https://doi.org/10.1098/rsif.2012.0138. CDC. 2020. “Heat Stress.” Centers for Disease Control and Prevention. 2020. https://www.cdc.gov/niosh/topics/ heatstress/default.html. Cervigni, Raffaello, Andres Losos, Paul Chinowsky, and James E. Neumann. 2017. “Enhancing the Climate Resilience of Africa’s Infrastructure: The Roads and Bridges Sector.” Africa Development Forum. Washington, DC: World Bank. https://doi.org/10.1596/978-1-4648-0466-3. Chen, Ping-Yu, Chi-Chung Chen, LanFen Chu and Bruce McCarl. 2015. “Evaluating the Economic Damage of Climate Change on Global Coral Reefs.” Global Environmental Change 30:12–20. https://doi.org/10.1016/j.gloenvcha.2014.10.011. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  97 Craig, M. H., R. W. Snow, and D. le Sueur. 1999. “A Climate-Based Distribution Model of Malaria Transmission in Sub- Saharan Africa.” Parasitology Today 15 (3): 105–11. https://doi.org/10.1016/S0169-4758(99)01396-4. Davis, Lucas, Paul Gertler, Stephen Jarvis, and Catherine Wolfram. 2021. “Air Conditioning and Global Inequality.” Global Environmental Change 69 (July):102299. https://doi.org/10.1016/j.gloenvcha.2021.102299. Desmet, P. J. J., and G. Govers. 1996. “A GIS Procedure for Automatically Calculating the USLE LS Factor on Topographically Complex Landscape Units.” Journal of Soil and Water Conservation 51 (5): 427–34. Doorenbos, J., and A. H. Kassam. 1979. “Yield Response to Water.” FAO Irrigation and Drainage Paper 33. Rome: FAO - Food and Agriculture Organization of the United Nations. Droogers, Peter, and Richard G Allen. 2001. “Estimating Reference Evapotranspiration Under Inaccurate Data Conditions.” Irrigation and Drainage Systems 16:33–45. Durigon, V.L., D.F. Carvalho, M.A.H. Antunes, P.T.S. Oliveira, and M.M. Fernandes. 2014. “NDVI Time Series for Monitoring RUSLE Cover Management Factor in a Tropical Watershed.” International Journal of Remote Sensing 35 (2): 441–53. https://doi.org/10.1080/01431161.2013.871081. Ebi, Kristie L., Jessica Hartman, Nathan Chan, John Mcconnell, Michael Schlesinger, and John Weyant. 2005. “Climate Suitability for Stable Malaria Transmission in Zimbabwe Under Different Climate Change Scenarios.” Climatic Change 73 (3): 375–93. https://doi.org/10.1007/s10584-005-6875-2. FAO. 2015. “Crop Ecological Requirements Database (ECOCROP).” Food and Agriculture Organization of the United Nations. 2015. https://www.fao.org/land-water/land/land-governance/land-resources-planning-toolbox/category/ details/en/c/1027491/. FAO. 2022a. “Crop Calendar: Information Tool for Crop Production.” Food and Agriculture Organization of the United Nations. 2022. https://cropcalendar.apps.fao.org/#/home. FAO, 2022b. “Global Livestock Environmental Assessment Model Version 3.0.” Rome, Italy: Food and Agriculture Organization of the United Nations. Fischer, Günther, Freddy Nachtergaele, Harrij van Velthuizen, Federica Chiozza, Gianluca Franceschini, Matieu Henry, Douglas Muchoney, and Sylvia Tramberend. 2021. “Global Agro-Ecological Zone v4 – Model Documentation.” Rome: FAO - Food and Agriculture Organization of the United Nations. https://doi.org/10.4060/cb4744en. Gourdji, Sharon M, Adam M Sibley, and David B Lobell. 2013. “Global Crop Exposure to Critical High Temperatures in the Reproductive Period: Historical Trends and Future Projections.” Environmental Research Letters 8 (2): 024041. https://doi.org/10.1088/1748-9326/8/2/024041. Hammer, W. I. 1981. “Soil Conservation Consultant Report.” Technical Note No.7. Bogor, Indonesia: Centre for Soil Research. Hatfield, Jerry L., and John H. Prueger. 2015. “Temperature Extremes: Effect on Plant Growth and Development.” Weather and Climate Extremes 10 (December):4–10. https://doi.org/10.1016/j.wace.2015.08.001. Honda, Yasushi, Masahide Kondo, Glenn McGregor, Ho Kim, Yue-Leon Guo, Yasuaki Hijioka, Minoru Yoshikawa, et al. 2014. “Heat-Related Mortality Risk Model for Climate Change Impact Projection.” Environmental Health and Preventive Medicine 19 (1): 56–63. https://doi.org/10.1007/s12199-013-0354-6. HOTOSM. 2020. “HOTOSM Points of Interest (OpenStreetMap Export).” Humanitarian OpenStreetMap Team (HOT); Humanitarian Data Exchange (HDX). https://data.humdata.org/organization/hot. Hudson, Norman. 1981. Soil Conservation. London: Batsford. Huizinga, Jan, Hans de Moel, and Wojciech Szewczyk. 2017. “Global Flood Depth-Damage Functions: Methodology and the Database with Guidelines.” EUR 28552. European Commission, Joint Research Centre. https://data.europa.eu/ doi/10.2760/16510. ILO. 2019a. “ILOSTAT.” International Labour Organization [Data set]. Available at ILOSTAT explorer. https://ilostat.ilo. org/data/. 98  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania ILO. 2019b. “Working on a Warmer Planet: The Impact of Heat Stress on Labour Productivity and Decent Work.” Geneva: International Labour Office. IMF. 2021. “Investment and Capital Stock Dataset (ICSD).” International Monetary Fund. 2021. https://data.imf. org/?sk=1CE8A55F-CFA7-4BC0-BCE2-256EE65AC0E4. ISO. 1989. “Hot Environments—Estimation of the Heat Stress on Working Man, Based on the WBGT-Index (Wet Bulb Globe Temperature).” ISO Standard 7243. Geneva: International Standards Organization. JICA. 2017. “Bridge Inspection and Diagnosis Manual.” Japan International Cooperation Agency (JICA); Oriental Consultants Global Co., Ltd. https://openjicareport.jica.go.jp/615/615/615_107_12288031.html. JICA. n.d. “Road Maintenance System Under the Framework of the Kenya Roads Board.” Accessed May 15, 2024. https://openjicareport.jica.go.jp/pdf/11716933_06.pdf. John, Denny, M. S. Narassima, Jaideep Menon, Jammy Guru Rajesh, and Amitava Banerjee. 2021. “Estimation of the Economic Burden of COVID-19 Using Disability-Adjusted Life Years (DALYs) and Productivity Losses in Kerala, India: A Model-Based Analysis.” BMJ Open 11 (8): e049619. https://doi.org/10.1136/bmjopen-2021-049619. Karamage, Fidele, Chi Zhang, Tong Liu, Andrew Maganda, and Alain Isabwe. 2017. “Soil Erosion Risk Assessment in Uganda.” Forests 8 (2): 52. https://doi.org/10.3390/f8020052. Kassam, A. H., H. T. van Velthuizen, P. H. Sloane, G. W Fischer, and M. M. Shah. 1991. “Agro-Ecological Land Resources Assessment for Agricultural Development Planning: A Case Study of Kenya.” Resources Data Base and Land Productivity. Rome: Food and Agriculture Organization of the United Nations (FAO) and International Institute for Applied Systems Analysis (IIASA). Khan, M. A., H. J. Lee, W. S. Lee, H. S. Kim, S. B. Kim, K. S. Ki, S. J. Park, J. K. Ha, and Y. J. Choi. 2007. “Starch Source Evaluation in Calf Starter: I. Feed Consumption, Body Weight Gain, Structural Growth, and Blood Metabolites in Holstein Calves.” Journal of Dairy Science 90 (11): 5259–68. https://doi.org/10.3168/jds.2007-0338. Kjellstrom, Tord, Chris Freyberg, Bruno Lemke, Matthias Otto, and David Briggs. 2018. “Estimating Population Heat Exposure and Impacts on Working People in Conjunction with Climate Change.” International Journal of Biometeorology 62 (3): 291–306. https://doi.org/10.1007/s00484-017-1407-0. Kong, Qinqin, and Matthew Huber. 2022. “Explicit Calculations of Wet-Bulb Globe Temperature Compared with Approximations and Why It Matters for Labor Productivity.” Earth’s Future 10 (3): e2021EF002334. https://doi. org/10.1029/2021EF002334. Lehner, Bernhard, and Günther Grill. 2013. “Global River Hydrography and Network Routing: Baseline Data and New Approaches to Study the World’s Large River Systems.” Hydrological Processes 27 (15): 2171–86. LeSueur, D., M. Craig, C. Fraser, B. Sharp, and C. Martin. 1998. “Towards an Atlas of Malaria Risk in Africa: First Technical Report of the MARA/ARMA Collaboration.” Durban: Mapping Malaria Risk in Africa/Atlas du Risque de a Malaria en Afrique. http://hdl.handle.net/10625/31644. Levy, Karen, Shanon M. Smith, and Elizabeth J. Carlton. 2018. “Climate Change Impacts on Waterborne Diseases: Moving Toward Designing Interventions.” Current Environmental Health Reports 5 (2): 272–82. https://doi.org/10.1007/ s40572-018-0199-7. Liljegren, James C., Richard A. Carhart, Philip Lawday, Stephen Tschopp, and Robert Sharp. 2008. “Modeling the Wet Bulb Globe Temperature Using Standard Meteorological Measurements.” Journal of Occupational and Environmental Hygiene 5 (10): 645–55. https://doi.org/10.1080/15459620802310770. Lima, Natália Ludmila Lins, Carolina Resende de Freitas Ribeiro, Hemilly Cristina Menezes de Sá, Izac Leopoldino- Júnior, Luigi Francis Lima Cavalcanti, Ronan Aparecido Valadares Santana, Iraides Ferreira Furusho-Garcia, and Idalmo Garcia Pereira. 2017. “Economic Analysis, Performance, and Feed Efficiency in Feedlot Lambs.” Revista Brasileira de Zootecnia 46 (10): 821–29. https://doi.org/10.1590/s1806-92902017001000005. Lo, A., S. El-Swaify, E. Dangler, and L. Shinshiro. 1985. “Effectiveness of EI30 as an Erosivity Index in Hawaii.” In Soil Erosion and Conservation, by S. El-Swaify, W. Moldenhauer, and A. Lo, 384–92. Ankeny, IA: Soil Conservation Society of America. https://www.semanticscholar.org/paper/Effectiveness-of-EI30-as-an-erosivity-index-in-Lo-El%E2%80%90Swa ify/2d4b1c8511e49c9d24c831cba7636db50b9a2a67. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  99 Mackle, T. R., C. R. Parr, G. K. Stakelum, A. M. Bryant, and K. L. MacMillan. 1996. “Feed Conversion Efficiency, Daily Pasture Intake, and Milk Production of Primiparous Friesian and Jersey Cows Calved at Two Different Liveweights.” New Zealand Journal of Agricultural Research 39:357–70. https://doi.org/10.1080/00288233.1996.9513195. Mahgoub, O., C. D. Lu, and R. J. Early. 2000. “Effects of Dietary Energy Density on Feed Intake, Body Weight Gain and Carcass Chemical Composition of Omani Growing Lambs.” Small Ruminant Research: The Journal of the International Goat Association 37 (1–2): 35–42. https://doi.org/10.1016/s0921-4488(99)00132-7. Mauger, Guillaume, Yoram Bauman, Tamilee Nennich, and Eric Salathé. 2015. “Impacts of Climate Change on Milk Production in the United States.” The Professional Geographer 67 (1): 121–31. https://doi.org/10.1080/00330124. 2014.921017. McCool, D. K., G. R. Foster, C. K. Mutchler, and L. D. Meyer. 1989. “Revised Slope Length Factor the Universal Soil Loss Equation.” Transactions of the American Society of Agricultural Engineers 32:1571–76. https://doi.org/doi: 10.13031/2013.31192. Moran, D. S., K. B. Pandolf, A. Laor, Y. Heled, W. T. Matthew, and R. R. Gonzalez. 2003. “Evaluation and Refinement of the Environmental Stress Index for Different Climatic Conditions.” Journal of Basic and Clinical Physiology and Pharmacology 14 (1): 1–16. https://doi.org/10.1515/JBCPP.2003.14.1.1. Nambajimana, Jean de Dieu, Xiubin He, Ji Zhou, Meta Francis Justine, Jinlin Li, Dil Khurram, Richard Mind’je, and Gratien Nsabimana. 2020. “Land Use Change Impacts on Water Erosion in Rwanda.” Sustainability 12 (1): 50. https://doi.org/10.3390/su12010050. NASA. 2022. “MOD13Q1—MODIS/Terra Vegetation Indices 16-Day L3 Global 250m SIN Grid.” NASA Goddard Space Flight Center. 2022. https://ladsweb.modaps.eosdis.nasa.gov/. Nichols, Gordon, Iain Lake, and Clare Heaviside. 2018. “Climate Change and Water-Related Infectious Diseases.” Atmosphere 9 (10): 385. https://doi.org/10.3390/atmos9100385. Noblet, J., H. Fortune, X. S. Shi, and S. Dubois. 1994. “Prediction of Net Energy Value of Feeds for Growing Pigs.” Journal of Animal Science 72 (2): 344–54. https://doi.org/10.2527/1994.722344x. Nuttall, James G., Kirsten M. Barlow, Audrey J. Delahunty, Brendan P. Christy, and Garry J. O’Leary. 2018. “Acute High Temperature Response in Wheat.” Agronomy Journal 110 (4): 1296–1308. https://doi.org/10.2134/ agronj2017.07.0392. OECD. 2023. “Working Age Population (Indicator).” Text. 2023. https://doi.org/10.1787/d339918b-en. Panagos, Panos, Pasquale Borrelli, Katrin Meusburger, Bofu Yu, Andreas Klik, Kyoung Jae Lim, Jae E. Yang, et al. 2017. “Global Rainfall Erosivity Assessment Based on High-Temporal Resolution Rainfall Records.” Scientific Reports 7 (1): 4175. https://doi.org/10.1038/s41598-017-04282-8. Philipsborn, Rebecca, Sharia M. Ahmed, Berry J. Brosi, and Karen Levy. 2016. “Climatic Drivers of Diarrheagenic Escherichia Coli Incidence: A Systematic Review and Meta-Analysis.” The Journal of Infectious Diseases 214 (1): 6–15. https://doi.org/10.1093/infdis/jiw081. Ponce, Victor M. 2014. “The Muskingum-Cunge Method.” In Fundamentals of Open-Channel Hydraulic. Online textbook. Potter, Philip, Navin Ramankutty, Elena M. Bennett, and Simon D. Donner. 2010. “Characterizing the Spatial Patterns of Global Fertilizer Application and Manure Production.” Earth Interactions 14 (2): 1–22. https://doi. org/10.1175/2009EI288.1. Prasad, P. V. V., S. A. Staggenborg, and Z. Ristic. 2015. “Impacts of Drought and/or Heat Stress on Physiological, Developmental, Growth, and Yield Processes of Crop Plants.” In Advances in Agricultural Systems Modeling, edited by L.R. Ahuja, V.R. Reddy, S.A. Saseendran, and Qiang Yu, 301–55. Madison, WI, USA: American Society of Agronomy and Soil Science Society of America. https://doi.org/10.2134/advagricsystmodel1.c11. Rahimi, Jaber, John Yumbya Mutua, An M. O. Notenbaert, Karen Marshall, and Klaus Butterbach-Bahl. 2021. “Heat Stress Will Detrimentally Impact Future Livestock Production in East Africa.” Nature Food 2 (2): 88–96. https://doi. org/10.1038/s43016-021-00226-8. 100  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Ramankutty, N., A.T. Evan, C. Monfreda, and J. A. Foley. 2010. “Global Agricultural Lands: Croplands, 2000.” NASA Socioeconomic Data and Applications Center (SEDAC). http://sedac.ciesin.columbia.edu/es/aglands.html. Renard, K. G., G. R. Foster, G. Weesies, D. McCool, and D. Yoder. 1997. “Predicting Soil Erosion by Water: A Guide to Conservation Planning with the Revised Universal Soil Loss Equation (RUSLE).” Washington, DC: U.S. Dept. of Agriculture, Agricultural Research Service. Roberts, E H. 1988. “Temperature and Seed Germination.” Symposia of the Society for Experimental Biology 42 (January):109–32. Romanello, Marina, Alice McGushin, Claudia Di Napoli, Paul Drummond, Nick Hughes, Louis Jamart, Harry Kennard, et al. 2021. “The 2021 Report of the Lancet Countdown on Health and Climate Change: Code Red for a Healthy Future.” The Lancet 398 (10311): 1619–62. https://doi.org/10.1016/S0140-6736(21)01787-6. Salama, A. A. K., G. Caja, S. Hamzaoui, B. Badaoui, A. Castro-Costa, D. A. E. Façanha, M. M. Guilhermino, and R. Bozzi. 2014. “Different Levels of Response to Heat Stress in Dairy Goats.” Small Ruminant Research, Special Issue: Industrial and Rural Activities in the Goat Sector including Science, Innovation and Development, 121 (1): 73–79. https://doi. org/10.1016/j.smallrumres.2013.11.021. Salman, M., M. García-Vila, E. Fereres, D. Raes, and P. Steduto. 2021. “The AquaCrop Model – Enhancing Crop Water Productivity: Ten Years of Development, Dissemination and Implementation 2009–2019.” FAO Water Report No. 47. Rome, Italy: FAO. https://doi.org/10.4060/cb7392en. Seo, Seongwon, Kyewon Kang, Seoyoung Jeon, Mingyung Lee, Sinyong Jeong, and Luis Tedeschi. 2021. “Development of a Model to Predict Dietary Metabolizable Energy from Digestible Energy in Beef Cattle.” Journal of Animal Science 99 (7): skab182. https://doi.org/10.1093/jas/skab182. Steduto, Pasquale, Theodore C. Hsiao, Elias Fereres, and Dirk Raes. 2012. “Crop Yield Response to Water.” FAO Irrigation and Drainage Paper 66. Rome: FAO. https://www.fao.org/3/i2800e/i2800e00.htm. St-Pierre, N. R., B. Cobanov, and G. Schnitkey. 2003. “Economic Losses from Heat Stress by US Livestock Industries.” Journal of Dairy Science 86 (June): E52–77. https://doi.org/10.3168/jds.S0022-0302(03)74040-5. United Nations. 2022. “Database on Household Size and Composition 2022.” Department of Economic and Social Affairs. https://www.un.org/development/desa/pd/data/household-size-and-composition. Wade Rain Irrigation Systems. n.d. “How Much Does a Center Pivot Irrigation System Cost?” Wade Rain Irrigation Systems. Accessed May 13, 2024. https://www.waderain.com/index.php?page=pivot_faqs#:~:text=The%20cost%20 per%20acre%20for,%2475%2C000.00%20and%20%2480%2C000.00%2C%20plus%20freight. Wahid, A, S Gelani, M Ashraf, and M Foolad. 2007. “Heat Tolerance in Plants: An Overview.” Environmental and Experimental Botany 61 (3): 199–223. https://doi.org/10.1016/j.envexpbot.2007.05.011. Wang, Tingting, and Fubao Sun. 2022. “Global Gridded GDP Data Set Consistent with the Shared Socioeconomic Pathways.” Scientific Data 9 (1): 221. https://doi.org/10.1038/s41597-022-01300-x. Wen, Chaoliang, Wei Yan, Jiangxia Zheng, Congliang Ji, Dexiang Zhang, Congjiao Sun, and Ning Yang. 2018. “Feed Efficiency Measures and Their Relationships with Production and Meat Quality Traits in Slower Growing Broilers.” Poultry Science 97 (7): 2356–64. https://doi.org/10.3382/ps/pey062. WHO. 2014. “Quantitative Risk Assessment of the Effects of Climate Change on Selected Causes of Death, 2030s and 2050s.” World Health Organization. https://apps.who.int/iris/handle/10665/134014. Williams, J. R. 1995. “The EPIC Model.” In Computer Models of Watershed Hydrology, by V. P. Singh, 909–1000. Highlands Ranch, CO: Water Resources Publications. Wischmeier, Walter H., and Dwight David Smith. 1978. “Predicting Rainfall Erosion Losses: A Guide to Conservation Planning.” Washington, DC: U.S. Department of Agriculture. https://naldc.nal.usda.gov/catalog/CAT79706928. Wolf, Jennyfer, Richard Johnston, Paul R. Hunter, Bruce Gordon, Kate Medlicott, and Annette Prüss-Ustün. 2019. “A Faecal Contamination Index for Interpreting Heterogeneous Diarrhoea Impacts of Water, Sanitation and Hygiene Interventions and Overall, Regional and Country Estimates of Community Sanitation Coverage with a Focus on Low- and Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  101 Middle-Income Countries.” International Journal of Hygiene and Environmental Health 222 (2): 270–82. https://doi. org/10.1016/j.ijheh.2018.11.005. World Bank. 2018a. “Second Bridges Improvement and Maintenance Program.” Report No: RES55302. World Bank. World Bank. 2018b. “When Water Becomes a Hazard: A Diagnostic Report on The State of Water Supply, Sanitation, and Poverty in Pakistan and Its Impact on Child Stunting.” Washington, DC: World Bank. https://openknowledge.worldbank. org/handle/10986/30799. World Bank. n.d. “Households and NPISHs Final Consumption Expenditure.” International Comparison Program, World Bank | World Development Indicators database, World Bank | Eurostat-OECD PPP Programme. https://data.worldbank. org/indicator/NE.CON.PRVT.PP.KD. 102  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Appendix C: Data Sources C1. Heat and labor productivity Data Source Labor Supply Working age population (ages 15–64) United Nations Population Prospects, 1995–2050, medium variant estimate (United Nations 2022b) Labor force (total employment and by sector and Data provided by World Bank team. occupation) Weekly hours worked Data provided by World Bank team. Exposure and Air Conditioning Outdoor exposure by occupation Occupational requirements survey (BLS 2022) Household size United Nations Household Size and Composition, 2022 (United Nations 2022a) Household income World Bank World Development Indicators, Purchasing Power Parity constant 2017 international $ (World Bank, n.d.) Air conditioning coverage Household Budget Survey (National Bureau of Statistics, Tanzania 2022) Labor Force Spatial Allocation Population distribution (including Zanzibar) Administrative Units Population Distribution Report (Ministry of Finance and Planning National Bureau of Statistics, Tanzania and Presidents’ Office – Finance and Planning Office of the Chief Government Statistician, Zanzibar 2022) Population weighting (for combining Mainland and National Population Projections (United Republic of Tanzania National Bureau of Statistics, Zanzibar labor data) Ministry of Finance and Planning, Office of the Chief Government Statistician 2020) GDP distribution Global gridded GDP data set consistent with the shared socioeconomic pathways (Wang and Sun 2022) Cropland distribution NASA Global Agricultural Lands: Croplands data set (Ramankutty et al. 2010) C2. Human health Data Source Labor Supply & Population Working age population (ages 15–64) United Nations Population Prospects, 1995–2050, medium variant estimate (United Nations 2022b) Labor force (total employment and by sector and International Labor Organization Labor Force Statistics, 2016–20 (ILO 2023) occupation) Weekly hours worked International Labor Organization Labor Force Statistics, 2016–20 (ILO 2023) Historical Illness Incidence Death and incidence rate of diseases Global Health Data Exchange (GHDx), 2016–19, medium estimate, Institute for Health Metrics and Evaluation, University of Washington (IHME 2021). http://ghdx.healthdata. org/about-ghdx. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  103 Data Source Baseline WASH coverage Joint Monitoring Programme for Water Supply, Sanitation and Hygiene, 2021 Development Scenario WASH Targets Rural population (% of total population), Tanzania World Bank Group, 2023. https://data.worldbank.org/indicator/SP.RUR.TOTL. ZS?locations=TZ Rural safely managed water access, rural improved United Republic of Tanzania, National Five-Year Development Plan 2021/22 – 2025/26. sanitation access, and urban safely managed water https://faolex.fao.org/docs/pdf/tan205461.pdf access Sustainable Development Goals in Tanzania: SDG 6 United Nations, Tanzania. https://tanzania.un.org/en/sdgs/6 C3. Crop production Data Source Historical crop area, production, and revenue statistics Crop area United Republic of Tanzania Ministry of Agriculture. 2018/2019. “Basic Data: Crop Sub Sector (Tanzania Mainland.” Provided by the World Bank Crop area and production Food and Agriculture, FAOSTAT – Crops and Livestock Products, https://www.fao.org/ faostat/en/#data/QCL, (accessed July 2023) Crop revenues FAO, FAOSTAT – Value of Agricultural Production, https://www.fao.org/faostat/en/#data/ QCL, (accessed July 2023) Spatial distribution of crop production International Food Policy Research Institute, 2020, "Spatially-Disaggregated Crop Production Statistics Data in Africa South of the Sahara for 2017", https://doi. org/10.7910/DVN/FSSKBW, Harvard Dataverse, V2 Crop calendars, coefficients, and thresholds Crop calendars FAO, Global Agro-Ecological Zones (GAEZ v4), https://gaez-services.fao.org/apps/theme-3/ Crop coefficients Allen, Richard G., Luis S. Pereira, Dirk Raes, and Martin Smith. 1998. “Crop Evapotranspiration - Guidelines for Computing Crop Water Requirements.” FAO Irrigation and Drainage Paper 56. Rome: FAO - Food and Agriculture Organization of the United Nations. https://www.fao.org/3/X0490E/x0490e00.htm#Contents Crop yield response to water coefficients Steduto, Pasquale, Theodore C. Hsiao, Elias Fereres, and Dirk Raes. 2012. “Crop Yield Response to Water.” FAO Irrigation and Drainage Paper 66. Rome: FAO. https://www.fao. org/3/i2800e/i2800e00.htm. Temperature thresholds FAO. 2015. “Crop Ecological Requirements Database (ECOCROP).” Food and Agriculture Organization of the United Nations. 2015. https://www.fao.org/land-water/land/land- governance/land-resources-planning-toolbox/category/details/en/c/1027491/. C4. Livestock production Data Source Historical headcounts, production, and revenue statistics Livestock production, headcounts, and productivity Food and Agriculture, FAOSTAT – Crops and Livestock Products, https://www.fao.org/ faostat/en/#data/QCL, (accessed July 2023) Revenues and producer prices FAO, FAOSTAT – Value of Agricultural Production, https://www.fao.org/faostat/en/#data/ QCL, (accessed July 2023) 104  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Data Source Spatial distribution Distribution of ruminants by production systems Robinson, Timothy P., Philip Thornton, Gianluca Franceschini, Russ Kruska, Federica Chiozza, An Notenbaert, Giuliano Cecchi, et al. 2018. “Global Distribution of Ruminant Livestock Production Systems V5 (5 Minutes of Arc).” Harvard Dataverse. https://doi. org/10.7910/DVN/WPDSZE. Distribution of chickens by production systems Gilbert, Marius; Nicolas, Gaëlle; Cinardi, Giusepina; Van Boeckel, Thomas P.; Vanwambeke, Sophie; Wint, G. R. William; Robinson, Timothy P., 2018, "Global chickens distribution in 2010 (5 minutes of arc)", https://doi.org/10.7910/DVN/SUFASB Distribution of pigs by production systems Gilbert, Marius; Nicolas, Gaëlle; Cinardi, Giusepina; Van Boeckel, Thomas P.; Vanwambeke, Sophie; Wint, William G. R.; Robinson, Timothy P., 2018, "Global pigs distribution in 2010 (5 minutes of arc)", https://doi.org/10.7910/DVN/33N0JG. Distribution of animal headcounts FAO, Gridded Livestock of the World v4, 2015 Agricultural Transformation Playbook, PETS, 2023 (corrections and projections for cattle) Feed Feed intake statistics FAO. 2022. “Global Livestock Environmental Assessment Model Version 3.0.” Rome, Italy: Food and Agriculture Organization of the United Nations. FAO. 2020. “Supply Utilization Accounts.” Food and Agriculture Organization of the United Nations (FAO). 2020. http://www.fao.org/faostat/en/#data/PP/. Feed prices World Integrated Trade Solution (WITS). 2017. “Trade Statistics by Product.” World Bank; UNCTAD. https://wits.worldbank.org/trade/country-byhs6product.aspx?lang=en. C5. Soil erosion Data Source Land use land cover Land cover Li, M., Cao, S., Zhu, Z., Wang, Z., Myneni, R. B., and Piao, S. (2023) Spatiotemporally consistent global dataset of the GIMMS Normalized Difference Vegetation Index (PKU GIMMS NDVI) from 1982 to 2022, Earth Syst. Sci. Data Discuss. [preprint], https://doi. org/10.5194/essd-2023-1. Soil characteristics and management Physical properties of soil Poggio, L., de Sousa, L. M., Batjes, N. H., Heuvelink, G. B. M., Kempen, B., Ribeiro, E., and Rossiter, D. (2021) SoilGrids 2.0: producing soil information for the globe with quantified spatial uncertainty, SOIL, 7, 217–240, https://doi.org/10.5194/soil-7-217-2021. Slope and slope length Amatulli, G., Domisch, S., Tuanmu, M.-N., Parmentier, B., Ranipeta, A., Malczyk, J., and Jetz, W. (2018) A suite of global, cross-scale topographic variables for environmental and biodiversity modeling. Scientific Data volume 5, Article number: 180040. DOI: doi:10.1038/sdata.2018.40. Historical rainfall-erosivity Panagos, P.; Borrelli, P.; Meusburger, K.; Yu, B.; Klik, A.; Jae Lim, K.; Yang, J.E.; Ni, J.; Miao, C.; Chattopadhyay, N.; et al. Global rainfall erosivity assessment based on high-temporal resolution rainfall records. Sci. Rep. 2017, 7, 4175. Historical crop area, production, and revenue statistics Crop area, production, and revenues International Food Policy Research Institute, 2020, “Spatially-Disaggregated Crop Production Statistics Data in Africa South of the Sahara for 2017”, https://doi. org/10.7910/DVN/FSSKBW, Harvard Dataverse, V2 Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  105 C6. Inland flooding Data Source Hydrology and topography Flood curve numbers Wischmeier, Walter H., and Dwight David Smith. 1978. Predicting Rainfall Erosion Losses: A Guide to Conservation Planning. Washington, DC: U.S. Department of Agriculture. Annual exceedance probability of flooding events World Bank. 2021. Climate Change Knowledge Portal https://climateknowledgeportal. worldbank.org/. Land cover and soil infiltration Buchhorn, Marcel, B. Smets, L. Bertels, M. Lesiv, N. E. Tsendbazar, D. Masiliunas, L. Linlin, M. Herold, and S. Fritz. 2020. “Copernicus Global Land Service: Land Cover 100m: Globe (Version V3.0.1).” River network Lehner, Bernhard, and Günther Grill. 2013. “Global River Hydrography and Network Routing: Baseline Data and New Approaches to Study the World’s Large River Systems.” Hydrological Processes 27 (15): 2171–86. Hydrological basins Linke, Simon, Bernhard Lehner, Camille Ouellet Dallaire, Joseph Ariwi, Günther Grill, Mira Anand, Penny Beames, et al. 2019. “Global Hydro-Environmental Sub-Basin and River Reach Characteristics at High Spatial Resolution.” Scientific Data 6 (1): 283. https://doi. org/10.1038/s41597-019-0300-6. Resources at risk Capital stock and capital-output ratio IMF. 2021. “Investment and Capital Stock Dataset (ICSD).” International Monetary Fund. 2021. https://data.imf.org/?sk=1CE8A55F-CFA7-4BC0-BCE2-256EE65AC0E4. Capital spatial distribution Wang, Tingting, and Fubao Sun. 2022. “Global Gridded GDP Data Set Consistent with the Shared Socioeconomic Pathways.” Scientific Data 9 (1): 221. https://doi.org/10.1038/ s41597-022-01300-x. Flood damage curves Huizinga, Jan, Hans de Moel, and Wojciech Szewczyk. 2017. “Global Flood Depth-Damage Functions: Methodology and the Database with Guidelines.” EUR 28552. European Commission, Joint Research Centre. https://data.europa.eu/doi/10.2760/16510. Floodproofing effectiveness and costs 360° Resilience : A Guide to Prepare the Caribbean for a New Generation of Shocks – Overview of Engineering Options for Increasing Infrastructure Resilience in the Caribbean (English). Washington DC: World Bank Group. http://documents.worldbank.org/curated/ en/260061635280496287/360-Resilience-A-Guide-to-Prepare-the-Caribbean-for-a- New-Generation-of-Shocks-Overview-of-Engineering-Options-for-Increasing-Infrastructure- Resilience-in-the-Caribbean. C7. Bridges Data Source Bridge inventory Current road inventory (used to construct bridge GloBio, “GRIP Global Roads Database”. https://www.globio.info/download-grip-dataset inventory) River network (used to construct bridge inventory) Lehner, Bernhard, and Günther Grill. 2013. “Global River Hydrography and Network Routing: Baseline Data and New Approaches to Study the World’s Large River Systems.” Hydrological Processes 27 (15): 2171–86. Maintenance, repair, and disruption costs Unit costs for maintenance and repair Arnoldi, Marleny. 2023. “Professors Suggest Vital Bridge Assessment Improvements to Contain Costs”. Engineering News. https://www.engineeringnews.co.za/article/professors- suggest-vital-bridge-assessment-improvements-to-cont. 106  |  Background Paper for Country Climate and Development Report: United Republic of Tanzania Data Source World Bank. 2018. “Second Bridges Improvement and Maintenance Program.” Report No: RES55302. World Bank. https://documents1.worldbank.org/curated/ en/099081523021032171/pdf/P1619290f8b0190cb0b3af07c1215bb7431.pdf. Road traffic data Japan International Cooperation Agency (JICA) Study Team. N.d. “Final Report: Annex 7: Second Field Study”. https://openjicareport.jica.go.jp/pdf/11716933_06.pdf. C8. Roads Data Source Road inventory Current road inventory. GloBio, “GRIP Global Roads Database”. https://www.globio.info/download-grip-dataset. Data received from the World Bank from the Tanzania National Roads Agency (TANROADS) Road Vulnerability Assessment. Maintenance, repair, and disruption costs Unit costs for maintenance, repair and disruption. Cervigni, Raffaello, Andres Losos, Paul Chinowsky, and James E. Neumann. 2017. “Enhancing the Climate Resilience of Africa’s Infrastructure: The Roads and Bridges Sector.” Africa Development Forum. Washington, DC: World Bank. https://doi. org/10.1596/978-1-4648-0466-3. Compass International 2024 Global Construction Costs Database. Compass International, Inc. https://compassinternational.net/. Road traffic data Sub-Saharan Africa Transport Policy Program (SSATP). The Road Network Evaluation Tools model (RONET). The World Bank Group. https://www.ssatp.org/en/page/road-network- evaluation-tools-ronet. Background Paper for Country Climate and Development Report: United Republic of Tanzania  |  107 World Bank Tanzania 50 Mirambo Street P. O. Box 2054 Dar es Salaam, Tanzania Read more on linkages Tel : +255-22-216-3200 between climate and tanzaniaalert@worldbank.org development in Tanzania