The “governance index” is related to potential effectiveness of any governmental response during and after exposure to climate stress in 2010. Generally well-structured governmental institutions are able to enhance the resilience of its members and citizen promoting adaptation strategies and action. The index results from the first cluster of the Principal Component Analysis preformed among 18 potential variables. The analysis identifies seven dominant variables, namely “openness to external assistance”, “regulatory quality”, “government effectiveness”, “voice and accountability”, “failed state”, “perceived corruption” and “criminal rate”, assigning equal weight of 0.1525 except to “criminal rate” that retrieve a weight of 0.085. Before to perform the analysis “criminal rate” was log transformed to shorten the extreme variation and then all variables were score-standardized (converted to distribution with average of 0 and standard deviation of 1; inverse method was applied for “criminal rate”) in order to be comparable. All the variables are country base and the tabular data were linked by country to the national boundaries shapefile (FAO/GAUL) and then converted into raster format (resolution 0.5 arc-minute). The variable “openness to external assistance” was compute as the average values of the period 2008-2012 of the KOF Index of Globalization that serves as a proxy for a country’s level of global integration. It uses data on three dimensions of globalization (economic, social and political) to create an overall globalization score between zero and 100 that measures a country’s level of integration in the global system. The variables “regulatory quality”, “government effectiveness” and “voice and accountability” are gathered from the World Bank’s Worldwide Governance Indicators, considering the average of the period 2008-2012. “Regulatory quality” captures perceptions of the ability of the government to formulate and implement sound policies and regulations that permit and promote private sector development. “Government effectiveness” captures perceptions of the quality of public services, the quality of the civil service and the degree of its independence from political pressures, the quality of policy formulation and implementation, and the credibility of the government's commitment to such policies. “Voice and accountability captures perceptions of the extent to which a country's citizens are able to participate in selecting their government, as well as freedom of expression, freedom of association, and a free media. The “failed state” variable is based on the average 2008-2012 of the fragile state index that is a compilation of 12 socio-economic and political variables made by the Fund for Peace. It aims to capture the vulnerability to any kind of stress of a certain state. The “perceived corruption” variable is based on the homonym index published by Transparency International (NGO) averaged from 2008 to 2012. It assesses the level of corruption using qualitative surveys. Finally the “criminal rate” records the average rate (per 100,000 people) of unlawful death, as defined and assessed by the UN Office on Drugs and Crime, in the period 2008-2012. A country with a higher degree of global integration will be better positioned to obtain disaster assistance from the international community. Autarkic or less well-integrated countries may be less able or willing to receive outside assistance in the event of extreme weather events. Such assistance could take the form of aid from bilateral or multilateral donors or international humanitarian organizations, or it could be in the form of remittances from overseas family or diasporas of the affected population. The Worldwide Governance Indicators define the level of expected assistance to citizens to better coping and manage with climate stressors and disaster events in situations measuring the good governance degree of a state. The failed state index measures the state ability to implement adaptation strategies measuring its vulnerability. People living in countries with higher level of corruption are thought to have more difficulties recovering from climate change impacts, due to limited governmental support reaching affected population compared to states with lower level of corruption. Corruption can further be of particular importance when it comes to the distribution of and access to emergency relief resources. Finally the criminal rate is included because violence severely limits a government’s ability to deliver assistance and indicate a lack of authority by central government. This dataset has been produced in the framework of the “Climate change predictions in Sub-Saharan Africa: impacts and adaptations (ClimAfrica)” project, Work Package 4 (WP4). More information on ClimAfrica project is provided in the Supplemental Information section of this metadata.

The “natural resources sensitivity” symbolizes the ecosystem vitality and degree of conservation in 2010. Deforestation and loss of water resources quality may render certain areas more sensitive to climate stressors on account of the loss of normal vegetation cover, the depletion of biodiversity, the reduction in ecosystem services and significant loss of beneficial assets. The index results from the first cluster of the Principal Component Analysis preformed among 16 potential variables. The analysis identify five dominant variables, namely “water availability per capita”, “net primary production”, “forest accessibility”, “vegetation continuity” and “climatic resources availability”, assigning respectively the weights of 0.19, 0.21, 0.165, 0.21 and 0.225. Before to perform the analysis the variables “water availability per capita”, “forest accessibility” and “vegetation continuity” were log transformed to shorten the extreme variation and then together with the other two variables were score-standardized (converted to distribution with average of 0 and standard deviation of 1; all variables with inverse method) in order to be comparable. The 6 arc-minute grid “water availability per capita” of 2005 was computed by sum of the run-off and discharge grids produced by World Water Development Report II and then sampled at 0.5 arc-minutes. A focal statistic ran with a radius of 55 cells (about 50 Km). This had a smoothing effect and represents some of the extend influence of major rivers as a resources for local people. To calculate the available water per capita it was then divided by the population. The 5 arc-minute grid “net primary production” of 2000 was gathered from FAO GeoNetwork and sampled at 0.5 arc-minute. Also in this case a focal statistic ran with a radius of 22 cells (about 20 Km) in order to represents the extend effect of primary production as natural resources for local people. The 0.5 arc-minute grid “forest accessibility” was build using the grid of travel distance in minutes to large cities (which one with population greater than 50,000 people), produced by the European Commission and the World Bank to represent the connectivity in 2000, and the grid of forest occurrence, extracted from the FAO Global Land Cover-SHARE dataset of 2014. The result measures the distance in minutes between forest and cities, thus is a proxy for remoteness and naturalness of forest. The 0.125 arc-minute grid “vegetation continuity” of 2010 were collected from University of Maryland and NASA and sampled at 0.5 arc-minute. A focal statistic ran with a radius of 55 cells (about 50 Km). This had a smoothing effect and represents some of the extend influence of vegetation concentration as a resources for local people. Finally the 0.5 arc-minute grid “climatic resources availability” was produced within the ClimAfrica project. The “water availability per capita” represents the potential water available per people in a certain area. We can consider the area with small values more sensitive to climatic stress, because lack a buffer of water resources, precious in a prevalently rain-fed agricultural system like in Africa. The “net primary production” and the “vegetation continuity” are proxies of the potential vegetal productivity available in a certain area. Moreover “vegetation continuity” is an indicator of abundance of natural ecosystem services that can reduce the sensitivity of human-environment systems. The “forest accessibility” assessing the distance between human and natural system measure the anthropogenic degree of a forest. A forest recording a high anthropogenic degree (thus near in terms of minute from a city) may potentially be threaded by human activity and thus represent a fragile ecosystem. Finally the “climatic resources availability” is an indicator of the climatic potential for biomass production. It is based on the climatically determined biomass productivity index that is a proxy for the atmospheric energy available for biomass production, as expressed by accumulated temperature, adjusted for drought stress. This dataset has been produced in the framework of the “Climate change predictions in Sub-Saharan Africa: impacts and adaptations (ClimAfrica)” project, Work Package 4 (WP4). More information on ClimAfrica project is provided in the Supplemental Information section of this metadata.

The raster dataset consists of a 1km score grid for cereal storage location, produced under the scope of FAO’s Hand-in-Hand Initiative, Geographical Information Systems - Multicriteria Decision Analysis for value chain infrastructure location. The location score is achieved by processing sub-model outputs that characterize logistical factors for selected crop warehouse location: • Supply: Crop. • Demand: Human population density, Major cities population (national and bordering countries). • Infrastructure/accessibility: main transportation infrastructure. It consists of an arithmetic weighted sum of normalized grids (0 to 100): ("Crop Production" * 0.4) + ("Human Population Density" * 0.2) + (“Major Cities Accessibility” * 0.3) + (”Regional Cities Weighted Accessibility” *0.1 )

The 1km raster dataset represents top location score areas filtered by exclusive criteria: access to finance, distance to major roads and access to IT. The layer is produced under the scope of FAO’s Hand-in-Hand Initiative, Geographical Information Systems - Multicriteria Decision Analysis for value chain infrastructure location. The location score is achieved by processing sub-model outputs characterizing logistical factors for crop warehouse siting: Supply, demand, Infrastructure/accessibility. Access to finance and roads are defined using a linear distance threshold: • Banks - approx. 20 km (0.18 degree) buffer radius. • Major roads - approx. 2 km (0.018 degree) buffer radius. Access to IT is characterized applying the mobile broadband coverage map.

The raster dataset consists of a 1km score grid for major cereals storage location, produced under the scope of FAO’s Hand-in-Hand Initiative, Geographical Information Systems - Multicriteria Decision Analysis for value chain infrastructure location. Major cereals include barley, millet, wheat, maize, sorghum, and rice. The location score is achieved by processing sub-model outputs that characterize logistical factors for selected crop warehouse location: • Supply: Crop. • Demand: Human population density, Major cities population (national and bordering countries). • Infrastructure/accessibility: main transportation infrastructure. It consists of an arithmetic weighted sum of normalized grids (0 to 100): ("Crop Production" * 0.4) + ("Human Population Density" * 0.2) + (“Major Cities Accessibility” * 0.3) + (”Regional Cities Weighted Accessibility” *0.1 )

The 1km raster dataset represents top location score areas filtered by exclusive criteria: access to finance, distance to major roads and access to IT (mobile broadband connection). The layer is produced under the scope of FAO’s Hand-in-Hand Initiative, Geographical Information Systems - Multicriteria Decision Analysis for value chain infrastructure location. The location score is achieved by processing sub-model outputs characterizing logistical factors for crop warehouse siting: Supply, demand, Infrastructure/accessibility. Access to finance and roads are defined using a linear distance threshold: • Banks - approx. 20 km (0.18 degree) buffer radius. • Major roads - approx. 2 km (0.018 degree) buffer radius. Access to IT is characterized applying the mobile broadband coverage map.

The raster dataset consists of a 1km score grid for vegetables storage location, produced under the scope of FAO’s Hand-in-Hand Initiative, Geographical Information Systems - Multicriteria Decision Analysis for value chain infrastructure location. The location score is achieved by processing sub-model outputs that characterize logistical factors for selected crop warehouse location: • Supply: Crop. • Demand: Human population density, Major cities population (national and bordering countries). • Infrastructure/accessibility: main transportation infrastructure. It consists of an arithmetic weighted sum of normalized grids (0 to 100): ("Crop Production" * 0.4) + ("Human Population Density" * 0.2) + (“Major Cities Accessibility” * 0.3) + (”Regional Cities Weighted Accessibility” *0.1 )

The combination of yield gap and poverty rates into a single map enables to identify best feasible modalities for agricultural development, potential investment, and resource allocation. - Yield gap provides important information for identifying causes of food insecurity and addressing rural poverty. Yield and production gaps have been estimated by comparing at a spatially detailed level of 5 arc-minutes the agro-ecological attainable yield and production of 22 major crops/crop groups, simulated under the historical climate of 1981-2010, with actual yields and production obtained by downscaling for the years 2000 and 2010 statistical data of main food, feed, and fiber crops. - The poverty map portrays poverty rates at the country level. Poverty rates vary from 12.7 to 76.2 percent. Accordingly, the higher percentage of poverty in a particular area, the poorer population is located there.

The “richness index” represents the level of economical wellbeing a country certain area in 2010. Regions with higher income per capita and low poverty rate and more access to market are wealthier and are therefore better able to prepare for and respond to adversity. The index results from the second cluster of the Principal Component Analysis preformed among 9 potential variables. The analysis identifies four dominant variables, namely “GDPppp per capita”, “agriculture share GDP per agriculture sector worker”, “poverty rate” and “market accessibility”, assigning weights of 0.33, 0.26, 0.25 and 0.16, respectively. Before to perform the analysis all variables were log transformed (except the “agriculture share GDP per agriculture sector worker”) to shorten the extreme variation and then were score-standardized (converted to distribution with average of 0 and standard deviation of 1; inverse method was applied for the “poverty rate” and “market accessibility”) in order to be comparable. The 0.5 arc-minute grid total GDPppp is based on the night time light satellite imagery of NOAA (see Ghosh, T., Powell, R., Elvidge, C. D., Baugh, K. E., Sutton, P. C., & Anderson, S. (2010).Shedding light on the global distribution of economic activity. The Open Geography Journal (3), 148-161) and adjusted to national total as recorded by International Monetary Fund for 2010. The “GDPppp per capita” was calculated dividing the total GDPppp by the population in each pixel. Further, a focal statistic ran to determine mean values within 10 km. This had a smoothing effect and represents some of the extended influence of intense economic activity for the local people. Country based data for “agriculture share GDP per agriculture sector worker” were calculated from GDPppp (data from International Monetary Fund) fraction from agriculture activity (measured by World Bank) divided by the number of worker in the agriculture sector (data from World Bank). The tabular data represents the average of the period 2008-2012 and were linked by country unit to the national boundaries shapefile (FAO/GAUL) and then converted into raster format (resolution 0.5 arc-minute). The first administrative level data for the “poverty rate” were estimated by NOAA for 2003 using nighttime lights satellite imagery. Tabular data were linked by first administrative unit to the first administrative boundaries shapefile (FAO/GAUL) and then converted into raster format (resolution 0.5 arc-minute). The 0.5 arc-minute grid “market accessibility” measures the travel distance in minutes to large cities (with population greater than 50,000 people). This dataset was developed by the European Commission and the World Bank to represent access to markets, schools, hospitals, etc.. The dataset capture the connectivity and the concentration of economic activity (in 2000). Markets may be important for a variety of reasons, including their abilities to spread risk and increase incomes. Markets are a means of linking people both spatially and over time. That is, they allow shocks (and risks) to be spread over wider areas. In particular, markets should make households less vulnerable to (localized) covariate shocks. This dataset has been produced in the framework of the “Climate change predictions in Sub-Saharan Africa: impacts and adaptations (ClimAfrica)” project, Work Package 4 (WP4). More information on ClimAfrica project is provided in the Supplemental Information section of this metadata.

The “element at risk index” describes the distribution in 2010 of the density of element (i.e. people, livestock unit or crop land) that are potentially threaded by climatic stress. When a climatic stress impact a densely populated (in terms of element at risk) the impact is likely to be more severe than it would be in areas with fewer element. The index results from the second cluster of the Principal Component Analysis (PCA) performed among potential 10 variables. The analysis identify three dominant variables, namely “people”, “livestock unit” and “crop land”, assigning a weight of 0.3 to people and weights of 0.35 to the other two variables. Before to perform the analysis the variables were log transformed to shorten the extreme variation and then score-standardized (converted to distribution with average of 0 and standard deviation of 1) in order to be comparable. The 0.5 arc-minute grid of population distribution in 2010 were gathered from Worldpop project website. The methodological approach used by Worldpop Project is described in Tatem et al 2007 (DOI: 10.1371/journal.pone.0001298). The values were adjusted in order to have national population totals equal to the UN Population and Demographic Office estimation per Country (World Population Prospect - the 2012 Revision). The 3 arc-minute grid of livestock distribution were gathered from FAO GeoNetwork (cattle, sheep, goats, pigs and poultry distribution), sampled at 0.5 arc-minute. The values were adjusted to national total livestock abundance in 2010 (FAO stats). The single species abundances were converted to livestock unit (LSU) according FAO methodology. The 0.5 arc-minute grid of crop land distribution were collected from FAO Global Land Cover-SHARE of 2014. The area with less than 1 people and 1 LSU for cell (about 1 Km square) and without cropland were masked and removed from the analysis for all layer because considered not exposed to climatic stress. This dataset has been produced in the framework of the “Climate change predictions in Sub-Saharan Africa: impacts and adaptations (ClimAfrica)” project, Work Package 4 (WP4). More information on ClimAfrica project is provided in the Supplemental Information section of this metadata.