Development and validation of risk profiles of West African rural communities facing multiple natural hazards

West Africa has been described as a hotspot of climate change. The reliance on rain-fed agriculture by over 65% of the population means that vulnerability to climatic hazards such as droughts, rainstorms and floods will continue. Yet, the vulnerability and risk levels faced by different rural social-ecological systems (SES) affected by multiple hazards are poorly understood. To fill this gap, this study quantifies risk and vulnerability of rural communities to drought and floods. Risk is assessed using an indicator-based approach. A stepwise methodology is followed that combines participatory approaches with statistical, remote sensing and Geographic Information System techniques to develop community level vulnerability indices in three watersheds (Dano, Burkina Faso; Dassari, Benin; Vea, Ghana). The results show varying levels of risk profiles across the three watersheds. Statistically significant high levels of mean risk in the Dano area of Burkina Faso are found whilst communities in the Dassari area of Benin show low mean risk. The high risk in the Dano area results from, among other factors, underlying high exposure to droughts and rainstorms, longer dry season duration, low caloric intake per capita, and poor local institutions. The study introduces the concept of community impact score (CIS) to validate the indicator-based risk and vulnerability modelling. The CIS measures the cumulative impact of the occurrence of multiple hazards over five years. 65.3% of the variance in observed impact of hazards/CIS was explained by the risk models and communities with high simulated disaster risk generally follow areas with high observed disaster impacts. Results from this study will help disaster managers to better understand disaster risk and develop appropriate, inclusive and well integrated mitigation and adaptation plans at the local level. It fulfills the increasing need to balance global/regional assessments with community level assessments where major decisions against risk are actually taken and implemented.

S3 Text. Results of the main components of vulnerability and risk. S3.1.Exposure of the rural communities to the multiple hazards Exposure to hazards is an important dimension of the overall risks faced by a system or community. The implementation of an SES approach means that the exposure index represents both the exposure of the environmental sub-system to droughts and floods as well as the exposure of the social subsystem. In Table A below, the exposure of all the community clusters studied in the three countries have been presented. In the Vea study area (Ghana), the Kula river community cluster is the most exposed community, followed by communities in the Vea main drain and Valley zone in that order. Communities in the Kanga has the least exposure with an index value of just 0.134. Similarly, in the Dano study area (Burkina Faso), communities in the Loffing-Yabogane cluster are the most exposed to the multiple hazards followed by those in Batiara, Bolembar and Gnipkiere in that order. In this study area, Meba Pari has the lowest exposure index of 0.225. Also in the Dassari study area (Benin), Porga cluster of communities are the most exposed followed by Tankouri and Sechendiga clusters. The results show the mean exposure index is highest for communities in the Dano study area (0.45), that for Vea is 0.30 and Dassari communities has mean exposure index of 0.19. Exposure of communities in Dano is also more variable within communities. The variability is estimated at 0.14 around the mean in Dano and 0.12 in Vea. The higher variability of the exposure index in Dano means significant differences exist between the communities in terms of exposure.
It is interesting to note that exposure of communities followed the same pattern of the Flood Hazard Index maps developed by Asare-Kyei et al. (2015b) where the distribution of flood hazard in the study areas was modelled. In their study, the Kula River and Vea main drain in Vea; Porga in Dassari and Loffing-Yabogone in Dano were reported to be falling in high flood intensity zones. This study reinforces this finding and shows that the exposure index followed the pattern of flood hazard intensity zones. Although, there are other determinants of exposure as can be seen in the indicators used to construct the index, this fact shows the strong effect proximity to high flood intensity zones has on the overall SES exposure to the hazards. Another major driving factor influencing community exposure to multiple occurrences of drought and flood is the indicator measuring the share of the population engaged in agriculture. This indicator measures populations whose livelihood depends solely on agriculture and which have no other income or food sources. As expected, 72% of people in the Dano area belong to this category of 'Agricultural Dependent Population' (ADP) , 42% in Dassari and Vea having the least number of people (35%) engaged in only agriculture. Although this indicator was ranked second in Dano and first in both Vea and Dassari (Asare-Kyei et al., 2015a), its effect on exposure is still significant.

S.3.2. Susceptibility of the communities to drought and floods
Susceptibility is measured as inherent conditions within the communities that predispose them to be adversely affected by the two hazards. The SES approach measures susceptibility for both the socioeconomic and environmental sub-systems. Within the social-economic sub-system, four dimensions comprising 'poverty and dependencies', housing, public infrastructure and health and nutrition are considered. Table B below provides details about the susceptibility indices of the communities.
In Table B, the three most susceptible community clusters have been highlighted in grey. Interestingly, all the highly susceptible communities in the Vea area are in the Bongo district. In this study area, Tarongo has the highest susceptibility of 0.693 and Kolgo-Anateem clusters having the least susceptibility. Susceptibility indices in the Dassari area are generally high with a mean of 0.44. Communities in Dano area has a mean of 0.37. However, there are sharp differences in susceptibility indices in the Vea area measured by the standard deviation of 0.12 while communities in the Dassari area record less variability (0.6) from each other.

S3.3. Lack of capacity index
Community lack of capacity to cope and adapt to the hazards occurrence is an integral part of the overall vulnerability of the community. Total lack of capacity in this study has been computed from three sub-indices, lack of coping capacity, lack of ecosystem vitality and lack of adaptive capacity to respond to long-term hazards. In terms of capacity, there is no significant difference between the three study areas with mean lack of capacity. All are greater than 50% with minimal differences in variability. Lacks of coping and adaptive capacities are major contributors to the total lack of capacity.