Rapidly changing socioeconomic conditions can have important consequences for environmental resources and ecosystem services. In developing countries, rural communities rely on natural resources and environmental products for food, fuel, building materials and medicines. Therefore, understanding poverty and ecosystem services relationships is important for sustainable human development and ecological conservation. There has been increasing interest in the use of remotely sensed satellite data for monitoring and estimating human poverty and ecosystem services to bridge data gaps and improve our understanding of the dynamics of socio-ecological systems. In this multi-disciplinary project we explored two research questions; (1) How can remotely sensed satellite data be used to estimate aspects of environmental resources and ecosystem provisioning services important for livelihoods, and; (2) what are the relationships between household poverty/livelihoods and remotely sensed environmental conditions? To answer these questions we used object based image analysis to classify land use in fine resolution (<1m spatial resolution) satellite data in Sauri, Kenya. To estimate environmental resources available to local households we developed a multi-resolution approach which accounts for the daily interactions between people and the local environment. We used this approach to link the environmental data from land use classifications to household panel data. To establish the relationships between poverty and environment we used classification and regression trees (CART) to compare household socioeconomic conditions (and asset poverty) with the availability of local environmental resources and ecosystem services.
Results/Conclusions:
The use of a single polygon, such as a radial buffer zone, to link people and pixels is inappropriate in coupled human-natural systems as processes and interactions between people and the environment operate over multiple scales. Interactions in these systems occur at multiple levels from individual (household) access to agricultural plots to communal access to common pool resources (forestry, pasture) at the village level. Results indicate that the relationships between household poverty and livelihoods are complex. We find that high resolution satellite data explains about 50%-60% of the variation of household poverty in the lowest poverty quintile. Measures of the proportion of building coverage within a homestead (financial capital); amount of communal woodland resources (natural capital) within the village/cluster boundary and distance to main road were strong correlates with poverty. Conclusions are that the multi-scale approach to linking households with remote sensing data could enhance the modelling of human-environment interactions and help improve monitoring of the sustainable development goals for human livelihoods and biodiversity conservation.