Investigating rural poverty and marginality in Burkina Faso using remote sensing-based products

• Published in: International journal of applied earth observation and geoinformation 2014-02, Vol.26, p.322-334
• Main Authors: Imran, M., Stein, A., Zurita-Milla, R.
• Format: Article
• Language: English
• Subjects: Applied geophysics; Composite asset index; Earth sciences; Earth, ocean, space; Exact sciences and technology; Food security; Geographical weighted regression; Internal geophysics; SPOT NDVI; TAMSAT rainfall
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2013.08.012

•We model agro-ecological marginality using remote sensing products.•We develop a new indicator to examine poverty and food-security spatial patterns.•We link the agro-ecological marginality and the indicator using local regression.•Interpreting local coefficients explains the poverty factors at fine resolutions.•The method generates timely and cost-effective poverty map at the national scale. Poverty at the national and sub-national level is commonly mapped on the basis of household surveys. Typical poverty metrics like the head count index are not able to identify its underlaying factors, particularly in rural economies based on subsistence agriculture. This paper relates agro-ecological marginality identified from regional and global datasets including remote sensing products like the normalized difference vegetation index (NDVI) and rainfall to rural agricultural production and food consumption in Burkina Faso. The objective is to analyze poverty patterns and to generate a fine resolution poverty map at the national scale. We compose a new indicator from a range of welfare indicators quantified from Georeferenced household surveys, indicating a spatially varying set of welfare and poverty states of rural communities. Next, a local spatial regression is used to relate each welfare and poverty state to the agro-ecological marginality. Our results show strong spatial dependency of welfare and poverty states over agro-ecological marginality in heterogeneous regions, indicating that environmental factors affect living conditions in rural communities. The agro-ecological stress and related marginality vary locally between rural communities within each region. About 58% variance in the welfare indicator is explained by the factors of rural agricultural production and 42% is explained by the factor of food consumption. We found that the spatially explicit approach based on multi-temporal remote sensing products effectively summarizes information on poverty and facilitates further interpretation of the newly developed welfare indicator. The proposed method was validated with poverty incidence obtained from national surveys.