Qualitative satellite image analysis: mapping spatial distribution of farming types in Ethiopia.

摘要

Satellite images have enormous potential for qualitative land use analysis. This paper presents empirical results that demonstrate how normally invisible dimensions produced by land use can be identified by enriching satellite data with qualitative information from field studies. Land use can be defined as the intentional use of a specific piece of land resulting in patterns of ecological responses that are visible in the land cover and landscape. Responses to land use often result in a heterogeneous combination of reflectance in satellite images. Statistical methods used in the classification of satellite imagery are limited in their capacity to handle categories consisting of heterogeneous combinations of spectral values. To overcome this limitation, a contextual post-classification method has been used to map land cover configurations as related to different agricultural practices in the district of Sodo, Ethiopia. The results show that it is possible to map socio-spatial distribution of different agricultural and socioeconomic practices on a regional level by combining field observations and spatial contextual information. The empirical findings show local agricultural activity variations in cash crop production and subsistence agriculture in the Sodo district of Ethiopia.