This paper reports a successful preliminary application of machine processing of LANDSAT data to the identification of swidden farming in East Africa. Analysis of "slash and burn" or shifting cultivation via LANDSAT requires recognition of the characteristics intrinsic to swiddening: by Western standards, field sizes are small, their borders are irregularly shaped and merge with natural features of the terrain; in the area of East Africa with which we are concerned, multiple cropping of as many as 25 cultigens is common, and the small fields in which these crop complexes are grown are interspersed among land at various stages of fallow and regeneration of plant cover.

These characteristics of swidden farming combine to achieve what Geertz has called a "canny imitation" of the natural landscape. This mimetic effect makes swidden fields indistinguishable from surrounding plant cover by visual inspection of standard LANDSAT imagery. In contrast, areas of Western-style agriculture are readily apparent. However, our analysis of the digital LANDSAT data does allow swidden areas to be differentiated.

The special problems which swidden farming pose require "a reorientation of techniques and typologies." Such a reorientation is justified for two reasons. First is the number of people for whom swidden farming is the life-support system: in 1963, Conklin put this number at 200 million persons in Africa, Asia, and the New World. Second, the homogeneous quality of the LANDSAT data, and the capability of making repeated observations, facilitate analysis of spatio-temporal variables intrinsic to swiddening.

Date of this Version