Abstract

Discriminating cotton and cereals in Mali smallholder fields with WorldView2
Track: Agriculture
Authors: Pierre C. Traore, Madina Diancoumba

Recognizing crop species is critical for reliable agricultural statistics, and important for food security. In the rainfed cotton belt of West Africa, land allocation to cotton or cereals will impact i/ market prices, the food budget and income of households, and ii/ total biomass production by annual crops, soil fertility and the C cycle. However, in smallholder conditions, high agricultural landscape fragmentation and heterogeneous field conditions preclude crop identification unless very high resolution imagery is available. Other confounding factors include variability in management practices, such as staggering of planting dates. This paper reports on the use of WorldView2 spectral and contextual information to discriminate between cotton, millet, maize, and sorghum, using a single scene around the time of peak biomass in smallholder farming systems of West Africa. WorldView2's performance is compared with that of QuickBird, and its potential discussed in the larger context of emerging agricultural information systems for smallholders.