Spectral Classification of Soil Degradation in Tropical Watersheds: Case Study in Kenya and Rwanda.
Athanase Mukuralinda1, Thine Omuto2, Louis Verchout3, and keith shepherd3. (1) ISAR, B.P 138, Butare, Butare, 250, Rwanda, (2) University of Nairobi, Upper Kabete, Box 30197, Nairobi, Kenya, (3) World Agroforestry Center, Box 30677, 0100 Unep avenue, Gigiri, Nairobi, Kenya
In the tropical watersheds, soil degradation is a major constraint limiting agricultural productivity and eventual contribution to food insecurity. Particularly in Eastern and Central Africa, the increasing soil degradation is evident in the rising decline in agricultural productivity even to areas that were originally perceived as high potential. The current conventional approach of assessing soil productive decline is however limited in both cost and technological capacity to characterize large areas in a way that can permit comparison over time. This study tested the application of infrared spectroscopy in rapid analysis and classification of numerous soil samples according to their degradation status. The soils, which were drawn from multiple points in two watersheds in eastern Kenya and southern Rwanda, were tested for their physical constraints (using hydrologic properties) and chemical constraints (using fertility indicators). Point-measurement of the soil properties were calibrated with soil spectral reflectance for rapid analysis and prediction of other samples in the watersheds. A soil quality index was developed as a distance measure in spectral data space of soil spectra compared to sediment spectra. This index displayed distinctly different soil chemical and physical properties and was used to develop a classification model for the prediction of future samples into their degradation classes. The soil quality index developed can provide a basis for rapid and cost –effective identification of land degradation hot-spots and as early warning indicators to assist land managers in targeting management interventions.