Mapping and Forecasting Soil Erosion Patterns from Landsat on a Microcomputer-based Image Processing Facility.

Abstract

The paper summarises recent work in the development and implementation of broad scale conceptual and mathematical models of the soil erosion process in flat arid lands. The conceptual model is based on erosion cell behaviour. Its mathematical counterpart is the simultaneous autoregressive random field model recently developed for image modelling.The data used in the mathematical model are derived by transforming standard Landsat MSS data to produce a soil stability index. Practical application of the methods requires the appropriate computer software and an image processing facility. The erosion modelling routines have therefore been implemented as part of a user- friendly microcomputer based image processing package which is now available commercially. The package runs on an IBM XT or AT computer using a Vectrix graphics board and the hardware includes high resolution display and hard copy facilities. Both hardware and software cost less than $30,000 making it ~ossible to decentralise image processing technology to branch offices of government departments or to agricultural consultants.