干旱区生态环境知识资源中心

The paper describes an attempt to relate patterns of vegetation cover with topography and a set of biological and grazing intensity variables in a mountain and piedmont area of arid central Australia. Vegetation cover, as measured by an index based on data from the Landsat satellite, can also be used as an erosion/deposition surrogate so the results have implications for distributed erosion models. A simple, analytically based erosion model derived from the continuity equation does not reproduce observed patterns of vegetation cover, and neither do various topographically based moisture indices. A regression approach shows that patterns of vegetation cover are related to topography but the most important predictors are biological ones, with percentage of bare ground upslope being the strongest. Tests with variable drainage area show that relationships between cover and topography, bare area upslope and grazing effects change systematically with basin size and that scale effects are present. Distributed erosion models are not yet capable of handling biological processes very well, yet these processes must be incorporated if erosion prediction is to be successful.