The invention provides a soil
erosion evaluation and prediction method based on a public
data set, and belongs to the field of soil
erosion. An improved RUSLE model is constructed by obtaining
public land utilization data, rainfall data,
vegetation coverage data,
soil texture data and DEM data of a target area, and the soil
erosion condition of the target area is evaluated; four prediction scenes including a natural development scene, a city extension suppression scene, an ecological priority scene and a
cultivated land protection scene are put forward according to policies, adaptive atlas under the four scenes are made, existing
land utilization data and the adaptive atlas of all the scenes are input into a CA-Markov model, and
land utilization conditions under all the scenes in the future are obtained; and on the basis of the predicted
land utilization data, a water and
soil conservation measure factor under each scene is obtained in combination with an assignment method, and then an improved RUSLE model is constructed in the grid, so that the soil erosion condition of the target area under multiple scenes in the future can be obtained.