Abstract: Farmlands of sandy loam soil in the Daliuta mining in Shaanxi Province was chosen as object of the study. Soil compactness and moisture content was measured in situ for statistical analysis and linear fitting, and furthermore, for comparative analysis in combination with four semi-empirical models in simulation and prediction. Results show that soil at the early harvesting stage was the lowest in compactness, and the highest in soil moisture content; Planting of mung bean reduced soil compactness fairly; soil moisture content in the range of 8.30%-16.14% was linearly and negatively related to soil compactness; The linear model and the four semi-empirical models all fitted well in Area B (corn), with correlation coefficients being all higher than 0.7; Among the models, the Ayers model performed the best in prediction, which means that the model can be used to predict soil conditions of the semi-arid mining areas.