Abstract: In order to provide the basic data of lithology modeling with high resolution in the early stage of coal precise mining geological guarantee, this paper established the p-wave impedance model of wedge-shaped coal seam according to the thickness and storage conditions of coal measure strata based on geostatistical inversion theory. The influence of different probability density functions and transverse range parameters on thickness inversion of coal seam are analyzed through multiple inversion experiments. The effect of inversion under different amount and location constraints is further discussed. Experimental results show that geostatistical inversion parameters have a great influence on the inversion results. Reasonable probability density function, lateral range and constrained well selection can improve the accuracy of inversion. Results shows that geostatistical inversion can predict about 1 m thin coal seam, and the error range between the inversion coal seam thickness and the logging data is 1.82 % ~12.24 %. It is therefore feasible to predict the thickness of thin coal seam, and can provide valuable modeling data for the early lithologic geological modeling of "transparent mine".