The coal seams of the Pinghu Formation in X area of Xihu Depression are widely developed, which are characterized by thin thickness, multiple layers, and low impedance. The strong amplitude presents during earthquakes masks the seismic response characteristics of sandstone and mudstone, thereby reducing the prediction accuracy of sandstone reservoirs. In response to the above issues, a pre-stack coal seam suppression method was proposed based on wavelet decomposition. Firstly, the pre-stack gathers were decomposed into a series of narrowband gathers with different center frequencies and Ricker wavelet meanings. Then, corresponding processing was implemented for the different coal seam response characteristics of the narrowband gathers. Among them, the low channel gathers had weak response in thin coal seams and did not require targeted coal compression processing. The seismic response characteristics of high channel coal seams were obvious. The strong amplitude attenuation method was used to suppress the strong energy of the coal seams. Finally, the processed signals of each frequency band were reconstructed to obtain the suppressed traces of the coal seams. Based on this, reservoir prediction was carried out, and the prediction accuracy was significantly improved, providing strong support for the exploration of lithological oil and gas reservoirs in coal bearing strata.