Abstract: Seismic resolution is crucial to structure interpretation and reservoir characterization. To enhance seismic dominant frequency for thin-bed identification through high-resolution attribute analysis and inversion, we propose an approach of multi-filter combination for frequency expanding based on a new reflectivity series derived from poststack seismic data to establish multi-well broadband synthetic seismograms, which serve as the target to optimize the convolution of the filter bank and reflectivity series through stepwise regression analysis. Efforts focus on the way of multi-filter combination and weighting factor optimization to uncover and reconstruct high-frequency signals and extract seismic amplitude and frequency information within different frequency bands. Field applications to tight sandstone gas and coal-rock gas in the south of the east marginal Ordos Basin show significantly improved seismic data quality in higher resolution and better amplitude preservation. Processed data provide a reliable support for subsequent structure description and thin-reservoir prediction.