Abstract: To identify hydrocarbon-bearing sandstones under the influence of thin coal seams, an integrated study using drilling and seismic data was conducted to perform a progressive reservoir prediction workflow, driven by geological understanding, which includes: identification of sand-rich zones → delineation of composite sand bodies → characterization of individual sand bodies. First, based on the slope-trough composite structural background of the study area, the slope-break controlled sand deposition model was summarized. Leveraging the advantages of structural seismic attributes such as gradient tensor, dip, and curvature, fine identification of flexural slope breaks was carried out. The overlap of slope-break zones and low-potential geomorphic areas indicated favorable sand-rich zones. Subsequently, within these sand-rich zones, pre-stack seismic data with reduced coal influence were obtained through a method involving "seismic-scale coal seam identification, filtering, and coal removal." Pre-stack inversion based on this processed data effectively eliminated coal-induced artifacts in the inversion results, achieving accurate delineation of composite sand bodies. Finally, constrained by the composite sand bodies, the advantages of multi-attribute well and seismic data were utilized to delineate the boundaries of individual sand bodies within the composite sand bodies. A research strategy of "inversion for sand identification, instantaneous attributes (frequency, phase, amplitude) for boundary definition" was proposed, enabling detailed characterization of individual sand bodies. This method effectively addresses the challenges of poor reliability and low resolution in reservoir prediction within coal-bearing strata of the Xihu Sag, improving prediction accuracy by 23%. It holds significant importance for the progressive exploration and potential tapping of structural-lithological reservoirs in the study area.