Abstract: Using the adaptive-threshold MODE method, four satellite precipitation products, FY-4B, GPM-Late, MSWEP, and PERSIANN-CCS, were evaluated for six heavy precipitation cases in the Sichuan region. The results show that: (1) FY-4B significantly overestimates precipitation intensity and area, with an excessive expansion in precipitation margins for large-scale precipitation events and a general southward shift. (2) GPM-Late consistently underestimates intensity and area, but demonstrates the best performance in centroid position. As the observed precipitation area increases, the centroid position bias gradually decreases. (3) MSWEP achieves the best overall performance with relatively stable error fluctuations, showing a systematic west-southward shift, while its intensity closely matches observations and centroid deviations are small. (4) PERSIANN-CCS has poor spatial localization ability, with large deviations in small-scale precipitation events. It generally overestimates intensity and area, with more pronounced overestimation in large-scale precipitation events. (5) FY-4B and PERSIANN-CCS tend to overestimate large-scale precipitation systems, reflecting the tendency of infrared-based retrieval methods to excessively expand the spatial extent of large precipitation systems.(6) GPM-Late shows severe underestimation of precipitation intensity, indicating that microwave signals are easily influenced by complex underlying surfaces.