Abstract: Single-source precipitation observation data typically exhibit considerable uncertainty, posing challenges for meeting the requirements of high-precision and high-resolution precipitation data in practical applications of the new era. To tackle this issue, this paper proposes a multi-source collaborative inversion method that integrates microwave link signals, satellite data, and ground-based rain gauge observations. By adopting a multi-grid variational analysis approach, the method fuses precipitation data from automatic stations, microwave links, and IMERG satellite inversions at both hourly and daily scales, thereby constructing precipitation fields that are more consistent with actual conditions. Results show that, in comparison with satellite products, the hourly and daily precipitation fields reconstructed via three-source fusion exhibit increases in the correlation coefficient by 0.23 and 0.13, respectively, while their relative errors are reduced by 22.8 and 0.41. This confirms that the proposed method can effectively enhance the original low-resolution satellite precipitation data, rendering it more representative of real-world conditions. Analysis of visualization results reveals that the precipitation field constructed through three-source data fusion possesses superior capability in capturing extreme precipitation events, with the capture rate for extreme rainfall intensity thresholds rising to 78.3%. Furthermore, precipitation events across adjacent time intervals demonstrate improved continuity. The proposed method effectively alleviates errors in satellite products during severe convective weather, providing a new approach for the accurate acquisition of precipitation resources in river basins.