Multi-Source Daily Precipitation Merging over the Yangtze River Basin Using Triple Collocation

The Yangtze River Basin (YRB) is highly vulnerable to both floods and droughts, with precipitation playing a key role in driving these hydrological extremes. Understanding and reducing uncertainty in precipitation estimates is therefore crucial for effective water management and hazard mitigation. The study evaluated the error characteristics of daily precipitation estimates from three datasets (CRA40, IMERG, and SM2RAIN) using the triple collocation (TC) approach. A least-squares merging framework was then applied to integrate these datasets and generate merged precipitation estimates with improved accuracy and reduced uncertainty over the YRB. Results showed that all examined datasets exhibited higher fractional root-mean-squared error (fRMSE) in the source region of the Yangtze River, indicating a greater influence of random errors and reduced sensitivity to precipitation changes in this area. Among the datasets, SM2RAIN exhibited the weakest agreement with ground-based measurements, while IMERG performed best in capturing extreme precipitation events. CRA40 and the TC-based merged precipitation estimates exhibited overall higher accuracy, with a station-average correlation coefficient of approximately 0.71. Despite comparable accuracy, the merged precipitation data is relatively more robust than CRA40, with a lower average error standard deviation of 2.07 mm.