基于多源数据融合的海河流域降水资源评价.

To increase the robustness of the estimation of Haihe River basin precipitation resources, we divided the basin into three regions: windward mountainside, leeward mountainside, and plains. Assistant by the artificial neural network machine learning and consistency correction principle, we constructed a comprehensive multisource fusion dataset of Haihe River basin precipitation from 2001 to 2019. The results showed that the original satellite precipitation products overestimated precipitation in the basin. The fusion dataset present that the average precipitation of the Haihe River basin from 2001 to 2019 was 515.2 mm, and the rainfall resources were 163.94 billion m~3,respectively. The evaluation parameter of the fusion dataset represented a significant improvement in the accuracy of precipitation estimation. The fusion dataset also better explored the spatial distribution of precipitation in the basin and showed that more rainfall was collected in the northeastern, southeastern, southwestern, and west-central regions of the basin, whereas less was collected in the northwestern and east-central areas. Precipitation over the plains regions of the basin showed a distinct spatial pattern, and precipitation on both windward and leeward mountainsides was related to elevation. [ABSTRACT FROM AUTHOR]