不同雨量站点密度条件下多源降水融合精度评估OA

In order to systematically evaluate the influence of rainfall station density on multi-source precipitation merging accuracy,the Huanren Basin is used as a case study,where eight station-density schemes are established,the random forest(RF)algorithm is applied to merge the near-real-time IMERG and GSMaP product series,and the resulting merged outputs are comprehensively evaluated in terms of statistical correlation,error characteristics,and event-detection capability.The results show that:(a)the RF-based merged products consistently outperform individual satellite products across all density scenarios,even under very sparse observational networks,and the merged estimates exhibit clear improvements in correlation,root-mean-square error,mean absolute error and detection rate compared with the best single-source product;and(b)the merging accuracy improves nonlinearly as the station density decreases,with a critical threshold identified at approximately 1 700 km2/station,and when the station density is greater than this threshold,the accuracy improvement is significant,whereas below this threshold the gains become limited.