Spatial Error Distribution and Error Cause Analysis of TMPA-3B42V7 Satellite-Based Precipitation Products over Mainland China

Abstract

With a high spatial resolution and wide coverage, satellite-based precipitation products have compensated for the shortcomings of traditional measuring methods based on rain gauge stations, such as the sparse and uneven distribution of rain gauge stations. However, the accuracy of satellite precipitation products is not high enough in some areas, and the causes of their errors are complicated. In order to better calibrate and apply the product’s data, relevant research on this kind of product is required. Accordingly, this study investigated the spatial error distribution and spatial influence factors of the Tropical Rainfall Measuring Mission (TRMM) Multi-Satellite Precipitation Analysis (TMPA) post-process 3B42V7 (hereafter abbreviated as 3B42V7) data over mainland China. This study calculated accuracy indicators based on the 3B42V7 data and daily precipitation data from 797 rain gauge stations across mainland China over the time range of 1998–2012. Then, a clustering analysis was conducted based on the accuracy indicators. Moreover, the geographical detector (GD) was used to perform the error cause analysis of the 3B42V7. The main findings of this study are the following. (1) Within mainland China, the 3B42V7 data accuracy decreased gradually from the southeast coast to the northwest inland, and shows a similar distribution for precipitation. High values of systematic error (>1.0) is mainly concentrated in the southwest Tibetan Plateau, while high values of random error (>1.0) are mainly concentrated around the Tarim Basin. (2) Mainland China can be divided into three areas by the spectral clustering method. It is recommended that the 3B42V7 can be effectively used in Area I, while in Area III the product should be calibrated before use, and the product in Area II can be used after an applicability study. (3) The GD result shows that precipitation is the most important spatial factor among the seven factors influencing the spatial error distribution of the 3B42V7 data. The relationships between spatial factors are synergistic rather than individual when influencing the product’s accuracy.