Toward a Better Understanding of Uncertainty for Satellite Precipitation Products over Complex Terrain with Sparse Rain Gauge Data

Yingying Gai,Xunjian Long,Zhihao Wei,Xinxin Sui,Bo Jiang,Xi Chen,Dingfeng Yu,Shengguang Qin,Yaokui Cui
DOI: https://doi.org/10.1016/j.atmosres.2023.106742
IF: 5.965
2023-01-01
Atmospheric Research
Abstract:Satellite precipitation product quality is critical for various meteorological and hydrological applications. Products are subjected to more uncertainties in sparsely-gauged regions than in densely-gauged regions due to insufficient error correction. However, ground-truth-based validation cannot accurately quantify the spatially distributed uncertainty of satellite precipitation products over complex terrain with sparse rain gauge data. In this study, we explored a gauge-independent method, called the Three-Cornered Hat (TCH), to overcome the quantification issue and better understand the uncertainty of satellite precipitation products in such situations. Firstly, the reliability of the TCH method was explored over a topographically complex area. Then, the performance of the Soil Moisture TO RAIN-Advanced SCATterometer (SM2RAIN-ASCAT) and Integrated Multi-Satellite Retrievals for Global Precipitation Measurement (GPM) mission (IMERG) in the Chuan-Yu region was evaluated using TCH, with the help of the China Merged Precipitation Analysis (CMPA) precipitation product as the third input. We found that: (1) The high consistency between error standard deviation (ESD) estimated by TCH and root mean square error (RMSE) estimated by ground-truth-based validation demonstrated that TCH is reliable for evaluating precipitation datasets over complex terrain; (2) The ESDs of SM2RAIN-ASCAT and IMERG had different spatial distributions, and the quality of IMERG was overall higher than that of SM2RAIN-ASCAT over a mountainous area; (3) IMERG performed better than SM2RAIN-ASCAT both at low (under 3000 m) and high elevations (over 3000 m) in most years, as well as for most main land cover types (woody savannas, mixed forests, and grasslands) in the study area. This method and these findings provide a new insight for better use of satellite precipitation products over complex terrain.
What problem does this paper attempt to address?