Evaluation of nine sub-daily soil moisture model products over China using high-resolution in situ observations

Abstract

Reliable soil moisture (SM) estimates are not only crucial for the climate variability but are also significant for agricultural, ecological, climatic, and hydrometeorological predictions. With the development of land surface models (LSMs) and multi-source data merging techniques, various SM model products can serve as alternatives to obtain spatially continuous and temporally complete SM estimates. The nine SM model products (0–10 cm), including one analysis, three reanalyses, and five Land Data Assimilation System (LDAS) products, are evaluated using the dense (2437 stations after quality control) in situ observations at different timescales (annual to sub-daily) in eight climate regions over mainland China during 2010–2017. The Kling-Gupta efficiency (KGE), a performance metric combining correlation, bias, and variability, is used to quantify the strengths and weaknesses of these products. CLDAS (China Meteorological Administration’s LDAS) performs the best with the optimal median values of KGE and correlation in most subregions from sub-daily to annual timescales, followed by Global LDAS (GLDAS) CLM product. The SM products show large variability for different subregions, timescales, and spatial resolutions. The SM products have better performance in summer than winter. In addition, high-resolution products and LDAS products have the potential to improve the performance. For example, ERA5 outperforms ERA Interim in most parts of China. The markedly better performance of regional product CLDAS than other global products highlights the importance of incorporating hourly rain gauge and surface temperature observations at a finer model resolution. This intercomparison study sheds some light on the capability of the SM products for climate monitoring, weather prediction and model validation.