Evaluation of satellite-based and reanalysis soil moisture products using ground-observations in the Qilian Mountains

Abstract

Critical to the reliable application of gridded soil moisture (SM) products is a thorough assessment of their quality, particularly in water conservation and supply areas such as Qilian Mountains (QLM). The study systematically evaluated the accuracy of surface soil moisture (SSM) and root-zone soil moisture (RZSM) using 14 ground-based sites in the QLM from 2019 to 2021. The study evaluated seven SSM products, comprising four satellite-based datasets (AMSR2-LPRM, SMOS-IC, SMAP-L3, and ESA CCI) and three reanalysis datasets (SMAP-L4, GLDAS-Noah, and ERA5-Land). Furthermore, the performance of three RZSM products derived from these reanalysis datasets was also examined. For SSM products, reanalysis products generally exhibit higher R (0.56) and lower ubRMSE (0.04 m3/m3) compared to satellite-based products (0.44, 0.05 m3/m3). The SMAP-L3 showed the best performance (R = 0.71), followed by the ERA5-Land and SMAP-L4. For RZSM products, compared to GLDAS and ERA5-Land, SMAP-L4 demonstrated better performance (R = 0.59) and shown greater potential for application in QLM. The average R (0.65) of SSM was highest for grassland sites, attributed to fewer land cover heterogeneity. For forest sites, the SSM performance was generally lowest (average R = −0.11). In Arou site, the superior spatial representativeness of Cosmic ray neutron sensor (CRNS) measurements enhanced the evaluation performance (higher R, lower ubRMSE) of nearly all products. Notably, the comparative evaluation with in-situ data underscored the robustness of CRNS as a reference for SM evaluation and proved its mesoscale applicability. The research aimed to provide a reference for the application of gridded SM products in water resources monitoring and ecosystem services in QLM.