Automatic soil moisture data with a temporal resolution of 1 hr, a spatial resolution of 30 km, from the period June 1-September 1, 2017, and from Henan province in China, and simulation results from the Community Land Model (CLM4.0) were first compared, and a calibration study was further conducted and investigated. The operational status of the instruments was confirmed, and the data passed quality control. The corresponding simulation was performed using the CLM4.0. Small spatial scale variations in observed soil moisture contents were found, but relatively large spatial variations were observed in the simulated soil moisture contents. Since a relatively large bias and weak correlation were found between the simulation and observations, it is necessary to bias-correct the model outputs in order to improve the prediction capability of the numerical models. After bias correction was conducted, the outliers accounting for around 10% of the actual samples were identified and removed for each layer. The observed and simulated data were found to be linearly distributed, and the probability density distribution of deviations between observed and simulated values showed a normal distribution. The mean bias (BIAS) and root mean square error (RMSE) decreased by 100% and 50%, respectively, and the correlation co-efficient (CORR) increased by about 10 times. These improvements showed that the bias-corrected data met the requirements for the evaluation and application of soil moisture data, which can be of great significance in improving future land simulation and assimilation.
K E Y W O R D Sautomatic soil moisture observation, bias correction, bi-weight method, land-surface model
| INTRODUCTIONSoil moisture can affect the regional climate by changing the reflectance of the surface, the heat capacity of the ground, the sensible heat and latent heat which are Funding informationThe study was supported by the CMAÁHenan Key Laboratory of Agrometeorological Support and Applied Technique (AMF201904).