Based on in-situ dekad soil relative moisture observations of automatic weather stations nationwide from 2008 to 2017, this study makes a comprehensive applicability evaluation of CLDAS (CMA Land Data Assimilation System) merged soil relative moisture product at 0-20 cm layer. The evaluation shows CLDAS soil relative humidity product has large systematic error compared with in-situ observations in Northeast China, Northwest China, most areas of Jiangnan, and South China, so it generally has poor applicability. To eliminate the systematic error of CLDAS soil relative humidity product, the regression correction method, seven ten-day moving average correction method, and near-weighted last ten-day correction method are used to conduct error correction. The evaluation of the correction result shows the correction of CLDAS soil relative humidity product makes a significant increase of its correlation with in-situ observations, the systematic error is basically eliminated, and its applicability has significantly improved. Among the above-mentioned three correction methods, the near-weighted last ten-day correction method performs the best. Finally, the corrected CLDAS soil relative humidity products with the three methods are used to reproduce one meteorological drought case that happened in Northeast–North China in May 2017. The verification results show that compared with the other two error correction methods, the CLDAS soil relative humidity product processed by the near-weighted last ten-day correction method can more accurately reproduce the area and intensity of the meteorological drought in Northeast–North China in May 2017.