SGD-SM 2.0: an improved seamless global daily soil moisture long-term dataset from 2002 to 2022

Abstract. The drawbacks of low-coverage rate in global land inevitably exist satellite-based daily soil moisture products because the satellite orbit covering scopes and limitations retrieving models. To solve this issue, Zhang et al. (2021a) generated seamless (SGD-SM 1.0) for years 2013–2019. Nevertheless, there are still several shortages SGD-SM 1.0 products, especially temporal range, sudden extreme weather conditions sequential time-series information. In work, we develop an improved 2.0) dataset 2002–2022, to overcome above-mentioned shortages. 2.0 uses three sensors, i.e. AMSR-E, AMSR2 WindSat. Global precipitation fused into proposed reconstructing model. We propose integrated long short-term memory convolutional neural network (LSTM-CNN) fill gaps missing regions products. situ validation testify accuracy availability (R: 0.672, RMSE: 0.096, MAE: 0.078). curves consistent with original distribution. Compared 1.0, outperforms on consistency. recorded https://doi.org/10.5281/zenodo.6041561 (Zhang al., 2022).