Generating seamless global daily AMSR2 soil moisture (SGD-SM) long-term products for the years 2013–2019

Abstract. High-quality and long-term soil moisture products are significant for hydrologic monitoring agricultural management. However, the acquired daily Advanced Microwave Scanning Radiometer 2 (AMSR2) incomplete in global land (just about 30 %–80 % coverage ratio), due to satellite orbit limitations of retrieval algorithms. To solve this inevitable problem, we develop a novel spatio-temporal partial convolutional neural network (CNN) AMSR2 product gap-filling. Through proposed framework, generate seamless (SGD) from 2013 2019. further validate effectiveness these products, three verification methods used as follows: (1) situ validation, (2) time-series (3) simulated missing-region validation. Results show that have reliable cooperativity with selected values. The evaluation indexes reconstructed (original) dataset correlation coefficient (R) 0.685 (0.689), root-mean-squared error (RMSE) 0.097 (0.093), mean absolute (MAE) 0.079 (0.077). temporal consistency is ensured original distribution valid spatial continuity regions accordance information (R: 0.963–0.974, RMSE: 0.065–0.073, MAE: 0.044–0.052). This can be downloaded at https://doi.org/10.5281/zenodo.4417458 (Zhang et al., 2021).