SGD-SM: Generating Seamless Global Daily AMSR2 Soil MoistureLong-term Productions (2013–2019)

Abstract. High quality and long-term soil moisture productions are significant for hydrologic monitoring and agricultural management. However, the acquired daily soil moisture productions are incomplete in global land (just about 30 %∼80 % coverage ratio), due to the satellite orbit coverage and the limitations of soil moisture retrieving algorithms. To solve this inevitable problem, we develop a novel 3D spatio-temporal partial convolutional neural network (CNN) for Advanced Microwave Scanning Radiometer 2 (AMSR2) soil moisture productions gap-filling. Through the proposed framework, we generate the seamless global daily (SGD) AMSR2 soil moisture long-term productions from 2013 to 2019. To further validate the effectiveness of these productions, three verification ways are employed as follow: 1) In-situ validation; 2) Time-series validation; And 3) simulated missing regions validation. Results show that the seamless global daily soil moisture productions have reliable cooperativity with the selected in-situ values. The evaluation indexes of the reconstructed (original) dataset are R: 0.683 (0.687), RMSE: 0.099 m3/m3 (0.095 m3/m3), and MAE: 0.081 m3/m3 (0.078 m3/m3), respectively. Temporal consistency of the reconstructed daily soil moisture productions is ensured with the original time-series distribution of valid values. Besides, the spatial continuity of the reconstructed regions is also accorded with the context information (R: 0.963∼0.974, RMSE: 0.065∼0.073 m3/m3, and MAE: 0.044∼0.052 m3/m3). More details of this work are released at https://qzhang95.github.io/Projects/Global-Daily-Seamless-AMSR2/ . This dataset can be downloaded at https://zenodo.org/record/3960425 (Zhang et al., 2020. DOI: https://doi.org/10.5281/zenodo.3960425 ).