MSM4D: a sub-daily soil moisture dataset with four nominal overpass slots from AMSR2-SMAP fusion (2015–2024)

Abstract. Soil moisture (SM) is an essential variable for hydrological processes, agricultural monitoring, drought and flood assessment, and land–atmosphere interactions. Existing passive microwave SM products provide global observations but are generally limited to daily or twice-daily sampling, leaving rapid intraday wetting and drying processes insufficiently sampled. In this study, we developed the “Multi-source Sub-daily Soil Moisture dataset with four daily nominal overpass slots (MSM4D)”, a 36 km sub-daily SM dataset that integrates Advanced Microwave Scanning Radiometer 2 (AMSR2)-derived estimates and Soil Moisture Active Passive (SMAP) observations from 2015 to 2024. MSM4D provides four nominal overpass slots, including T_AMSR2-D, T_SMAP-D, T_AMSR2-A, and T_SMAP-A, together with a daily-mean product on the Equal-Area Scalable Earth Grid 2.0. A ten-day-scale training strategy establishes grid-wise relationships between AMSR2 microwave observations and SMAP Level-3 (L3) SM retrievals, enabling SMAP-consistent SM estimates for the two AMSR2-derived temporal slots. By adding these temporally complementary estimates to the two SMAP-based slots, the global mean proportion of days with at least two valid observations increased from 21.31 % for SMAP alone to 71.78 % for MSM4D. Across the core validation grids, MSM4D achieved an overall unbiased root mean square error (ubRMSE) of 0.048 cm³ cm⁻³, with the AMSR2-derived slots showing accuracy comparable to the SMAP-based slots. Validation across additional sites further supported the robustness of the dataset over diverse land-surface conditions. Event-based analyses further showed that the enhanced temporal sampling provides additional information on intraday wetting and drydown processes that may be obscured by daily aggregation, smoothed in model-based products such as SMAP Level-4 (L4), or insufficiently sampled by twice-daily observations. MSM4D therefore provides a decade-long, observation-constrained data record for investigating sub-daily SM variability and associated hydrological processes. The dataset is available at https://doi.org/10.6084/m9.figshare.34700667 (Song and Wang, 2026).