Ensemble deep learning towards high-resolution soil-moisture mapping for enhanced water management in California's Central Valley

Authors: AZEDOU, ALI - Universite Ibn Tofail; AMINE, AOUATIF - Universite Ibn Tofail; LAHSSINI, SAID - National School Of Forestry Engineering; Osterman, Gordon; ARBOLEDA-ZAPATA, MAURICIO - University Of California, Davis; Cosh, Michael; KISEKKA, ISAYA - University Of California, Davis

Submitted to: Environmental Modelling & Software
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/4/2025
Publication Date: 12/5/2025
Citation: Azedou, A., Amine, A., Lahssini, S., Osterman, G.K., Arboleda-Zapata, M., Cosh, M.H., Kisekka, I. 2025. Ensemble deep learning towards high-resolution soil-moisture mapping for enhanced water management in California's Central Valley. Environmental Modelling & Software. 197. Article 106824. https://doi.org/10.1016/j.envsoft.2025.106824.
DOI: https://doi.org/10.1016/j.envsoft.2025.106824

Interpretive Summary: Understanding the spatial distribution of soil moisture in the root zone is important for understanding hydrological and agricultural processes, but is difficult to map with high accuracy, high spatial and temporal resolution, and extensive spatial coverage at the same time. Satellite remote sensing data can provide extensive spatial coverage, but lack the resolution or accuracy, while point measurements can be very accurate and provide continuous monitoring, but often sacrifice spatial coverage and resolution. In this study, we explore the use of deep ensemble learning frameworks to downscale the satellite data from the kilometer-scale to the scale of approximately 30 meters. By synthesizing several remote sensing products, we can select the most predictive variables to train a deep neural network and validate the model with point measurements of soil moisture. We find that we can reproduce the temporal dynamics of soil moisture at the desired resolution. These results have implications for the use of remote sensing products to understand processes dependent on soil moisture, such as evapotranspiration, which are important for achieving sustainable water use in agricultural settings.

Technical Abstract: Soil moisture (SM) plays a vital role in both hydrological and agricultural processes and is critical for achieving groundwater sustainability in agriculture through demand management. NASA's Soil Moisture Active Passive (SMAP) satellite measures the SM across the Earth and provides data on both the surface and root zone SM but at a coarse spatial resolution of 9 km, thereby limiting detailed analyses. This study aimed to develop an optimized deep ensemble learning framework to downscale the resolution of SMAP observations of California's Central Valley from 9 km to 30 m for both the surface and root-zone SM. Sensitivity analysis was employed to identify key explanatory variables. The models were then combined into an ensemble DNN trained on multiscale SMAP data and validated against in-situ SM measurements. The results demonstrated that the ensemble model achieved the highest correlations of 0.789 and 0.683 for surface and root-zone SM, respectively, with the lowest root mean square errors of 0.0281 and 0.0814 cm3/cm3, respectively, thereby reliably reproducing the temporal dynamics. Seasonal analysis revealed distinct patterns linked to climate and management practices at a spatial resolution of 30 m, thereby capturing seasonal variations in soil moisture among major crops. Additionally, SM maps can be used to refine the estimated evapotranspiration due to groundwater pumping, which can be used to monitor water allocation. SM can also be used to inform agronomic practices, such as delayed irrigation in early spring, which can reduce groundwater demand.