Rabiei et al. (2026) Deep learning-based downscaling of SMAP data for surface and root-zone soil moisture mapping

Identification

• Journal: Journal of Hydrology
• Year: 2026
• Date: 2026-06-25
• Authors: Saman Rabiei, Ebrahim Babaeian, Sabine Grunwald
• DOI: 10.1016/j.jhydrol.2026.135901

Research Groups

• Department of Soil, Water, and Ecosystem Sciences, University of Florida

Short Summary

This study presents a deep learning downscaling framework that improves the accuracy of surface and root-zone soil moisture mapping using SMAP data at 100 m resolution.

Objective

• Investigate the feasibility of using deep learning techniques to downscale SMAP data for high-resolution soil moisture mapping.

Study Configuration

• Spatial Scale: Local to regional scale (sub-kilometer resolution)
• Temporal Scale: Daily

Methodology and Data

• Models used: Convolutional Neural Networks (CNN) and Long Short-Term Memory Networks (LSTM)
• Data sources: SMAP data, MODIS data, elevation, land cover, and Soil Landscapes of the U.S. (SOLUS100) databases

Main Results

• The CNN-LSTM model improved surface soil moisture accuracy with a median correlation coefficient (R) increasing from 0.61 to 0.82 and unbiased root mean square error (ubRMSE) decreasing from 0.06 to 0.05 cm3 cm−3.
• Root-zone soil moisture achieved R ~ 0.72, ubRMSE ~ 0.05 cm3 cm−3.

Contributions

• The study bridges the spatial resolution gap in SM products and supports small-scale agricultural and hydrological applications.

Funding

• This research was funded by [project/program name] (reference code).

Citation

@article{Rabiei2026Deep,
  author = {Rabiei, Saman and Babaeian, Ebrahim and Grunwald, Sabine},
  title = {Deep learning-based downscaling of SMAP data for surface and root-zone soil moisture mapping},
  journal = {Journal of Hydrology},
  year = {2026},
  doi = {10.1016/j.jhydrol.2026.135901},
  url = {https://doi.org/10.1016/j.jhydrol.2026.135901}
}