Shafiei et al. (2026) Downscaling SMAP Soil Moisture to 1 km with Machine Learning and MODIS Data for Agricultural Drought Assessment in Békés County, Hungary

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: AgriEngineering
• Year: 2026
• Date: 2026-09-04
• Authors: Mahrokh Shafiei, István Waltner, Z. Vekerdy, Gábor Halupka
• DOI: 10.3390/agriengineering8090373

Research Groups

Not specified

Short Summary

This study compares three machine learning frameworks to downscale SMAP soil moisture data from 9 km to 1 km in Békés County, Hungary, concluding that Random Forest provides the most accurate high-resolution mapping.

Objective

• To improve the spatial resolution of SMAP soil moisture (SM) retrievals from 9 km to 1 km using machine learning to enhance agricultural drought monitoring at regional and local scales.

Study Configuration

• Spatial Scale: Békés County, Hungary (Regional/Local scale; target resolution 1 km)
• Temporal Scale: Growing seasons (April to October) from 2020 to 2023

Methodology and Data

• Models used: Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Gradient Boosting Machine (GBM)
• Data sources: SMAP (passive microwave retrievals), MODIS-derived variables (NDVI, EVI, daytime and night-time land surface temperature, and evapotranspiration), land cover classification, and topographic elevation

Main Results

• The Random Forest (RF) model demonstrated the highest accuracy during testing and validation ($R^2 = 0.71$, $\text{RMSE} = 0.0295\text{ m}^3/\text{m}^3$).
• Daytime land surface temperature (LST) was identified as the most significant predictor across all models.
• The 1 km Standardized Soil Moisture Index (SSI) maps generated via the RF model successfully captured inter-annual variability, specifically identifying the severe drought of July 2022.

Contributions

• Provides a validated downscaling approach for generating high-resolution soil moisture data specifically suited for Central European agricultural environments.
• Offers a practical tool for decision-makers to optimize irrigation scheduling and reduce agricultural losses during drought periods.

Funding

Not specified

Citation

@article{Shafiei2026Downscaling,
  author = {Shafiei, Mahrokh and Waltner, István and Vekerdy, Z. and Halupka, Gábor},
  title = {Downscaling SMAP Soil Moisture to 1 km with Machine Learning and MODIS Data for Agricultural Drought Assessment in Békés County, Hungary},
  journal = {AgriEngineering},
  year = {2026},
  doi = {10.3390/agriengineering8090373},
  url = {https://doi.org/10.3390/agriengineering8090373}
}