Bouizrou et al. (2025) The potential of novel remote sensing evapotranspiration data and global soil maps for SWAT+ agro-hydrological modeling in data-scarce regions of the North Mediterranean

Identification

• Journal: Agricultural Water Management
• Year: 2025
• Date: 2025-08-29
• Authors: Ismail Bouizrou, Giulio Castelli, Gonzalo Cabrera, Lorenzo Villani, Stavros Solomos, Giorgos Maneas, Christos Pantazis, Elena Bresci
• DOI: 10.1016/j.agwat.2025.109761

Research Groups

• Institute for Ecology and Landscape, Hochschule Weihenstephan-Triesdorf University of Applied Sciences, Germany
• Department of Agriculture, Food, Environment and Forestry (DAGRI), University of Florence, Italy
• Environmental Governance and Territorial Development Hub (GEDT), University of Geneva, Switzerland
• UNESCO Chair in Hydropolitics, University of Geneva, Switzerland
• Research Centre of Atmospheric Physics and Climatology, Academy of Athens, Greece
• Navarino Environmental Observatory (N.E.O.), Messenia, Greece

Short Summary

This study utilized the SWAT+ model with novel remote sensing evapotranspiration (RS-ET) data, a high-resolution global soil map (DSOLMap), and detailed agricultural practices for agro-hydrological modeling and multisite calibration in four ungauged watersheds in the data-scarce Messinia region, Greece. The findings demonstrate that integrating DSOLMap and GLEAM RS-ET significantly improved model performance (Nash-Sutcliffe Efficiency > 0.5; Percent Bias < ±15 %) compared to local soil maps and MODIS RS-ET, providing a valuable tool for water resource management.

Objective

• To explore the potential of multiple remote sensing evapotranspiration (RS-ET) products for SWAT+ model evaluation using a multisite calibration approach.
• To evaluate the performance of a recent high-resolution global soil database (DSOLMap) in constraining hydrological models compared to data from a local soil map.
• To investigate modifying agricultural land use using crop distribution statistics and integrating management details into SWAT+ to enhance model performance and improve agricultural water management in data-scarce regions.

Study Configuration

• Spatial Scale: Messinia region, southwestern Peloponnese, Greece. Four ungauged watersheds (Sellas, Xerolagados, Gianouzagas, Xerias) covering a total area of 210 square kilometers. Individual watershed areas range from 25 square kilometers to 90 square kilometers. Input data resolutions: Digital Elevation Model (DEM) at 5 meters, Land Use at 100 meters, DSOLMap at 250 meters, MODIS AET at 500 meters, and GLEAM AET at 25000 meters (0.25 degrees).
• Temporal Scale: Study period from 2013 to 2022 (10 years). Warm-up period: 2013–2014. Calibration period: 2015–2018. Validation period: 2019–2022. Model operates on a daily timestep, with monthly actual evapotranspiration (AET) used for calibration and validation.

Methodology and Data

• Models used:
  • SWAT+ (Soil and Water Assessment Tool Plus) for agro-hydrological modeling.
  • Penman-Monteith method (FAO-56) for Potential Evapotranspiration (PET) estimation within SWAT+ and for independent AET comparison.
  • SCS Curve Number II method for runoff estimation.
  • Variable Storage method for channel water routing.
  • WGEN statistical weather generator module for filling missing meteorological data.
  • Dynamic Dimensional Search (DDS) algorithm for automatic calibration.
  • Random Balance Designs Fourier Amplitude (RBD-FAST) algorithm for sensitivity analysis.
• Data sources:
  • Topographic/Geospatial: Digital Elevation Model (DEM) from OpenDEM.
  • Soil: Digital Soil Open Land Map (DSOLMap) and a local soil map (LOCSOL) based on national soil map and field measurements.
  • Land Use: CORINE Land Cover, supplemented by crop distribution statistics.
  • Climate: Daily ground-based meteorological observations (precipitation, solar radiation, maximum/minimum temperature, relative humidity, wind speed) from three stations in Messinia.
  • Evapotranspiration (RS-ET): Global Land Evaporation Amsterdam Model (GLEAM) v3.7 and Moderate Resolution Imaging Spectroradiometer (MODIS) MOD16A2 v6.1.
  • Agricultural Management: Planting, irrigation, and harvesting schedules from local sources.

Main Results

• GLEAM AET estimates showed a strong correlation with the FAO-56 Penman-Monteith method (R² = 0.91), outperforming MODIS AET (R² = 0.59), which tended to underestimate AET.
• The most sensitive SWAT+ parameters for AET simulation were soil available water capacity (AWC), soil bulk density (BD), plant uptake compensation factor (EPCO), curve number 2 (CN2), curve number 3 (CN3_SWF), and percolation coefficient (PERCO).
• During calibration, the SWAT+ model using GLEAM AET and DSOLMap (SWAT+GLEAM_DSOLMap) achieved very good performance (mean Nash-Sutcliffe Efficiency (NSE) = 0.75, R² range 0.74–0.92, Percent Bias (PBIAS) < 9.5 % for most models).
• In validation, SWAT+GLEAM_DSOLMap remained the best combination, demonstrating satisfactory performance (mean NSE = 0.55, NSE range 0.45–0.64). All MODIS-based simulations showed unsatisfactory performance (NSE < 0.5).
• Incorporating DSOLMap, a high-resolution global soil map with a detailed six-layer profile, consistently improved simulations over the single-layer local soil map (LOCSOL).
• Model performance varied with watershed characteristics, generally better in larger watersheds (Sellas, Gianouzagas, Xerias) than in the smallest (Xerolagados, 25 km²), especially in areas with dominant agricultural land use and water bodies.
• Mean annual AET values ranged from 554 millimeters to 927 millimeters, with irrigation ranging from 77 millimeters to 272 millimeters across the four watersheds.

Contributions

• This study is the first to assess multiple remote sensing evapotranspiration (RS-ET) datasets for SWAT+ model evaluation using a multisite calibration approach in the Messinia region, Greece.
• It provides the first comparison in the Mediterranean region of the effects of a novel high-resolution global soil database (DSOLMap) and local soil maps on hydrological modeling of evapotranspiration.
• The research demonstrates the significant value of integrating freely available high-resolution global RS datasets and detailed agricultural management schedules into agro-hydrological models for data-scarce regions.
• It offers a scalable and replicable approach for enhancing water resource management and developing site-specific management strategies in other data-scarce regions globally.
• The findings highlight the superiority of GLEAM over MODIS for AET estimation in Mediterranean agricultural landscapes, attributing this to GLEAM's better accounting for soil moisture and irrigation effects.

Funding

• SALAM-MED project
• Partnership for Research and Innovation in the Mediterranean Area (PRIMA) programme
• European Union
• Grant Agreement number: [2123] [SALAM-MED] [Call 2021 Section 1 Water RIA]

Citation

@article{Bouizrou2025potential,
  author = {Bouizrou, Ismail and Castelli, Giulio and Cabrera, Gonzalo and Villani, Lorenzo and Solomos, Stavros and Maneas, Giorgos and Pantazis, Christos and Bresci, Elena},
  title = {The potential of novel remote sensing evapotranspiration data and global soil maps for SWAT+ agro-hydrological modeling in data-scarce regions of the North Mediterranean},
  journal = {Agricultural Water Management},
  year = {2025},
  doi = {10.1016/j.agwat.2025.109761},
  url = {https://doi.org/10.1016/j.agwat.2025.109761}
}