Bongiovanni et al. (2025) Mapping soil water properties using soil samples and satellite images in the irrigated area of Biota (Spain)

Identification

• Journal: Irrigation Science
• Year: 2025
• Date: 2025-11-20
• Authors: Michel Bongiovanni, A. Laguet, P. Paniagua, Eva García, Calogero Romano, J. Fernández-Pato, N. Zapata, Enrique Playán Jubillar
• DOI: 10.1007/s00271-025-01050-9

Research Groups

• EEA INTA Hilario Ascasubi, Buenos Aires, Argentina
• Department of Soil and Water, EEAD-CSIC, Zaragoza, Spain
• Polytech Nice-Sophia, Biot, France
• ARAID Foundation, Zaragoza, Spain

Short Summary

This study developed and assessed quantitative and qualitative Total Available Water (TAW) maps for a shallow, stony irrigated area in Spain using Sentinel-2 imagery and field data. The quantitative maps, derived from multi-temporal satellite data, provided finer spatial detail and correlated with crop yield, offering enhanced value for irrigation management compared to traditional qualitative approaches.

Objective

• To develop two thematic TAW soil maps for the Monte Saso de Biota (MSB) irrigated area: a classic qualitative map with soil units and a quantitative map with TAW (and other soil water properties) values assigned to pixels from satellite imagery.
• To assess the quality and coherence of both maps by comparing them with crop yield maps and by their agreement with farmers’ perception and knowledge.

Study Configuration

• Spatial Scale: Monte Saso de Biota (MSB) irrigated area, 11,850,000 square meters (1185 hectares), Zaragoza, Spain (42°13’01” N, 1°13’46” W, WGS84 datum). Quantitative maps were generated at a 10 meter pixel resolution.
• Temporal Scale:
  • Field soil sampling: November 2023 to February 2024.
  • Satellite imagery: 82 bare-soil Sentinel-2 images collected from December to May, annually from 2018 to 2024.
  • Crop yield data: 2024 growing season.

Methodology and Data

• Models used:
  • Total Available Water (TAW) calculation: TAW = 10 * Z * ((FC - WP) / 100) * (ρb / ρw) * (1 - STO / 100), where Z is effective soil depth, FC is field capacity, WP is wilting point, ρb is bulk density, ρw is water density, and STO is volumetric stoniness.
  • Regression models: Ordinary Least Squares (OLS) regression with backward stepwise selection for single-date models. Combined (combi) multi-temporal regression models using estimates from single-date models as independent variables.
  • GIS tools: SAGA GIS for Total Catchment Area (TCA) analysis (Deterministic 8 and Multiple Flow Direction methods), QGIS for data conversion, extraction, and map generation.
  • Crop yield map generation: R functions within QGIS, employing Kriging interpolation.
• Data sources:
  • Field data: 172 soil samples (0.0–0.2 meter depth) for gravimetric water content at field capacity (FC), wilting point (WP), bulk density (ρb), stoniness (STO), and texture (laser diffraction). Effective soil depth (Z) was determined from 14 soil pits, 12 auger boreholes, and 50 historical soil pits.
  • Satellite data: 82 atmospherically corrected Sentinel-2 L2A images (10 meter and 20 meter spatial resolution bands) from the Copernicus program. Frequent predictor bands included B02, B08, B11, and B12.
  • Auxiliary spatial data: Digital Terrain Model (DTM), Google Earth maps, and Total Catchment Area (TCA) analysis.
  • Crop yield data: Maize yield maps (kilograms per square meter) from combine harvesters equipped with GPS and yield monitors for two center pivots.
  • Climate data: Average annual rainfall of 0.421 meters and reference evapotranspiration of 1.335 meters (2011–2024 average from El Bayo station, SIAR network).

Main Results

• Soil Characteristics: MSB soils are shallow (0.30–0.60 meters effective depth), stony (average 32% volumetric stoniness, with 42% of samples exceeding 35%), and have limited TAW (ranging from 0.019 to 0.096 meters, average 0.046 meters).
• Model Performance (Calibration): Combined (combi) multi-temporal models consistently outperformed single-date models. The highest coefficients of determination (R²) for combi models were: WP (92%), Z (89%), STO (84%), TAW (77%), and FC (72%).
• Model Performance (Validation): Independent validation of the quantitative maps yielded R² values in the range of 19–45%. Specifically, TAW validation R² was 28% (minimum SE strategy) and 19% (maximum R² strategy).
• Spectral Bands: Sentinel-2 bands B02 (blue), B08 (near-infrared), B11 (shortwave infrared), and B12 (shortwave infrared) were identified as frequent and important predictors for TAW, STO, Z, and WP.
• Map Comparison: The qualitative TAW map captured general terrain-related patterns. The quantitative TAW map provided finer spatial detail and showed a significant positive correlation with maize yield in one pivot (r = 0.35, p > 0.001), indicating its utility for understanding yield variability.

Contributions

• Developed and assessed both quantitative (pixel-level, Sentinel-2 based) and qualitative (soil unit based) TAW maps, specifically addressing the challenges of mapping in shallow, stony irrigated soils where existing literature is limited.
• Demonstrated the effectiveness of integrating multi-temporal bare-soil Sentinel-2 imagery with extensive field data and auxiliary spatial information for improved accuracy in soil property mapping.
• Provided evidence that quantitative TAW maps offer finer spatial detail and a measurable correlation with crop yield, thus enhancing their value for precision irrigation management and numerical analyses (e.g., irrigation and crop yield simulations).
• Enhanced the understanding of local soil hydrological responses in the Monte Saso de Biota area, contributing to sustainable water and energy management in modernized irrigation systems.

Funding

• Spanish Ministry of Science and Innovation: Project “Irrigation modernization sustainability: controlling zebra mussel in pressurized irrigation networks and optimizing on-farm sprinkler irrigation” (PID2021-124095OB-I00). Co-financed by the European Regional Development Fund (ERDF).
• Instituto Nacional de Tecnología Agropecuaria (INTA, Argentina) and the project “BIRF, sistemas agroalimentarios inclusivos e inteligentes desde el punto de vista climático” (P176905-BIRF) for funding a doctoral stay.
• Open Access funding provided through the CRUE-CSIC agreement with Springer Nature.

Citation

@article{Bongiovanni2025Mapping,
  author = {Bongiovanni, Michel and Laguet, A. and Paniagua, P. and García, Eva and Romano, Calogero and Fernández-Pato, J. and Zapata, N. and Jubillar, Enrique Playán},
  title = {Mapping soil water properties using soil samples and satellite images in the irrigated area of Biota (Spain)},
  journal = {Irrigation Science},
  year = {2025},
  doi = {10.1007/s00271-025-01050-9},
  url = {https://doi.org/10.1007/s00271-025-01050-9}
}