Galli et al. (2026) Integrating biophysical models and remote sensing to evaluate irrigation practices in four global hubs

Identification

• Journal: Agricultural Water Management
• Year: 2026
• Date: 2026-03-15
• Authors: Nikolas Galli, Francesco Capone, Jacopo Dari, Davide Danilo Chiarelli, Maria Cristina Rulli, Clément Abergel, Carla Saltalippi, Renato Morbidelli, Luca Brocca
• DOI: 10.1016/j.agwat.2026.110284

Research Groups

• Department of Civil and Environmental Engineering, Politecnico di Milano, Milano, Italy
• Department of Civil and Environmental Engineering, University of Perugia, Perugia, Italy
• CNR-IRPI, Perugia, Italy
• ECSAT, European Space Agency, Oxford, UK

Short Summary

This study compares irrigation demand from an agro-hydrological model (WATNEEDS) with irrigation water use from five satellite products across four global irrigation hubs, finding significant correlations (above 0.6 for 3/4 cases) and using discrepancies to identify hydroclimatic and anthropogenic irrigation drivers.

Objective

• To present a novel comparison of irrigation demand simulations from a spatially distributed agro-hydrological model and irrigation water use estimates from five satellite retrievals over four global irrigation hubs.
• To provide a proof-of-concept of the value of analyzing discrepancies between process-based and Earth Observation-based irrigation time series to identify margins for mutual improvement and investigate responses to exogenous (climatic and anthropogenic) drivers.

Study Configuration

• Spatial Scale: Four global agricultural hubs: Continental United States (CONUS), India, Ebro River Basin (Spain), and Murray–Darling Basin (Australia). All data resampled to a 0.25° regular grid. Analysis restricted to pixels with at least 15% irrigated area.
• Temporal Scale: 2003–2022. Daily model simulations and satellite retrievals, cumulated monthly for comparison.

Methodology and Data

• Models used:
  • Agro-hydrological model: WATNEEDS (spatially distributed, physically based, dynamic, crop-specific soil water balance model).
  • Remote sensing irrigation estimation: SM-Inversion algorithm (soil moisture-based inversion framework).
• Data sources:
  • Satellite soil moisture products (for SM-Inversion): Advanced SCATterometer (ASCAT) v7, Soil Moisture Ocean Salinity (SMOS L2) v700, Soil Moisture Active Passive (SMAP L2) v8, Climate Change Initiative (CCI) Combined v08.1, and Climate Change Initiative (CCI) Passive v08.1.
  • Other satellite/reanalysis data: Global Land Evaporation Amsterdam Model (GLEAM) v3.7b for potential evapotranspiration, ERA5 (European ReAnalysis v5) for rainfall rates.
  • Ancillary data: Portmann et al. (2008) calendar for irrigation seasons, Mehta et al. (2022) and MIRCA-OS (Kebede et al., 2024) for areas equipped for irrigation, meteoclimatic, soil, and crop inputs for WATNEEDS (Chiarelli et al., 2020b).

Main Results

• WATNEEDS Blue Water (BW) simulations showed comparable seasonal patterns and magnitudes with satellite-derived Irrigation Water Use (IWU) estimates across the four case studies.
• Statistically significant linear correlations (p-value < 0.001) above 0.6 were found for CONUS, Ebro Basin, and Murray-Darling Basin, with CONUS showing the highest correlations (above 0.8 for all products).
• India exhibited the lowest agreement, with the highest Root Mean Square Error (RMSE) of 23.4 mm/month (CCI-PASSIVE) and near-zero/negative correlations for some products, primarily due to WATNEEDS underestimating the Rabi and Zaid rice cycles.
• Discrepancies between BW and IWU highlighted responses to hydroclimatic and anthropogenic drivers:
  • In the Murray-Darling Basin, WATNEEDS underestimated IWU peaks during major flood events (e.g., 2010-2011, 2017), which were captured by satellite products.
  • In CONUS (2012), WATNEEDS showed higher BW peaks than satellite IWU, consistent with a severe drought where demand exceeded actual water application.
  • In the Ebro Basin, satellite IWU showed a sharper increase at the onset of the irrigation season (April) compared to WATNEEDS, indicating early spring irrigation practices not fully captured as demand by the biophysical model.
• Correlation maps revealed site-specific spatial gradients, with agreement generally increasing in areas with a larger extent of irrigated land. Different satellite products showed varying performance across different regions within the same case study.

Contributions

• This study provides the first large-scale comparative assessment of biophysical irrigation demand modeling and satellite-driven irrigation water use estimates.
• It serves as a proof-of-concept for leveraging discrepancies between process-based models and Earth Observations to gain insights into human-water interactions and feedbacks in irrigation dynamics.
• The research identifies specific limitations and areas for mutual improvement in both agro-hydrological models (e.g., incorporating dynamic cropping calendars, accounting for flood events) and satellite products (e.g., disentangling irrigation signals in mixed pixels, identifying optimal product ensembles).
• It paves the way for developing more integrated and refined analytical techniques to combine process-based modeling and Earth Observation for comprehensive irrigation monitoring and assessment at regional-to-global scales.

Funding

• ESA under the CCI-AWU (Climate Change Initiative Anthropogenic Water Use) precursor project (contract n. 4000142449/23/I-NB).

Citation

@article{Galli2026Integrating,
  author = {Galli, Nikolas and Capone, Francesco and Dari, Jacopo and Chiarelli, Davide Danilo and Rulli, Maria Cristina and Abergel, Clément and Saltalippi, Carla and Morbidelli, Renato and Brocca, Luca},
  title = {Integrating biophysical models and remote sensing to evaluate irrigation practices in four global hubs},
  journal = {Agricultural Water Management},
  year = {2026},
  doi = {10.1016/j.agwat.2026.110284},
  url = {https://doi.org/10.1016/j.agwat.2026.110284}
}