Lannoy et al. (2026) Advancing crop modeling and data assimilation using AquaCrop v7.2 in NASA's Land Information System Framework v7.5

Identification

• Journal: Geoscientific model development
• Year: 2026
• Date: 2026-03-31
• Authors: Gabriëlle J. M. De Lannoy, Louise Busschaert, Michel Bechtold, Niccolò Lanfranco, Shannon de Roos, Zdenko Heyvaert, Martynas Bielinis, Jonas Mortelmans, Samuel Scherrer, Maxime Van den Bossche, Sujay V. Kumar, David M. Mocko, Eric Kemp, Lee Heng, Pasquale Steduto, Dirk Raes
• DOI: 10.5194/gmd-19-2551-2026

Research Groups

• KU Leuven, Department of Earth and Environmental Sciences, Heverlee, Belgium
• NASA/GSFC, Hydrological Sciences Laboratory, Greenbelt, MD, USA
• Politecnico di Torino, Department of Environment, Land and Infrastructure Engineering, Torino, Italy
• Vrije Universiteit Brussel, Department of Water and Climate, Brussel, Belgium
• European Centre for Medium-Range Weather Forecasts (ECMWF), Research Department, Reading, United Kingdom
• Science Applications International Corporation, Reston, VA, USA
• Food and Agriculture Organization (FAO), Land and Water Division, Rome, Italy (formerly)
• International Atomic Energy Agency, Vienna, Austria (formerly)

Short Summary

This paper integrates the AquaCrop v7.2 crop model into NASA's Land Information System Framework (LISF) v7.5, enabling high-performance, scalable crop modeling and satellite data assimilation. Through three showcases, it demonstrates improved canopy cover simulations with satellite-informed crop parameters, quantifies biomass uncertainty in relation to soil moisture, and explores the beneficial but limited impact of fractional vegetation cover (FCOVER) assimilation on yield estimates due to strong model constraints.

Objective

• To introduce the open-source AquaCrop v7.2 model as a dynamical state propagation model within NASA's LISF v7.5 and to pave the way for large-scale satellite data assimilation into a crop model.
• To demonstrate the current potential and limitations of (i) coarse-scale crop growth simulation with various crop parameterizations, (ii) coarse-scale ensemble simulations, and (iii) ensemble Kalman filtering of fine-scale satellite data into AquaCrop.

Study Configuration

• Spatial Scale:
  • Coarse-scale (approximately 11 km x 11 km grid cells, 0.1°) for generic crop growth and ensemble simulations over Europe.
  • Fine-scale (approximately 900 m x 900 m grid cells, 1/112°) for winter wheat simulations and data assimilation over the Piedmont region of Italy.
  • Satellite data used at 0.05° (VIIRS GLSP, approximately 5.5 km x 5.5 km) and 1/336° (CGLS FCOVER, approximately 300 m x 300 m).
• Temporal Scale:
  • Showcase 1 (Crop Parameterization): 2015–2020 (6 years).
  • Showcase 2 (Ensemble Simulations): 2015–2017 (3 years).
  • Showcase 3 (Data Assimilation): 2017–2023 (7 years).
  • Model simulations performed at daily time steps; evaluation and assimilation use 10-day averaged data.

Methodology and Data

• Models used:
  • AquaCrop v7.2 (open-source Fortran version).
  • NASA's Land Information System Framework (LISF) v7.5.
  • Ensemble Kalman Filter (EnKF) for data assimilation.
• Data sources:
  • Satellite:
    • Visible Infrared Imaging Radiometer Suite (VIIRS) Global Land Surface Phenology (GLSP) product (0.05°) for crop parameterization.
    • Copernicus Global Land Service (CGLS) Fraction of Vegetation Cover (FCOVER) (1/336°) for assimilation and evaluation.
    • Copernicus Global Land Service (CGLS) Dry Matter Productivity (DMP) for evaluation.
    • Copernicus High Resolution Layer Croplands product (10 m) and Geoportale Piemonte regional crop maps (1:2000) for masking FCOVER observations.
  • Reanalysis:
    • Fifth-generation European Center for Medium-Range Weather Forecasts (ECMWF) Reanalysis (ERA5) for meteorological forcings (temperature, precipitation, reference evapotranspiration).
  • Observation/Ancillary:
    • Harmonized World Soil Database 1.21 for soil texture.
    • University of Maryland 1 km global land cover data set for cropland identification.
    • Mauna Loa Observatory for atmospheric carbon dioxide (CO₂) concentrations.
    • Global Yield Gap Atlas (GYGA) and RICA (Italian survey data) for in situ yield reference data.

Main Results

• Crop Parameterization (Showcase 1): Spatially variable crop parameters derived from satellite-based land surface phenology (GDD mode) significantly improve canopy cover (CC) simulations across Europe (increased time series correlation) compared to uniform calendar-day parameters. Biomass (ΔB) simulations also improve for much of southern and eastern Europe.
• Ensemble Simulations (Showcase 2): Ensemble simulations perturbing meteorological forcings and soil moisture reveal that biomass uncertainty is often greater in water-limited regions, where root-zone soil moisture spread is higher. This highlights the sensitivity of biomass to soil moisture variability under stressed conditions.
• Satellite Data Assimilation (Showcase 3): Assimilation of high-resolution FCOVER observations into AquaCrop for winter wheat in the Piedmont region improves CC (by design) and intermediary biomass estimates. However, the impact on end-of-season yield estimates is small (correlation R increases from 0.07 to 0.12), primarily due to strong model constraints (e.g., fixed crop stages, maximum CC, fertility, and harvest index parameters) and potentially insufficient informativeness of FCOVER for yield. The model-only simulations exhibit low spatiotemporal variability in yield.
• Computational Efficiency: All showcase experiments, despite continental coverage or high resolution, were completed within a few hours of walltime on a high-performance Linux cluster, demonstrating the efficiency of the LISF-AquaCrop system.

Contributions

• First-time integration of the open-source AquaCrop v7.2 model into NASA's Land Information System Framework (LISF) v7.5, enabling scalable, high-performance crop modeling and data assimilation capabilities.
• Demonstration of a novel approach to derive spatially variable generic crop parameters from satellite-based phenology products, leading to improved coarse-scale canopy cover and biomass simulations over Europe.
• Quantification of the sensitivity of coarse-scale biomass uncertainty to meteorological and soil moisture perturbations, providing insights for future data assimilation strategies.
• Exploration of fine-scale satellite FCOVER data assimilation for state updating in AquaCrop, identifying key limitations related to model structural constraints and observation informativeness, and outlining pathways for future advancements (e.g., state-dependent crop stages, joint parameter-state updating, multi-sensor and multi-variate data assimilation).
• Release of the open-source Fortran version of AquaCrop v7.2 and its LISF implementation, promoting wider use and community-based development.

Funding

• Fonds Wetenschappelijk Onderzoek (grant nos. 1158423N and Storage4Climate)
• Horizon 2020 (grant no. 773903)
• KU Leuven (grant no. C14/21/057)
• Belgian Federal Science Policy Office (grant no. SR/00/412)

Citation

@article{Lannoy2026Advancing,
  author = {Lannoy, Gabriëlle J. M. De and Busschaert, Louise and Bechtold, Michel and Lanfranco, Niccolò and Roos, Shannon de and Heyvaert, Zdenko and Bielinis, Martynas and Mortelmans, Jonas and Scherrer, Samuel and Bossche, Maxime Van den and Kumar, Sujay V. and Mocko, David M. and Kemp, Eric and Heng, Lee and Steduto, Pasquale and Raes, Dirk},
  title = {Advancing crop modeling and data assimilation using AquaCrop v7.2 in NASA's Land Information System Framework v7.5},
  journal = {Geoscientific model development},
  year = {2026},
  doi = {10.5194/gmd-19-2551-2026},
  url = {https://doi.org/10.5194/gmd-19-2551-2026}
}