LAUSR

Coupling crop growth models and remote sensing data has shown great potential to improve crop yield and biomass forecasting. However, few such applications are used at field level. In this study, the plug-in version of AquaCrop, a water driven model initially designed for conditions in which water is the limiting factor in crop production, was adapted to assess maize in Belgium. Field campaigns were organised in 2015, 2016 and 2017, during which ground measurements were collected. At field level and under rainfed conditions, above ground total dry biomass and field management information (crop density, planting dates, biomass measurements etc.) was collected. Green fractional vegetation cover (fCover) data were retrieved from high spatial and temporal satellite images (DMC-2/ DEIMOS-1 and Sentinel-2). Maximum canopy cover and emergence date were derived from the time series and assimilated into AquaCrop. The model was calibrated with the 2015–2016 dataset and validated with 2017 data. The root mean square error (RMSE) for the biomass assessment was 0.7 ton/ha and the R² reached 0.85. The R² between the canopy cover simulated by the model and the remotely sensed fCover ranged from 0.76–0.98 in most of the cases. To simultaneously run and evaluate the ensemble of field-level simulations, a semi-automated R-environment was developed. This robust and automated approach offers vast potential for large-scale field-level yield assessments for temperate as well as tropical conditions.