Optimizing variable rate irrigation using model-based dynamic prescription maps

Variable rate irrigation is usually based on prescription maps delineated according to a static approach. Irrigation rate and timing are optimised by sensor and/or modelling-based methods applied within homogenous zones whose spatial distribution is kept constant during the crop season. The objective of this study was to develop a procedure based on the combination of the crop-energy-water balance model FEST-EWB-SAFY with remote sensing data of vegetation variables and land surface temperature to generate dynamic irrigation prescription maps. The crop-energy-water balance FEST-EWB-SAFY model couples the distributed energy-water balance FEST-EWB, which allows computing continuously in time and distributed in space both soil moisture and evapotranspiration fluxes, and the SAFY (simple model for yield prediction and plant development).

The model was tested in a 30-ha field cultivated with soybean in 2022 at Ceregnano, in the lower zone of the Po Valley (Italy). Irrigation was provided by 270m long lateral move irrigation machine, equipped with a precision irrigation system with a lateral resolution of 34 m. The model was pixelwise calibrated with remotely sensed land surface temperature (LST, RMSE 1.3 °C) and leaf area index (RMSE 0.45) as well as local measured soil moisture at 10cm and 50cm depth (RMSE 0.04). Four dynamic prescription maps were generated during the season, calculating the pixel-by-pixel difference between the field retention capacity and the daily average of the 50-cm soil moisture profile. Dynamic variable rate irrigation was compared with a conventional irrigation system according to an experimental block design with three replicates and evaluated in terms of crop yield, irrigation volumes and water use efficiency.

FEST-EWB-SAFY allowed the creation of dynamic maps that captured the crop water requirement variability originated by the interaction of ET, soil properties and field management. Compared with the conventional system, there was a significant increase in water use efficiency, but not in crop yield. These results confirm that the model-based dynamic prescription maps could be used to optimize variable irrigation in highly spatio-temporal dynamic cropping systems