Integration of Sentinel-2 Imagery with the AquaCrop-OSPy Model for Simulating Agricultural Crop Requirements and Growth in Desert Farming Systems: A Saudi Arabian Case Study

Integrating remote sensing technology into crop growth models is a viable approach for water resources management and agricultural sustainability assurance since it allows crop water requirements and yield within agricultural fields to be estimated. Saudi Arabia has severely limited renewable water resources and non-renewable groundwater reserves that are rapidly depleting. Unlike rain-fed agriculture, the majority of agricultural water demand in Saudi Arabia is pumped from deep aquifers (up to 1000 m) to irrigate center pivots. This situation entails continuous monitoring of agricultural water use to enhance agricultural water productivity (i.e., producing more crops per drop) and preserve the equilibrium among the water, food, and energy sectors. The main purpose of this study is to calibrate the AquaCrop-OSPy model (an open-source Python implementation of the FAO crop-water productivity model, known as AquaCrop) using field data and Sentinel-2 (S2) images for operational mapping of crop yield, water demand, and water productivity. Another objective is to spatiotemporally estimate the energy requirements and associated CO2 emissions related to groundwater pumping for irrigation. The study area is located in the north of Saudi Arabia. It is a commercial farm with an area of 30,000 hectares comprising more than 200 agricultural fields with center-pivot irrigation systems. The crops cultivated on the farm are wheat and tomato. Field data were collected over three consecutive growing seasons (2019-2020, 2020-2021, and 2021-2022) and include information on wells, pumps, irrigation technique, field management practices, soil parameters, crop parameters, daily meteorological data, actual crop yield, and water use. The AquaCrop-OSPy model was first calibrated and validated using the collected field data as well as S2 images over the three seasons. Subsequently, the fractional vegetation cover (FVC) derived from S2 images was assimilated into the AquaCrop-OSPy model by direct insertion in place of AquaCrop-OSPy's simulated canopy cover (CC). Later, the energy requirements and CO2 emissions associated with irrigation groundwater pumping were estimated using crop water demand information calculated with the calibrated AquaCrop-OSPy model along with pumps and wells data. Coupling the S2-derived FVC and the AquaCrop-OSPy model improved AquaCrop-OSPy predictions of crop water demand, yield, and water productivity as S2 images provide spatialized FVC information every 6-days. This integration further permitted a robust quantification of the energy requirements and CO2 emissions associated with groundwater pumping for irrigation. These results, when applied to larger scales and multiple crops, can help develop a comprehensive understanding of the water-energy-agriculture nexus and indicate potential improvements in AquaCrop-OSPy estimates that could be achieved once remote sensing data are integrated.