Leveraging Earth Observation for Accurate Early Forecasting of Irrigation Needs

Irrigation is a critical component of global agriculture, supporting 40% of food production on 22% of cultivated land. As climate change intensifies, the demand for irrigation water is expected to rise, particularly in vulnerable regions like the Mediterranean basin.This study presents the implementation and performance of a Spatial Decision Support System (SDSS) developed under the "EarTH Observation for the Early forecasT of Irrigation needS (THETIS)" project, funded by the Italian Space Agency, designed to forecast irrigation needs in semi-arid Mediterranean environments.The THETIS SDSS aims to provide irrigation forecasts at a basin scale, focusing on three critical stages: early, at the beginning, and during the summer season. The early stage is crucial for assessing water availability and managing irrigation efficiently. THETIS integrates a hydrological model (HM) and a crop growth model (CGM), leveraging Earth Observation (EO) data and artificial intelligence (AI) techniques to spatialize forecasted meteorological and climatic data.The SDSS combines soil water balance at two spatial scales. At the basin scale, the HM, calibrated with daily streamflow data, reliably reproduces soil moisture dynamics. At the district scale, the CGM, initialized by the HM, better models water dynamics at the local scale, accounting for factors like rain, irrigation, transpiration, evaporation, and drainage.The HM estimates soil water content at the beginning of the crop growing period, provided by the DREAM hydrological model. The CGM, based on AquaCrop and initialized by the HM, simulates crop development and forecasts evapotranspiration and irrigation needs based on meteorological forcing, hydrologic, and EO-derived information. Forecasted meteorological and climatic data are obtained from the C3S Copernicus Service. CGM outputs are early forecast water demand maps (m³/ha) at the field scale, refined as the cropping season progresses.The EO-derived information used in THETIS comes from both Synthetic Aperture Radar data (e.g., Sentinel-1, COSMO-SkyMed, SAOCOM) and optical data (e.g., Sentinel-2 and hyperspectral PRISMA). The obtained information includes maps of tilled fields , which, combined with historical land use information based on crop rotation, provide an initial estimate of irrigated areas. Maps of surface soil moisture and derived irrigated/non-irrigated fields refine the localization of irrigated areas after sowing, while vegetation index maps are used during the season for identifying sowing dates.The system has been set up over the Fortore irrigation district in the Apulian Tavoliere, Foggia, Italy, managed by the Reclamation Consortium of Capitanata, covering an area of 141 km². The SDSS performance was evaluated on tomato crops, focusing on cultivated area identification and water consumption. First results obtained for the 2022 irrigation season indicate that the water consumption of 600 m³/ha, estimated early by the THETIS SDSS using tillage change maps, is comparable to the measured value of 500 m³/ha, considering that additional water volumes from groundwater sources were likely used. The application of THETIS to the 2023 and 2024 seasons is in progress. Acknowledgment: THETIS is funded by ASI under the Agreement N. 2023-52-HH.0 in the framework of ASI’s program “Innovation for Downstream Preparation for Science” (I4DP_SCIENCE).