IrriLand, a digital twin assimilating biophysical parameters of vegetation to assess and forecast site-specific crop water requirements at irrigation district scale

In the Mediterranean region, agricultural water use accounts for a large share of the water demand and is key for food security and socio-economic stability in rural areas. At the same time, both managing irrigation in farms and managing water distribution to farms are not trivial tasks, since the water requirements by crops are site-specific and vary in time because of weather, agronomic management and other factors. In this context, the availability of EO data opens the opportunity to develop tools for the supervision, management and forecast of irrigation, scalable from farms to districts and basins. Time series of observed biophysical parameters of the vegetation and estimates of actual crop evapotranspiration (ETa) are promising resources for these applications. Those data can be assimilated into digital twins that integrate observations from different sources with models of crop development and soil water balance, enabling assessments of irrigation performance and management decision making. Here we describe a decision-making approach for irrigation district managers that assimilates EO data and simulates the water balance parameters of the soil-crop system at each individual plot. The goal is to obtain a dynamic view of irrigation performance scaling from individual plots to the basin, quantifying at real time the progress of crop growth and seasonal water balance, including forecasts of the forthcoming crop water demands under different meteorological scenarios. This approach has been implemented in the Catalan side of the Ebro basin (Spain), on an area of 2600 km² covering 105 municipalities. A separate digital twin was defined for each of over 130000 agricultural plots listed in the Land Parcel Identification System. For each plot, the agricultural scenario was set according to open data of EU CAP’s Single Farm Payment and a soil map of the area. This included the list of crops declared from 2015 to 2022, the irrigation system and the soil class. From these basic categoric data, more detailed parameters of the crop, soil and irrigation method were assigned according to the description of actual agricultural scenarios on the area. The development of the crop and its soil water balance at each individual plot is simulated at real time, using a customized model based in a rationale similar to FAO’s AquaCrop, but with additional adaptations to permanent crops, localized irrigation and discontinuous canopies. Simulations are updated every day, using online weather data from the Meteorological Service of Catalonia. In parallel, as soon as new Sentinel-2 images are available, fAPAR and LAI are computed through the Biophysical Processor available in the SNAP software and these parameters are assimilated in the model. The output are maps and time series with the estimated ETa, irrigation amounts and available soil water at each plot, accessible at www.irrilleida.cat. Time series cover the whole year, on a week basis, including the forecasts of crop water demands for the remaining part of the year.