Evaluation of ECMWF Seasonal Forecast Performance for Operational Irrigation Planning in the LEB Consortium

Climate change and increasing hydro-climatic variability pose growing challenges for irrigation water management in Mediterranean regions, where agricultural production and water availability are highly sensitive to seasonal weather conditions. In Northern Italy, recent drought events have exposed the limitations of irrigation planning approaches based on historical records and short-term forecasts. This work is part of a broader effort initiated by the Lessinio–Euganeo–Berico (LEB) Consortium to improve the reliability of seasonal forecasting for irrigation management in Northern Italy. Seasonal ensemble forecasts from the European Centre for Medium-Range Weather Forecasts (ECMWF) are used to drive a district-scale soil–water balance model that estimates irrigation demand up to 45 days ahead across the 26 irrigation districts managed by the consortium. Forecast-based simulations are compared with observation-driven simulations derived from ARPAV meteorological data. The analysis evaluates the full forecasting chain—from meteorological inputs to irrigation demand—across five representative years (2010, 2014, 2017, 2022, 2024). Forecast performance is assessed using Percentage Bias (PBIAS), Root Mean Square Error (RMSE), Mean Bias Error (MBE), and Pearson correlation coefficient (r) for precipitation, reference evapotranspiration, crop evapotranspiration, infiltration, and irrigation demand. Results show that forecast skill is strongly variable-dependent, with higher reliability for temperature-driven processes and substantial uncertainty in precipitation. These errors are amplified through nonlinear soil–water balance dynamics, leading to systematic overestimation of irrigation demand, particularly under dry conditions. Forecast performance degrades with lead time and varies across hydro-climatic regimes, scenarios, and districts, with spatial clustering indicating structured rather than random error patterns. Overall, while seasonal forecasts capture general hydro-climatic trends, their direct application to irrigation demand remains limited by precipitation uncertainty and nonlinear error propagation. This study provides a district-scale evaluation of ECMWF forecasts within an operational irrigation system and supports the development of more reliable climate-informed irrigation planning strategies.