Persistence versus dynamical seasonal forecasts of cereal crop yields

Climate change is expected to have impacts on the balance of global food trade networks and food security. Thus, seasonal forecasts of precipitation and temperature are an essential tool for stakeholders to make timely choices regarding the strategies required to maximize their expected cereal yield outcomes. The availability of state-of-the-art seasonal forecasts such as the European Centre for Medium-Range Weather Forecasts (ECMWF) system 5 (SEAS5) may be an asset to help decision making. However, uncertainties and reduced skill may hamper the use of seasonal forecasts in several applications. Hence, in this work, we aim to understand the added value of such dynamical forecasts when compared to persistent anomalies of climate conditions used to predict the production of wheat and barley yields. With that in mind, empirical models relating annual wheat and barley yields in Spain to monthly values of precipitation and temperature are developed by taking advantage of ECMWF ERA5 reanalysis. Then, dynamical and persistence forecasts are issued at different lead times, and the skill of the subsequent forecasted yield is verified through probabilistic metrics. The results presented in this study demonstrate two different outcomes: (1) wheat and barley yield anomaly forecasts (dynamical and persistent) start to gain skill later in the season (typically from April onwards); and (2) the added value of using the SEAS5 forecast as an alternative to persistence ranges from 6 to 16%, with better results in the southern Spanish regions.