Improving WOFOST model to simulate winter wheat phenology in Europe : Evaluation and effects on yield

This study describes and evaluates improvements to the MARS crop yield forecasting system (MCYFS) for winter soft wheat (Triticum aestivum) in Europe, based on the WOFOST crop simulation model, by introducing autumn sowing dates, realistic soil moisture initialization, adding vernalization requirements and photoperiodicity, and phenology calibration. Dataset of phenological observations complemented with regional cropping calendars across Europe is used. The calibration of thermal requirements for anthesis and maturity is done by pooling all available observations within European agro-environmental zones and minimizing an objective function that combines the differences between observed and simulated anthesis, maturity and harvest dates. Calibrated phenology results in substantial improvement in simulated dates of anthesis with respect to the original MCYFS simulations. The combined improvements to the system result in a physically more plausible spatial distribution of crop model indicators across Europe. Crop yield indicators point to better agreement with recorded national winter wheat yields with respect to the original MCYFS simulations, most pronounced in central, eastern and southern Europe. However, model skill remains low in large parts of western Europe, which may possibly be attributed to the impacts of wet conditions.