GLORIFY : a new forecasting system for rice grain quality in Northern Italy

A reliable forecast of the pre-harvest grain quality is requested by stakeholders in the rice sector, which is increasingly oriented to the achievement of superior standards to meet the market demand. Despite its economic importance, very few simulation models of the qualitative aspects of rice productions including the effects of weather conditions and farming practices are available. This paper presents GLORIFY, a forecasting system targeting the simulation of head rice yield (HRY), which represents the main determinant of rice market price at global level. A new HRY model was developed using experimental data collected in Northern Italy in 2006–2013 and referred to Loto (japonica) and Gladio (tropical japonica) cultivars, and it was coupled to the WARM rice simulator. Historical simulations were then performed in the period 1994–2013 to reproduce observed HRY variability, with model outputs and weather variables used as independent variables to build multi-regression models. At field level, model performances denoted a good agreement between observed and simulated HRY (R2 and modelling efficiency in the range 0.73–0.93). At province level, best results were obtained for Loto variety, as the regression model was able to explain 78% of the HRY variability, with a root mean square error (RMSE) of 0.77%. The model accuracy slightly decreased when leave-one-out cross-validation was applied (R2 = 0.61, RMSE = 1.04%). The present study lays the basis for a reliable estimation of HRY variability under different management and weather conditions.