Preliminary application of a simulation model to reproduce hazelnut development and growth in Italy, Georgia and Chile

Process-based simulation models have been increasingly used in agriculture for crop yield forecasting and in-season monitoring of crop growth, to support stakeholders, from farmers to agribusiness companies, and regional policies. The study shows the application of a modelling solution for hazelnut trees, which provides a dynamic simulation of the plant physiological processes associated with development and growth, as affected by variable agro-pedo-climatic conditions. The modelling solution was tested using reference data collected in 2015-2016 in four hazelnut orchards placed in Italy (Baldissero d’Alba, Oriolo Romano), Georgia (Chitaskari) and Chile (Camarico). The data collection (i.e., growth stage, LAI, fruit biomass, crown size, soil water content) was carried out on ten adjacent plants grown in a homogenous area of the field. Simulation results indicated a good correlation with observed phenological development, with R 2 >0.7 and relative root mean square error RRMSE <20% for vegetative and reproductive phases. The modelling solution was able to reproduce the LAI dynamics (mean absolute error MAE <0.77 m 2 m -2 ; RRMSE <23.24%), characterized by large differences across sites, being reproduced by adjusting parameters related to crown shape and leaf area density. Similarly, the yield trend, that shows an average production for the four sites higher in 2016 (6.32 kg plant -1 ) than in 2015 (4.45 kg plant -1 ), was correctly reproduced, with simulations leading to MAE between 0.25 and 1.02 kg plant -1 . This modelling solution is a promising tool to provide information on the orchard status and to support evaluations on crop performance in different areas and climate change scenarios, even if further tests are needed to consolidate these preliminary results.