Predicting vineyard's evolution with the crop model IVINE driven by meteorological model forecasts: preliminary results.

The crop growth model IVINE (Italian Vineyard Integrated Numerical model for Estimating physiological values) was developed at our Dept. of Physics in FORTRAN language as a research model in order to evaluate the environmental forcing effects on vine growth, being vines generally strongly sensitive to meteorological conditions, and with the idea of using it for assessing climate change effects on grape growth. IVINE requires a set of hourly meteorological and soil data as boundary conditions. Input data that are more relevant for the model to correctly simulate the plant growth are air temperature and soil moisture. Among the principal IVINE outputs, we mention: the main phenological stages (dormancy exit, bud-break, fruit set, veraison, and harvest), the Leaf Area Index, the yield, the berry sugar concentration and the predawn leaf water potential. IVINE model requires to set some experimental parameters depending on the cultivar; at present, IVINE is optimized for Nebbiolo and other northern Italy autocthonous and common varieties. In order to use the model for forecasting purposes, the set of input data required by IVINE must be retrieved by the simulation's outputs of a mesoscale model, in turn driven by a Global Circulation Model simulation. In our Department, a voluntary meteorological forecasting service has been working for several years; for this task four daily 5-days simulations are performed over Piedmont Italian region with WRF (Weather Research and Forecast) mesoscale model driven by the GFS (Global Forecast System). Taking advantage of these runs, we have organized a system able to extract, for each simulation, the hourly values of the parameters needed by IVINE. The input dataset is updated every six hours using the values coming by the new simulation, while considering past values acquired. Since IVINE simulation must start from the previous season, in order to correctly simulate the dormancy exit, we have carried out several simulations with IVINE by starting in the same date (January 1^st 2018) and ending at the fifth day of the last available WRF simulation. In this way, we were able to made a sort of temporal ensemble meteogram for the last five days; where the results of the most recent simulation were displayed with those of previuos runs and the number of simulations was gradually decreasing from 20 to 1 with the progress of the time.

The simulations were performed for the whole 2019 year over 156 WRF grid points distributed in the Langhe, Roero and Monferrato wine areas of Piedmont. Here some pheno-physiological variables in vineyards are analyzed, relative to some significant points and events, and the main results are discussed.