Abstract/Summary

This contribution presents a novel approach to characterise uncertainty in the manual grape harvest of a winery in Tuscany (Italy). After identifying the potential sources of variability arising from randomness, weather, and management options, a model to define useful output variables is built. These output variables include the discrepancy in the harvest date of the vineyards (harvest date discrepancy), the discrepancy in the required workforce across harvest dates (labour discrepancy), and, finally, the potential deficit of working hours throughout the grape harvest campaign (labour deficit). The range spanned by these variables is first assessed through a Monte Carlo uncertainty analysis wherein the model is repeated approximately 16,000 times with variable combinations of the input parameters per their probability distribution. The assessed uncertainty is then apportioned to the input parameters through a global sensitivity analysis. In turn, a regional sensitivity analysis characterises the circumstances producing a deficit of working hours, which corresponds to sufferance in the grape harvest campaign. The discussed approach could be implemented in a user-friendly decision-support tool for risk characterisation and efficient grape harvest management.