Can Grapevine Leaf Water Potential Be Modelled from Physiological and Meteorological Variables? A Machine Learning Approach

Abstract

Climate change is affecting global viticulture, increasing heatwaves and drought. Precision irrigation, supported by robust water status indicators (WSIs), is inevitable in most of the Mediterranean basin. One of the most reliable WSIs is the leaf water potential ((Formula presented.)), which is determined via an intrusive and time-consuming method. The aim of this work is to discern the most effective variables that are correlated with plants’ water status and identify the variables that better predict (Formula presented.). Five grapevine varieties grown in the Alentejo region (Portugal) were selected and subjected to three irrigation treatments, starting in 2018: full irrigation (FI), deficit irrigation (DI), and no irrigation (NI). Plant monitoring was performed in 2023. Measurements included stomatal conductance ((Formula presented.)), predawn water potential (Formula presented.), stem water potential ((Formula presented.)), thermal imaging, and meteorological data. The WSIs, namely (Formula presented.) and (Formula presented.), responded differently according to the irrigation treatment. (Formula presented.) measured at mid-morning (MM) and mid-day (MD) proved unable to discern between treatments. MM measurements presented the best correlations between WSIs. (Formula presented.) showed the best correlations between the other WSIs, and consequently the best predictive capability to estimate (Formula presented.). Machine learning regression models were trained on meteorological, thermal, and (Formula presented.) data to predict (Formula presented.), with ensemble models showing a great performance (ExtraTrees: (Formula presented.), (Formula presented.) ; Gradient Boosting: (Formula presented.) ; (Formula presented.)).