High-resolution climate modelling for the Cognac region under climate change

Abstract

Context and purpose of the study. Climate change has varied effects across French vineyards, with marked regional differences in temperature shifts. Fine-scale studies highlight significant local climate variability, emphasizing the need for precise regional characterization to adapt vineyard management at the regional scale. This study focuses on the Cognac region, France’s leading area for white grape production, covering nearly 87,000 hectares and known for its oak-barrel-aged brandy, Cognac. It aims to establish a high-resolution climate zoning model through extensive on-site climate measurements, facilitating the calculation of agro- and eco-climatic indicators tailored to viticulture.

Material and method. Data for this study were gathered from three distinct measurement networks across the Cognac region. The first network consists of 26 Météo-France weather stations with historical data spanning 20 to 70 years. The second network includes 50 connected weather stations operational since 2018, and the third, a network of 100 temperature sensors, was deployed in the vineyards in 2022. Using this data, three statistical models were tested to interpolate temperature and rainfall on a 1 km² grid: artificial neural networks (ANNs), support vector regression (using non-linear regressions), and multiple linear regression model with smoothing of residuals that incorporates environmental variables such as altitude, slope, distance from the sea, etc.

Results. Preliminary model comparisons suggest slight differences in efficiency. Once interpolation at a fine scale was achieved, it allowed the calculation of agro- and eco-climatic indicators, revealing significant spatial variations across the Cognac region. This method will also be used to assess the impact of future climate projections, offering insights into potential regional adaptations for viticulture.