Analysis of spatio-temporal bias of Weather Research and Forecasting temperatures based on weather pattern classification

Viticulture is a key economic activity for many countries around the world, and temperature is one of the most important parameters in grapevine response. Many studies have shown that variability of temperature at the vineyard scale has a significant effect on physiological development of the grapevine and, ultimately, wine quality. The Weather Research and Forecasting (WRF) model has been widely used to dynamically downscale from the synoptic and larger scale atmospheric circulation in order to provide a high-resolution analysis of weather and climate in regions of complex terrain. The temperature bias of the WRF model in a vineyard region is analysed using 18 automatic weather stations in order to understand the spatial variation in the bias due to local conditions. The WRF-predicted temperatures exhibited an average bias that was relatively consistent between measurement sites, although the consistency of this bias was found to vary in relation to weather type, time of day and season. These factors therefore need to be taken into account in order to properly correct temperatures produced by the WRF model.