Modelling frost generates insights for managing risk of minimum temperature extremes

Frosts and extreme minimum temperature events cause significant damage to grapevines. In the Victorian vine growing regions of Australia, these events may cause complete wipe-out of a season's grape production, and millions of dollars of losses. Frost is an important risk to be managed by viticulturalists, and is likely to change under future climatic conditions with impacts on existing wine-grape growing regions. This case-study applies high spatial resolution analysis of minimum night-time temperatures to explore the impact of current and future frost risks. Multivariate Adaptive Regression Splines (MARS) are used to model high-resolution (30 m grid) minimum temperatures for a topographically diverse region in the Yarra Valley wine region in southeastern Australia. Remotely sensed Moderate Resolution Imaging Spectroradiometer (MODIS) night-time temperatures, elevation, terrain indices and meteorological station data were used to develop the model of minimum night-time temperatures. The model is formally validated against independent vineyard minimum temperature records (R² = 0.68) as well as current distribution of viticulture land-use, which was compared with viticulturists' perceptions of a frost-line in the region. Historical temperature records were then adjusted to produce high-resolution maps of frost occurrence under a hot-dry future climate scenario, and a warm-wet scenario, for the years 2030 and 2050. All the future climate scenarios project down-elevation movement of the frost line of between 10 m and 30 m depending on scenarios. This work has implications for viticulturists who might plan to expand growing areas to lower elevations, or might consider changing to different grape varieties. The method developed here could be applied to other regions or used to explore other future climate scenarios.