Drivers of the increased impact of extreme spring frosts on European vineyards

A number of unprecedented extreme late spring frosts has recently hit Western Europe, especially France, causing severe impacts on vineyards that had already undergone bud burst. Whether climate change may have contributed to making these events more likely is poorly understood, since it depends on a subtle balance between how much the phenological cycle anticipates due to winter warming and how much cold spring nights warm. Here we shed more light on the key driving factors behind observed changes by taking a novel approach in which the different components of past warming trends, i.e. mean warming, seasonal cycle changes, diurnal temperature changes and frequency of cold snaps, are isolated and separately analysed. Bud burst date is determined using a previously tested phenological model based on growing degree days. A statistical model is introduced to modulate the different components of warming and generate thousands of sample of plausible realisations of internal climate variability. The statistical model is fit to daily mean and night minimum temperature data in 1950-2022 from the ERA5 reanalysis and EOBS.

 

We show that in large parts of Europe even a simple seasonally and daily uniform warming leads to an increase in the risk of frosts after vineyard bud bursts. However, such simple thermodynamic effect is small compared to the observed changes in the frequency of such events. Including the effect of changes in the seasonal cycle, i.e. the winter warming amplification, and in the diurnal temperature range further increases frost risk, but it still explains only up to a third of the observed changes in the frost events. Likewise, changes in internal atmospheric variability are also insufficient to explain the observed trends. We demonstrate that only the increased frequency in an intra-seasonal  atmospheric circulation pattern causing warm winters followed by spring cold snaps, and acting on a warmed seasonal cycle, can explain the observed trends. We then analyse CMIP6 historical simulations to attribute the role of climate change versus internal climate variability. While mean warming and seasonal cycle changes are strongly influenced by climate change, the winter to spring swing in atmospheric circulation is more difficult to attribute and it may result from an extreme realisation of internal variability. We suggest that storylines can be used to illustrate plausible future risks for vineyards depending on the rate of regional warming and the type of circulation changes.