Aim: Climate is arguably one of the most important factors determining the quality of wine from any given grapevine variety. High rainfall during spring can promote growth of the vines but increases the risk of fungal disease, while vineyard operations can be disrupted, as machinery may be prevented from getting in the vineyard owing to muddy soils. Conversely, high rainfall during harvest time (August to October) also bears the potential for severe operational disruption and heavy economic losses. To date, the probability of unprecedented rainfall amounts in spring and the harvest season has not been assessed over northern Portugal, specifically the three wine-growing regions of Vinho Verde, Trás-os-Montes and Porto and Douro DOC. In a situation of higher climatic variability, establishing the probable limits of rainfall variation during critical moments of the vine growth cycle will allow for better readiness of farmers as well as higher resilience of the whole value chain.
Methods and Results: Observed rainfall totals for northern Portugal were extracted from version 21 of the E-OBS dataset. Monthly rainfall totals were archived from a series of 16 month-long hindcasts produced with the Met Office’s decadal prediction system DePreSys3. These hindcasts begin in November of each year, corresponding to the start of each viticultural campaign. The hindcasts are produced from 1980 to 2017, when satellite data are available for model initialisation. Forty ensemble members are available for each start time, providing 1520 (38 × 40) simulations of spring and late summer rainfall totals. The hindcast and observed rainfall totals are considered indistinguishable if the mean, standard deviation, skewness and kurtosis from the observations are within the respective 2.5th–97.5th percentile ranges from 10,000 model bootstraps. It was necessary to shift the modelled mean for spring rainfall owing to a wet bias in the simulations. The model results showed there was a probability of 0.02 ± 0.01 of an unprecedented rainfall event in spring and summer. However, the chance of another year with an exceptionally wet spring and late summer (as happened in 1993) is extremely small.
Conclusions: Rainfall totals in northern Portugal over the past 38 years have been very high in a few years, but higher values are possible in the current climate. The chance of another year similar to 1993, when both seasons were exceptionally wet, is very low. The uncertainty in extreme rainfall estimates is considerably reduced when the modelled data are used. A year with rainfall equal to the highest observed amounts in one of these two seasons could be expected to occur just once in the next 30-100 years.
Significance and Impact of the Study: This study is the first to assess the probability of unprecedented rainfall extremes over northern Portugal, allowing for a better estimate of the inherent risk. The results help inform the need for costly adaptation investments, such as better availability of spraying machinery and labour, high-gauge drainage, landslide controls or even abandonment of exposed vineyard areas.
Authors: Michael G. Sanderson1*, Marta Teixeira2, Natacha Fontes2, Sara Silva2, António Graça2
1Met Office, Fitzroy Road, Exeter, EX1 3PB, United Kingdom
2Sogrape Vinhos, S.A., Aldeia nova, 4430-809 Avintes, Portugal
Email: michael.sanderson@metoffice.gov.uk