Frost days

Description of the results

The three altitude bands analyzed all show a clear trend: frost days have decreased since 1980. The calculated trends are statistically significant for all bands, and the decrease is largest in absolute terms for altitudes above 1000 m, with a decrease of about 10 days per decade. In this range, the annual number of frost days has decreased by about 40 days between 1980 and 2022. Below 1000 m, the calculated decrease over the 43 years is about 8 days per decade. Below 500 m, the number of frost days has decreased by about 32 days.

Method 

The illustrations are based on the daily meteorological observations of more than 80 measuring points of the Office for Meteorology and Avalanche Warning of the Autonomous Province of Bolzano, supplemented by the observation data of some sites in Switzerland and Austria close to the national border. The collected series were interpolated to a regular grid with a resolution of 1 km for the whole national territory using a geostatistical method.

Before interpolation, all observation series were checked for measurement errors and temporal homogeneity. In addition, missing daily values were reconstructed using a statistical procedure to maximize the temporal continuity of the series.

The interpolation allows a regional mean to be obtained which is more representative and stable than that based on individual stations. Trends are calculated using the Theil-Sen method and significance is determined using the Mann-Kendall test. The trend is considered significant if the resulting p-value is less than 0.05.

Sectors affected 

• Snow and Ice 

• Water 

• Flora and Fauna 

• Soil 

• Natural hazard 

• Ecosystem service 

• Agriculture

• Forestry

• Tourism

• Traffic infrastructure  

Related Indicators 

Future scenarios

According to available climate projections, the number of annual frost days will continue to decrease in the coming decades. The largest decrease in frost days is observed in the most pessimistic climate scenario, RCP 8.5. According to this scenario, frost days at altitudes between 1000 and 2000 m could fall from an average of 190 days in the period 1981 to 2010 to 100 days in 2100. According to the same scenario, an average of 10 frost days per year are expected below 500 m by 2100, a decrease of about 80 days compared to reference conditions (1981 to 2010). According to the RCP 4.5 scenario, the decrease is more modest at all altitudes and slows down from the second half of the century onwards.

The spatial distribution of annual frost days, averaged over 30-year intervals in the reference period betwee 1981 to 2010, mid-century (2041 to 2070) and at the end of the century (2071 to 2100), shows a significant decrease in the lower areas of the Etsch/Adige and Vinschgau/Venosta valleys, with less than 25 frost days per year at the end of the century (RCP 8.5 scenario). Areas with frost for most of the year are also becoming increasingly rare compared to the historical reference period, and by the end of the century, according to the RCP 8.5 scenario, they will only exist at altitudes above 3000 meters.

Method  

The climate scenarios for annual and seasonal precipitation for South Tyrol were derived from EURO-CORDEX climate simulations throughout Europe for the two emission scenarios RCP 4.5 and RCP 8.5. RCP stands for “Representative Concentration Pathways“, i.e. projections of how greenhouse gas emissions in the atmosphere will develop in the future.

RCP 4.5 is an intermediate scenario in which greenhouse gas emissions are curbed, but atmospheric concentrations continue to rise over the next 50 years and the +2°C target is not met. RCP 8.5 represents the most pessimistic scenario, in which greenhouse gas emissions continue to increase and no action is taken to combat climate change.

The daily minimum temperature projections from 1971 to 2100 provided by 15 different climate models (ensembles) for the two scenarios were further processed using a downscaling procedure that allows the simulated values to be transferred from the original spatial resolution (in this case around 12 km) to a finer resolution (in this case 1 km). This step makes it possible to reduce systematic errors that occur in the model simulations due to the limited spatial resolution of the available models, which do not provide an adequate representation of local features, especially in mountainous regions with complex orography. The downscaling method used is based on the delta-quantile mapping method (QDM, Cannon et al., 2015), in which simulated values are compared with observations over a common reference period and corrected so that the probability distributions match. In addition, the QDM method makes the corrections in such a way that the long-term climate signal originally present in the simulations is not altered.

In this case, the reference period is 1981- to 2010 and the correction was made on the basis of the 1 km gridded observation data set.

From the corrected simulations, the annual frost days from 1971 to 2100 were calculated for each model in the ensemble and for both scenarios. The values of the indicator for the ensemble were then aggregated by calculating the median of the 15 model simulations for each year and using the interquartile range, i.e. the range of values between the 25th and 75th percentile, to obtain an estimate of the variability of the model simulations.

Contact

Eurac Research: Alice Crespi, Center for Climate Change and Transformation

Data provided by Weather and avalance service of the Autonomous Province of Bolzano