Provision of design temperatures

Abstract

To design a building adapted to local climate requires a number of different climate indicators, one of them is design temperatures (DUT) for summer and winter.  The classic definition of summer design temperature was the maximum temperature exceeded 50 hours a typical year, and for winter the coldest three day average temperature. Looking into different descriptions of the DUT, there are distinct discrepancies. One example is for DUT-winter, where some instances describe the three coldest consecutive days and others the DUT-winter as a return period (e.g. 30 year) based on observations, similar for DUT-summer where some define it as 50 consecutive hours, others as individual hours summed together. The above uncertainty of definition is combined with the uncertainty of representation. The classic construction of DUT is based on observations from a representative station, interpolated to e.g. the municipality of interest. This method opens for the uncertainty of representation of the observational site and correctness for the interpolation.  Standards Norway contacted MET Norway to update the values of the DUT summer and winter in Norway to be calculated for the latest normal period, 1991 - 2020. In this work a new method to calculate DUT summer and winter was proposed and accepted: Instead of using single observational sites as a base, national climate grids calculated on a daily basis at MET Norway covering the entire country with a 1x1 km resolution is used as a basis Instead of a single temperature representing e.g. a single day is a statistical based approach applied. The method that was selected was a Bayesian-GEV approach where the output was calculated for 1 to 5 days average for highest and lowest mean temperature, with return values for 2-200 years.  This new approach resolves partly the challenge of representativity and interpolation by using robust and well documented spatial interpolation. The statistical approach also provides more well documented and robust statistics than the older approach. This approach creates a challenge since the new datasets represent something different than the older approach, and thus challenges the standard built on these datasets.  The distribution of the datasets will be renewed. Previously one had to buy the datasets from e.g. Standards Norway or other commercial vendors, the new dataset will be distributed openly from MET Norway, and implemented in e.g. the API frost.met.no  Besides the DUT, an extended information package containing daily temperature range and absolute humidity is calculated based on representative stations.