Numerical weather prediction as a surrogate for climate observations in practical applications

Climate data is used in many practical applications including energy demand estimations for heating and cooling, agricultural applications, risk assessment, and many more. The required climate data is only available if meteorological observations exist at a given location. In this study, the possibility of replacing long observational records with a few years of numerical weather forecast data is investigated for practical applications requiring temperature data. Observational data from 1980-2010, measured at 700 weather stations in Central Europe are used together with model forecasts of the years 2008-2010. Depending on the station, forecast data capture 90-110% of the standard deviation observed for daily mean and maximum temperatures and slightly less for minimum temperature. Heating and cooling degree days can be estimated with an error of 5-15% in climates where they have a relevance. Based on model data, maps of heating and cooling degree days are computed and the regional uncertainties are quantified using the observational data. The results suggest that numerical weather forecast data can be used for certain practical applications, either as a surrogate of observational data or for quite reliable estimates in locations with no observations.