Multivariate testing of spatio-temporal consistence of daily precipitation records

Abstract. The project KLIDADIGI of the German Meteorological Service (DWD) systematically rescues historical daily climate data of Germany by keying and imaging. Up to now, daily nearly gap-free precipitation time series at 118 locations for the period 1901–2000 are collected and extended by digitalization of hand-written protocols. To screen the spatio-temporal consistence of these raw data, we apply principal component analysis (PCA) in S (spatial) mode for daily precipitation records as well as for indices such as the number of rainy days above a certain threshold, intensity and absolute daily maximum in monthly, seasonal or annual resolution. Results of this screening test indicate that the PCA is a useful tool for detection of questionable stations and data preprocessing for further quality control and homogenization.