Updated monthly homogenized precipitation analysis HOMPRA Europe 2 online

Reliable data is essential for robust precipitation trend analyses. However, long time series in particular often contain artificial changes due to changes in the measurement conditions (e.g. instrument changes or relocations). These can lead to an artificial signal in the data and mask the actual changes in the time series.


In order to perform long-term precipitation change analyses, the monthly grid product HOMPRA Europe 2 (HOMogenized PRecipitation Analysis of European in-situ data) was developed at the Global Precipitation Climatology Centre (GPCC) and is available for the period 1951-2015 with 0.5°, 1.0° and 2.5° spatial resolution. The database consists of monthly time series carefully selected from the GPCC's data archive with less than 20% missing values. All data have already passed the GPCC's semi-automatic quality control. Compared to its predecessor, the number of incoming stations has increased from 5373 to 7916 and the data set has been extended from 2005 to 2015. The selected time series have been homogenized. This process recognizes and corrects the artificial influences as far as possible while retaining the natural changes.


The actual algorithm is essentially based on three steps (Rustemeier et al., 2017):


• Selection of overlapping station networks located in the same precipitation regime, based on rank correlation and Ward's method of minimum variance. 

• The natural variability and trends were removed in time by highly correlated neighboring time series in order to detect artificial breaks in the annual totals. 

• In the final step, the detected breaks are corrected monthly using multiple linear regression. 


The final data set HOMPRA Europe 2 (DOI: 10.5676/DWD_GPCC/HOMPRA_EU_M_V2_050) is available on the GPCC homepage (gpcc.dwd.de).