HOMPRA Europe 2 – An update of a gridded precipitation data set from European homogenized time series

Reliable observational data are essential for robust climate analyses, especially long-term trends.

HOMPRA Europe 2 (HOMogenized PRecipitation Analysis of European in-situ data) is a gridded monthly precipitation dataset based on homogenized time-series. The carefully selected database of more than 5500 stations is a subset of the time series collected by the Global Precipitation Climatology Centre (GPCC). The subset is characterized by few missing values < 20% and intensive quality control.

Compared to its predecessor, the new HOMPRA Europe 2 product covers a longer period from 1951-2015 and is syncronized with the more recent data of the GPCC Full Data Monthly product, so that the data can be extended to the present without a break, although a homogeneous time series cannot be guaranteed for these more recent data.

The actual homogenization process consists of multiple steps: In the first step, for each station, comparable time series are selected. The decision is based on the correlation and Ward's method of minimum variance performed on the deterministic first derivative. For the artificial breakpoint detection, natural variability and natural trends are temporarily removed using the previously selected neighboring time series. This ensures that only artificial changes can be detected. The actual detection applied to annual values is based on the segmentation algorithm of Caussinus and Mestre (2004). In the final step, the breaks detected in the monthly data are corrected using multiple linear regression (Mestre, 2003).

An additional random variation of the neighboring stations shows the robustness of the correction of the individual time series.

Finally, the actual HOMPRA Europe 2 product is created by interpolating the homogenized series onto a 1° grid using the modified spheremap interpolation schemes in operation at the GPCC (Becker et al., 2013 and Schamm et al., 2014).