EUSTACE

EU Surface Temperature for All Corners of Earth

Eustace is a project funded by the European Union's Horizon 2020 Programme for Research and Innovation and aims at creating and giving publicly available daily estimates of surface air temperature since 1850 across the globe, by combining surface and satellite data using novel statistical techniques.

KNMI is participating in Work Package 1, that has the task to provide a homogenized dataset of ground-based measurements. The European Climate Assessment & Dataset (ECA&D), maintained by KNMI, has a fundamental role in this part of the project since it collects thousands of temperature series from all over Europe starting in the 18th century (details available at http:// www.ecad.eu).

In particular KNMI has been involved in the development of an homogenization process for ECA&D series in cooperation with the Department of Geography of the University of Bern. Different homogenization methods have been tested on the temperature series. A particular focus has been given to the blended series, which are the result of the merging of series coming from same or neighboring stations. The plots below represent the different effects of the tested methods on two sample series from Groningen (164) and Geneva (240), whose series have been moved to nearby airports in 1952 and 1962, respectively. Black lines represent the original series, red lines represent the series homogenized with the method from Vincent et al. (2001) (adjustments through calculation of a seasonal cycle) and blue lines represent the series homogenized with the method from Trewin et al. (2012) (quantile matching of temperature distributions). This last method can be applied on series having an overlapping period, this is the case of the green line in the plot on Groningen. These graphs show a clear improvement of the quality of the series, deleting part of the artificial signal due to the relocation of the stations. For these reasons the Trewin method has been applied to all the blended series in the ECA&D dataset, providing a higher spatial agreement among the trends in annual mean.

Further work will be done in the homogenization of the simple series, thanks to a break detection procedure developed in Bern, based on the method by Kuglistch et al. (2012).