MeteoSwiss contribution – Comparison of climate datasets
Work package 4.2 (led by MeteoSwiss) in C3Surf aims at evaluating the E-OBS dataset, which has been produced in the framework of ECA&D, and subsequently updated in the framework of this C3SURF project. Given that the new global reanalysis from ECWMF, ERA5, was simultaneously released and thanks to a similar grid resolution it is also examined. In order to evaluate them, both datasets are being compared to regional high-resolution grid datasets from three European regions (Alps, Carpathians and Fennoscandia). Not only are these mountainous regions interesting from a climatological perspective, but there are also high-quality observational datasets available covering these regions. Thanks to including the densest possible station network and locally adjusted methods, they act as a reference in this evaluation. In the Alpine region, only precipitation was examined (with the reference being the Alpine Precipitation Grid Dataset (APGD)), whereas in the Carpathian region (CARPATCLIM) and in Fennoscandia (Norwegian Gridded Climate Datasets (NGCD)) both, precipitation and temperature, were investigated.
The results of the evaluation include the calculation of climate indices (e.g mean annual precipitation, see Figure 2), correspondence measures and an uncertainty evaluation. Based on these results the potential, as well as the limitations of E-OBS and ERA5 were determined. The most dominant finding for the precipitation analysis is that E-OBS strongly depends on the data availability in a certain area. While E-OBS is able to show realistic meso-beta scale precipitation patterns over Germany, Slovenia, Slovakia and over most of Fennoscandia, it performs worse especially over Romania, Ukraine, Croatia, Northern Italy, parts of France, Switzerland and Austria. Overall, this manifests in an underestimation of precipitation sums. ERA5, on the other hand, generally tends to overestimate precipitation, but it realistically resolves most large-scale precipitation patterns. Additionally, both ensemble datasets, E-OBS-Ens and ERA5-Ens underestimate the uncertainty compared to APGD-Ens. Nevertheless, both datasets have significantly improved compared to their older versions (older E-OBS and ERA-interim).
Regarding the temperature analysis in Fennoscandia and the Carpathians, E-OBS shows larger deviations in complex topography of Fennoscandia due to the absence of locally adapted methods, but it generally performs well in the Carpathians. Considering ERA5, most notably the Diurnal Temperature Range (Tmax – Tmin) is underestimated due to smoother temperature fields.