Regional Evaluation of ERA-40 Reanalysis Data with Marine Atmospheric Observations in the North Sea Area
Schade, Nils H.; Heinrich, Hartmut; Rosenhagen, Gudrun
An important task of the departmental research programme KLIWAS is the evaluation and assessment of climate model results by means of a comprehensive reference data set. For validation purposes, and to create a North Sea wide maritime atmospheric and oceanographic reference database, in-situ observations of the Centre for Global Marine Meteorological Observations (GZS) of the National Meteorological Service DWD have been compared to the ERA-40 reanalysis. ERA-40 is used as forcing for the hindcast runs of the ENSEMBLES regional climate models, which is used within the KLIWAS model chain. The GZS hosts a regularly updated, quality controlled, world-wide data bank of weather observations from the oceans. It includes data from all sorts of observation platforms as Voluntary Observing Ships (VOS), drifting and moored buoys, light vessels, and offshore platforms, either from real-time (RT) via the Global Telecommunication System (GTS) or from international exchange in delayed-mode (DM). In addition to the automated set of programs applied for high quality control, erroneous data are also manually corrected to a certain extent, if possible. To assure reliable statistics for the evaluation, the corrected observations are gridded to a resolution of 2.25 degree, so each grid box includes four ERA-40 reanalysis grid points. The temporal coverage of the grid boxes depends on shipping routes and the positions of automated systems. Observed air temperatures, covering a period of 40 years (1961?2000), show noticeable differences to the reanalysis data for all land influenced boxes, specifically in the winter months. The same differences can be found if ERA-40 data alone are compared between land- and sea facing boxes. They can not be found in GZS data. It can be assumed that the differences are not resulting from measurement errors or uncertain fraction variabilities, since they are small during the winter months. A comparison of the differences basing on the 1981?2000 period to those of the 1961?1980 period shows an increase. Further investigations reveal that differences are largest in the landward part of the grid boxes, whereas in the parts facing to the open sea, observations and ERA-40 are in fair agreement. This leads to the conclusion, that the resolution of the ERA-40 reanalysis is not sufficient for detailed analyses of air temperatures near the coasts. However, this problem does not occur for the sea level pressure. The described land influence does not interfere with parameters, that are more or less insensitive to the land-sea distribution.