Evaluating Reanalysis - Independent Observations and Observation Independence

Wednesday, 17 December 2014: 11:50 AM
Sabrina Wahl1,2, Christoph Bollmeyer1,2, Christopher Danek1,2, Petra Friederichs1,2, Jan Dominik Keller2,3 and Christian Ohlwein1,2, (1)University of Bonn, Meteorological Institute, Bonn, Germany, (2)Hans Ertel Centre for Weather Research, Climate Monitoring Branch, Bonn, Germany, (3)Deutscher Wetterdienst Nieder, Offenbach, Germany
Reanalyses on global to regional scales are widely used for validation of meteorological or hydrological models and for many climate applications. However, the evaluation of the reanalyses itself is still a crucial task. A major challenge is the lack of independent observations, since most of the available observational data is already included, e. g. by the data assimilation scheme. Here, we focus on the evaluation of dynamical reanalyses which are obtained by using numerical weather prediction models with a fixed data assimilation scheme.

Precipitation is generally not assimilated in dynamical reanalyses (except for e.g. latent heat nudging) and thereby provides valuable data for the evaluation of reanalysis. Since precipitation results from the complex dynamical and microphysical atmospheric processes, an accurate representation of precipitation is often used as an indicator for a good model performance. Here, we use independent observations of daily precipitation accumulations from European rain gauges (E-OBS) of the years 2008 and 2009 for the intercomparison of various regional reanalyses products for the European CORDEX domain (Hirlam reanalysis at 0.2°, Metoffice UM reanalysis at 0.11°, COSMO reanalysis at 0.055°). This allows for assessing the benefits of increased horizontal resolution compared to global reanalyses. Furthermore, the effect of latent heat nudging (assimilation of radar-derived rain rates) is investigated using an experimental setup of the COSMO reanalysis with 6km and 2km resolution for summer 2011.

Further, we present an observation independent evaluation based on kinetic energy spectra. Such spectra should follow a k-3 dependence of the wave number k for the larger scale, and a k-5/3 dependence on the mesoscale. We compare the spectra of the aforementioned regional reanalyses in order to investigate the general capability of the reanalyses to resolve events on the mesoscale (e.g. effective resolution).

The intercomparison and evaluation of regional reanalyses is carried out by the climate monitoring branch of the Hans-Ertel-Centre for Weather Research.