676 
N.H. Schade et al.: Regional Evaluation of ERA-40 Reanalysis Data 
Meteorol. Z, 22, 2013 
Downloaded from www.schweizerbart.de 
Unauthorized distribution of this copyrighted material is strictly forbidden! 
ocean-atmosphere climatology in the North Sea and adja 
cent parts of the North Atlantic and the Baltic Sea has 
become an important task. The North Sea is of particular 
interest and importance for marine carrier operations and 
off-shore activities. Especially in this context, subsequent 
projects within the KLIWAS model chain investigate 
amongst others appropriate strategies for adaption to 
changed environmental conditions to safeguard the effi 
ciency of transport ways and to preserve the water quality 
and the habitats in coastal waters. Thus, the quality of 
reanalyses and operational analyses should be precisely 
checked and possible biases compared to in-situ observa 
tions should be addressed. 
For the validation of the RCM output over the North 
Sea area, high quality controlled surface marine in-situ 
observations and/or reanalyses data are needed as refer 
ence of the actual climate state. In-situ observations are 
the sole ground based measurements of atmospheric 
parameters over the open sea. They provide information 
that cannot be reliably measured from satellites (Kent 
and BERRY, 2005), e.g. surface air temperature and sea 
level pressure, although the data quality varies within 
the course of time. Several publications dealing with 
the popular International Comprehensive Ocean-Atmo 
sphere Data Set ICOADS (WORLEY et al., 2005; WOOD 
RUFF et al., 2011) are focusing on the ocean on a global 
scale. Some authors have specifically investigated the 
importance of ship log data in scientific research, also 
on a regional scale, e.g. KUETTEL et al. (2009) for the 
reconstruction of past sea level pressure fields derived 
from ship logbook wind data over the North Atlantic/ 
European area. Nevertheless, gridded data sets are mostly 
of an insufficient resolution for regional climate analyses 
and based on terrestrial observations, containing only a 
small number of island stations, if any (ALEXANDER 
et al, 2009; Jones et al., 1999). 
Here, we introduce the database of the Centre for Glo 
bal Marine Meteorological Observations, further referred 
to as GZS. This database is hosted by the German Mete 
orological Service DWD and aims to ensure the produc 
tion and supply of marine climatological data. 
For the evaluation of the ERA-40 data, areas were 
identified with a sufficient number and quality of obser 
vations in GZS. The investigated data in this study, cov 
ering a time period of 40 years beginning in 1961, are 
described in the following section 2. DWD’s high quality 
control procedures for the marine atmospheric observa 
tions are described briefly in section 2.1 and the data pre 
processing in section 2.2. Results of our investigations 
are presented in section 3, followed by a summary and 
conclusion in section 4. 
2 Data 
GZS is regularly updated with meteorological observa 
tions over the world’s oceans, reaching back to 1850. It 
contains the available data from all sorts of measurement 
platforms: Data from Voluntary Observing merchant 
Ships (VOS), research, naval and light vessels, buoys, 
and digitized data from historical weather journals 
(DWD project HISTOR, GLOEDEN, 2011) and registra 
tions. Altogether, GZS consists of more than 188 million 
real-time, distributed via the international Global Tele 
communication System (GTS), and non real-time weather 
reports. These data from national sources and bilateral or 
international exchange are complemented by data from 
ICOADS. ICOADS has an overall proportion of 
20,21% in GZS for the investigated period from 1961— 
2000, decreasing towards the end of the period 
(11.91% in the last decade). Since the proportion of 
GTS reports, inherent in both GZS and ICOADS, further 
increases, additional reports from ICOADS in GZS are 
now of about 3.5%. 
The ERA-40 Reanalyses (UPPALA et al., 2005) covers 
the time period from 1957 to 2002 and contains world 
wide meteorological parameters in 60 height levels on 
a 6 hourly basis. It is provided by the European Centre 
for Medium-Range Weather Forecast (ECMWF) in col 
laboration with other institutions, e.g. the MetOffice, 
Exeter, UK, or the Max-Planck-Institut fiir Meteorologie, 
Hamburg, Gemiany. Several data types are assimilated, 
i.e. operational real-time data distributed via GTS 
(1979-2002), observations from aircrafts, buoys, offshore 
platforms, radiosondes and ship data, taken from ICO 
ADS (1950-1999). Since 1973 satellite data are included 
also. The assimilation system uses an updated form of the 
older 3D variational analyses of the ECMWF with a 
spectral T159 resolution, corresponding to a 
1.125 x 1.125 degree geographical grid. The model has 
a reduced Gaussian grid of about 1.125 degree in latitude 
and a reduced number of longitudinal grid points from 
the Equator to the Poles. 
Each data set has its own particular problem: While 
GZS marine in-situ observations are unevenly distributed 
in space and time and are derived from several different 
observation types from different countries and with dif 
ferent methods for the measured parameters, the spatial 
resolution of the existing reanalyses data is coarse and 
the models suffer from biases in numerics and parameter- 
izations, inhomogeneities of data assimilation input, etc. 
In this context ERA-40 is used as an example for reanal 
yses data, since the KLIWAS model chain is based on the 
ENSEMBLES Hindcast runs which are forced by ERA- 
40. Further investigations with ERA-Interim and other 
reanalyses products are planned in the future. 
In the following, we describe the procedures to obtain 
GZS air temperatures and sea level pressures in the North 
Sea area of best possible quality to conduct reliable com 
parisons with ERA-40 reanalysis data. 
2.1 High Quality Control 
The high quality control (HQC) procedures used for 
GZS data are more or less the same as published in the