Atmosphere 2022, 13, 1634
3 of 21
in westerlies goes along with a decrease in easterly flows, especially in winter [21,25,26].
Future changes in storminess on the other hand are still subject to great uncertainties.
There is a considerable number of studies focusing on positive storm surges in the
North Sea area [27,28]. NSSs on the other hand are more common at the German Baltic
Sea coast (e.g., [5]). To our knowledge, there are no preceding investigations on NSSs in
the Elbe estuary, despite its high relevance for transport due to the connection to the Port
of Hamburg.
This study aims at addressing this gap by analysing the large-scale meteorological
conditions inducing past NSSs in the Elbe estuary and assessing how far the frequency
of these meteorological conditions may change in a potential future climate. The water
levels along the Elbe estuary are strongly influenced on the one hand by SLR and river
runoff into that region and on the other hand by the large-scale atmospheric circulation
and the associated wind field. It is obvious that under continued global warming, SLR will
become a non-negligible factor in the context of NSS and resulting ELWs. We address this
issue by additionally including a sensitivity study based on a hydrodynamic-numerical
simulation to investigate the influence of possible SLR scenarios on ELW caused by a
particular NSS event.
The following data and methods used in this paper are described in Section 2. Section 3
presents an analysis of past extreme low waters as well as a statistical analysis of the
mean meteorological conditions before ELWs, and shows first results for possible future
developments derived from an ensemble of global climate model simulations. It also
contains the results of the hydrodynamic sensitivity study. These results are followed by a
discussion (Section 4) and conclusions (Section 5).
2. Data and Methods
2.1. Data
2.1.1. Observational Data from Gauge Stations
As a fundamental data set for this investigation, we use observations of the low
water (LW) at the gauges Cuxhaven and St. Pauli for the period 1950-2019 which are
provided and checked for plausibility by the German Federal Waterways and Shipping
Administration (WSV) [29]. Based on the data at Cuxhaven, we develop a definition of
ELW (see Section 2.2.1). The points in time of LW levels matching our definition of an ELW
form the basis for the meteorological analyses. This data set is also used to analyse and
visualize the long-term development of the median and minimum LW of each year at the
two gauges Cuxhaven and St Pauli.
2.1.2. Atmospheric Reanalysis Data
For the analysis of past meteorological conditions related to the identified ELWs,
we use the ERA5 reanalysis which is the most recent reanalysis dataset produced by the
European Centre for Medium-Range Weather Forecasts [30,31]. As such, it is one of the
most modern and advanced global reanalyses, using the global weather forecasting system
IFS. The ERA5 data are available on a high-resolution grid with a spacing of 0.25° and
ın hourly resolution for the period from 1979 to the present and are updated at regular
intervals. To additionally cover the period before 1978 we included the ERA5 backward
extension [32]. The only variable considered here is the sea level pressure (SLP) which is
an instantaneous output, processed as follows for this investigation: (1) for daily mean
values, an average is calculated of the hourly values from 0UTC to 23UTC; (2) for a better
‚epresentation of the weather conditions leading up to the event, a 24-h mean is calculated
as an average over the 24 h directly before the event, including the hour of the event itself.
In a comparison of several reanalyses with observational SLP data over the North
Sea arca [33] it was shown that the ERA reanalyses (ERA40, ERA-Interim) performed best,
except for mixed sea/land grid points. Since to our knowledge there are no consistent ob-
servational data sets for both land and sea, we assume that the ERA reanalyses give the best
cepresentation in our research area, especially since the higher spatial resolution compared