Short summary

The overall objective of STOWASUS-2100 is to study severe storms, surges and waves in the present climate and in a scenario with increased CO₂-concentration. More specifically the project is a joint atmospheric/oceanographic numerical modelling effort aiming at constructing and analysing storm, wave and surge climatologies for the North Atlantic/European region in a climate forced by increasing amounts of greenhouse gases and to compare with present day conditions. It is investigated whether any systematic anomalies regarding frequency, intensity or area of occurrence are found for these extreme events. Also physical mechanisms responsible for possible scenario anomalies are investigated.

Off-shore industries, fisheries, shipping companies, and the insurance business are highly sensitive to extra-tropical strong wind events and the associated ocean waves and surges. It is likely that impacts of possible future changes in the occurrence of extreme type events like these and others will be more severe than modulations of the long term mean climate. This is the rationale behind the STOWASUS-2100 project which aims at setting up climate change scenarios for storms, waves and surges on a variety of spatial scales. On the larger scales, studies on storminess in the North Atlantic region will be performed, while detailed studies on storminess, surges and waves will be carried out in the Adriatic, The North Sea and the Norwegian Sea. On the local scales, storms and surges will be studied in estuaries, low lying coastal areas along the North Mediterranean and North-western European coasts. 

The project builds to considerable degree on the results obtained in another project called WASA, which has been described by "The WASA Group" (1998). In WASA it was found that the storm and wave climate has roughened in recent decades, but that the present intensity of the storm and wave climate seems comparable with that at the beginning of the 20th century. The WASA project furthermore analysed and used the output from a high-resolution (T106 spectral truncation) climate change scenario experiment, mimicking global warming due to increase greenhouse gas concentrations. It was found that storm and extreme wave activity was slightly increased in the Bay of Biscay and in the North Sea in a warmer climate, while this activity was slightly weakened at several other places. The experimental set-up of the climate model simulations on which these results were based has been described by Beersma et al. (1997) who pointed out that the projected anthropogenic changes in storm activity fall well within the limits of variability observed in the past considering the length of the (control and scenario) simulations which was only 5-years.

Recently, two so called time slice simulations with the ECHAM4 model also at T106 horizontal resolution have been performed at the Danish Meteorological Institute (DMI) in a collaboration between the Max Planck Institute for Meteorology in Hamburg and DMI. These simulations each covered a period of 30 years, i.e. 6 times longer than the simulations used in WASA. Thus they should be much more suited for studies of storminess and associated impacts since the sampling problem is considerably reduced. The STOWASUS-project therefore uses these new simulations as a backbone to drive very high resolution regional atmospheric climate models and wave and surge models of different resolution. It is these secondary simulations which will be used to set up climate change scenarios of storms, waves and surges along European shelf and in European Estuaries and to compare these with present day conditions. The project is logically divided into 12 working tasks some of which will be described briefly in the following three sections together with preliminary results - to the extend they are available at this early state of the project.

As mentioned above the backbone in STOWASUS-2100 consists of two 30 year time slice simulations with the ECHAM4 atmospheric climate model at T106 horizontal resolution. The experimental design of these simulations which are not part of the project is described in May and Roeckner (1998). The project includes an analysis of the storm and extreme wind climate in these simulations. Fig. 1 (left column) shows the long term mean sea level pressure (MSLP) in winter (DJF) as obtained from the European Re-Analysis data (ERA), from the control simulation and from the scenario simulation. It is seen, that the ECHAM4 model simulates the atmospheric mass field well except for a too high pressure over and immediately to the west of the Iberian peninsula which leads to a too zonal flow over NE Atlantic region. The figure also shows the difference in the MSLP between the scenario and control simulations in the bottom panel, and it is seen that there is a significant increase in the zonality over the northern part of the area associated with a decrease in the high latitude MSLP in the scenario. The right column in Fig. 1 similarly shows the standard deviation of the band pass filtered (2.5-6 day filter) 500 hPa winter height fields which is commonly used as an estimate of storm activity. Also here the model behaves well and a significant increase in storm activity is seen over Northern Europe in the scenario simulation relative to the control simulation together with a corresponding decrease in storm activity around the east coast of US. Fig.2 shows the 1% percentile wind speed for the winters (DJF). Comparing the ERA data and the control run shows that the ECHAM4 model has more severe storms along the south and east coast of Greenland than the ERA data. The difference between the scenario run compared to the control run (fig. 2d) shows more severe storms in the Atlantic north of 60N, and less severe storms south of this latitude. These changes are in accordance with the changes in the 500 hPa variability (fig. 1d). The changes in near surface wind (Fig. 2) are important to the wave and surge simulations (see below) as one can expect that the enhanced wind speeds will also lead to more severe wave and surge activity. This has, however, not yet been shown in the project.

The atmospheric investigations will also cover atmospheric modelling with very high resolution regional climate models (HIRHAM and BOLAM) to perform case studies of intensive systems that are not well resolved at T106 resolution: intensive extra-tropical baroclinic developments, polar lows and highly convective systems (with some apparent similarities to polar lows) in the Mediterranean. Fig. 3 shows the orography in the T106 model and in the BOLAM model over the Mediterranean region and it is seen that one may expect much larger impact from orographic effects at very high resolution than at T106. All the simulations with HIRHAM and BOLAM will take boundary conditions from the T106 time slice simulations and will to considerable degree focus on analysing and understanding the processes associated with possible changes in scenario cases relative to control cases.

A main task is to calculate surge statistics based on the full 30 year T106 time slices for the North-western European shelf seas. A new surge model covering the North Eastern Atlantic and European shelf has been set up for this purpose. The model (NEAC) has a resolution of ~35km, and 26 tidal constituents input at the open boundaries. 4 runs have been completed with the model so far: two with tide only and two with both tide and surge. The meteorological inputs to the latter two runs has been DNMI data from 1955 to 1997 ("present day") and output from the ECHAM4 control run, respectively.

Validation of the model runs has been undertaken by comparison with the runs from a former model (CSX) which have previously compared well with observations – see Flather et al., 1998. The 10 largest surges have been extracted for each year of the "present day" and "control" runs of NEAC based on the "r largest" method (Tawn, 1988). A Gumbel distribution has been fitted and excedence probabilities have been derived. The 50-year surge elevations for the 43-year present day and the 30-year control period are compared in Fig. 4, and it is seen that the "control" run represents the "present day" surge elevation distribution very well over much of the Shelf with deviations occurring in coastal locations. This is an encouraging result as in a previous EU project, WASA, "control" forcing data from another climate model experiment (ECHAM3, 5-year time slice) produced surges which were underestimated by ~60% (Flather and Smith, 1998).

The project will also include a long simulation of surges for the Adriatic Sea. Here the relatively coarse topography in the ECHAM4 model constitutes a major problem, since the highly non-geostrophic channelling wind effect responsible for large surge and wave heights - as illustrated in Fig. 5 - can not be well simulated. Therefore downscaling will be needed for these simulations. In addition to the long simulations a number of case studies will be performed. The purpose of this is to study climate change of the extreme surges in the Adriatic and to investigate if the high resolution atmospheric forcing provides a needed more realistic forcing.

Investigation of local impacts - e.g. in the low lying areas along the Dutch coasts - calls for very high resolution models. Such models are very expensive from a computational point of view and therefore a combined dynamical/statistical downscaling technique will be used. By running 20 to 30 storm cases with atmospheric forcing taken from ERA, the two T106-experiments and severe hypothetical ones, statistical relationships can be derived between the surge levels at deep water and at the location. These relationships are finally used to downscale the surges simulated in the two 30 year time slice surge simulations mentioned above.

Long relative coarse mesh simulations and case studies at high resolution will be performed with wave models over the Mediterranean as well as over the North East Atlantic region. The output from these will be analysed and used for setting up scenarios of changing wave climate. The case studies will to a large degree be for the same cases which are used for surge investigations (see above) and for the same reasons special treatment with downscaling is needed for the Adriatic Sea.

Included in the wave studies is development and application of a statistical model (`wave generator'), that describes the wave statistics at selected locations in the North Atlantic, North Sea and Adriatic conditioned on the atmospheric conditions subject to changing CO₂ concentrations. This model will be developed using the long original T106 atmospheric simulations and the corresponding (modelled) wave heights. So far, the model was developed using observed daily MSLP and daily wave heights at one selected grid point in the central North Sea from a 40-year hindcast performed in the WASA project. To describe the daily wave climate a first-order autoregressive model for the monthly anomalies of de-seasonalized heights was used (see Fig. 6). A single autocorrelation parameter was fitted from the whole data set of wave heights whereas the variance of the model was fitted conditional on the seasonal cycle and the intramonthly variability of SLP. A model for the monthly mean wave heights was then derived using a multivariate regression (CCA) on monthly SLP over the Atlantic region. The wave generator has been successfully validated against the observed wave statistics (not shown).

This research was supported by the "Environment and Climate Programme" under contract number ENV4-CT97-0498.

Eigil Kaas and Uffe Andersen are at the Danish Meteorological Institute, Lyngbyvej 100, DK-2100 Copenhagen, Denmark, e-mails: ek@dmi.dk and ua@dmi.dk

Roger Flather and Jane Smith are at Proudman Oceanographic Laboratory (POL), Bidston Observatory, Birkenhead, Merseyside L43 7RA, UK, e-mails: raf@pol.ac.uk and jane@pol.ac.uk

Piero Lionello is at University of Padua (UP), Via VIII Febbraio 2, 35100 Padova, Italy, e-mail: piero@borexo.pd.infn.it

Piero Malguzzi is at FISBAT-CNR, Via Gobetti 101, 40126 Bologna, Italy, e-mail: malguzzi@ocean.fisbat.bo.cnr.it

Arnt Pfizenmayer and Reiner Schnur are at GKSS Research Centre, Institute of Hydrophysics, Max-Planck-Str. 1, D-21502 Geesthacht, Germany, e-mails: pfiz@gkss.de and schnur@gkss.de

John de Ronde, Marc Philippart and Stephanie Holterman are at National Institute for Coastal and Marine Management/RIKZ, P.O.Box 20907, 2500 EX Den Haag, The Netherlands, e-mails: J.G.dRonde@rikz.rws.minvenw.nl, m.e.philippart@rikz.rws.minvenw.nl and S.R.Holterman@rikz.rws.minvenw.nl

Magnar Reistas and Knut Helge Midtbø are at The Norwegian Meteorological Institute (DNMI), Allegt. 70, N-5007 Bergen, Norway, e-mails: magnar.reistad@dnmi.no and k.h.midtbo@dnmi.no

Beersma, J., K. Rider, G. Komen, E. Kaas and V. Kharin, 1996: An analysis of extra-tropical storms in the North Atlantic region as simulated in a control and 2x CO2 time-slice experiment with a high resolution atmospheric model. Tellus, 49A, 347-361.

Flather, R.A. and Smith, J.A. 1998. First estimates of changes in extreme storm surge elevation due to doubling CO₂. Global Atmosphere and Ocean System. In press.

Flather, R.A., Smith, J.A., Richards, J.D., Bell, C. & Blackman, D.L. 1998. Direct estimates of extreme storm surge elevations from a 40-year numerical model simulation and from observations. Global Atmosphere and Ocean System. In Press.

May, W. and E. Roeckner, 1998: An enhanced resolution modelling study on anthropogenic climate change (ERACC). The present volume p. ??.

Tawn, J. A. 1988. An extreme value theory model for dependent observations. Journal of Hydrology, 101, 227-250.

The WASA Group, 1998: Changing waves and storms in the Northeast Atlantic? B.A.M.S., 79, 5, 741-760.