Günther Heinemann: Interaction of katabatic winds, mesocyclones and sea ice formation

Universität Trier, FB VI, Fach Umweltmeteorologie, 54296 Trier



Studies with the coupled atmospheric/sea ice model for the Angmassalik region of Greenland show that synoptically forced katabatic winds can result in a fast formation of a coastal polynia within 24 h. Sea ice advection is the main process during the first 12 h, but the production of frazil ice and the subsequent conversion to consolidated ice becomes important at later stages. Simulations for an scenario of an eastward moving cyclone being typical for piteraq events in the Angmassalik area reproduce the main features known from observational and realistic modeling studies, such as the development of the lee trough, its interaction with the katabatic wind and a development of a low-level mesoscale cyclone in the bay-like area southwest of Angmassalik. The simulations of the present studies were performed for the Greenland area, but because of their idealized setup the results are also applicable for the conditions of the Antarctic.

1. Introduction

The orography of the Greenland represents a barrier to the mean westerly flow. In addition, strong katabatic wind develop under cloud-free conditions over the ice sheet. The specific orography structures of the Greenland coast near Angmagssalik/Tasiilaq (see Fig.1) seems to favour the development of MCs (Klein and Heinemann, 2002), leading to an interaction of katabatic storms and MCs in this area. Besides the destructive potential of katabatic winds, a special point of interest is the modification of katabatically generated air flows when passing over the coastline and interacting with the sea ice or the open ocean. The formation of coastal polynias by the katabatic wind and the associated strong air-sea interaction processes have been considered to be relevant for the oceanic thermohaline circulation (e.g. Gordon und Comiso, 1988). A prominent example of a katabatically-driven coastal polynia is the Terra Nova Bay polynia in the Ross Sea of Antarctica, which is almost ice-free even during wintertime (Bromwich and Kurtz, 1984). The evidence of a polynia associated with channeled katabatic wind in the area of East Greenland is shown in Klein and Heinemann (2002). In the present study, forcing and feedback mechanisms between the katabatic wind, mesocyclone development and sea ice are investigated by idealized simulations with a non-hydrostatic 3D mesoscale model and a two-class sea ice model. The simulations are performed for  topographies for the Greenland area, but the results are of course valid also for the Antarctic without restriction.

2. Numerical models and design of the simulations

2.1 Atmospheric model

The limited area model used for this study is the non-hydrostatic model FOOT3DK (Flow Over Orographically structured Terrain 3 Dimensional, Kölner Version) developed at the University of Köln, which was adapted to the polar regions. The model has been used for applications in mid-latitudes and the tropics from mesoscale down to the microscale for simulating atmospheric flow and dispersion over complex topographies (e.g. Shao et al., 2001). In the present study, the model is run with a horizontal resolution of up to 60x60 grid points and a vertical resolution of up to 31 vertical levels (with 14 levels below 500 m). The vertical coordinate is terrain-following with a high resolution of the boundary layer. A one-way nesting procedure is used. A first run with 24 km resolution and a model domain of about 1200kmx1200km is performed (Fig.2), which is used as initial and boundary conditions for the model run with 12 km resolution. Details of the model and the nesting procedure can be found in Brücher (1997) and Sogalla and Kerschgens (2001). The main modification for polar regions was necessary in the land-surface scheme, since the treatment of soil moisture, which is not designed for ice and snow, was found to lead to unrealistic subsurface energy fluxes in the snow.

The initial state for all idealized simulations is an atmosphere with horizontally homogeneous fields of temperature and humidity, represented by the vertical profiles shown in Fig.3. Two different sets of profiles are chosen, each set representing a relative cold and a moderately cold situation. The profile of the relative humidity is the same for both profiles, except for one level (in order to reduce cloud formation for wintertime cases associated with strong convection). The temperature for the moderately cold situation is -8°C at 1000 hpa, which corresponds to typical springtime temperatures at the Greenland coast, but represents more summertime conditions in the Antarctic coastal areas. This profile is referred to as ‘summer profile‘. On the other hand, the temperature for the cold situation is -28°C at 1000 hpa, which represents wintertime conditions for the coastal areas of both ice sheets (‘winter profile‘). The simulations were started during the polar night at 00 UTC 1 January. In all simulations, katabatic winds develop after a few hours of simulation time.

2.2 Sea ice model

A sea ice model was implemented into FOOT3DK using similar parameterizations as in Gallée (1997) with some adaptions from Goodrick et al. (1998). The main differences to Gallée (1997) are that the ice momentum equation is formulated in flux form, and that the drag at the bottom of the ice is also parameterized. The model is a two prognostic two-class ice model for frazil ice and consolidated ice. The ocean is assumed to be at the freezing point (-1.8°C), and ocean areas are identical with frazil ice areas.

3. Results

Since in reality strong piteraq events in the Angmassalik area are associated with an eastward moving (rather than stationary) cyclone between Greenland and Iceland (Klein and Heinemann, 2002), a scenario of instationary synoptic forcing was investigated. This scenario of an eastward moving cyclone (CYC) is a geostrophic wind with 8 m/s during the first 6 h, with decreases to 2 m/s after 18 h and remains constant after that time. Accordingly, the wind direction changes from northerly to westerly directions between 0 and 12 h. The development of a lee trough, the katabatic wind and the initial polynia occurs during the first 12 h (Fig.4a). The eastward movement of the lee trough is small, and a low-level mesoscale cyclone is found in the bay-like area southwest of Angmassalik. The most intense winds in the southwestern part of this mesocyclone are associated with the katabatic flow of the a pronounced valley. This process of interaction between the synoptic environment, the formation of a lee trough and low-level vorticity generation by the channeled katabatic wind corresponds well to the simulation of realistic mesocyclones in that area (Klein and Heinemann, 2002). The mesocyclonic circulation is also reflected by the ice velocity vectors after 12 h (slightly deflected to the right with respect to the near-surface wind vectors, Fig.4b), and the initial polynia can be seen at that stage. The belt of frazil ice areas with up to 40% is also visible as warm anomalies of the surface potential temperature with values up to -10°C as the average over a grid box (Fig.3a). After 22 h simulation time (Fig.5), the width of the coastal polynia (here visualized again as warm anomalies of the surface potential temperature) has increased. The effect of the polynia on the katabatic wind near the coast is twofold: firstly the reduction of the consolidated ice results in a reduced surface roughness allowing the katabatic flow to extend over the coastline until it converges at the seaward side of the polynia where the roughness increases; secondly, the convection and warming of the boundary layer air over the polynia represent an additional thermal forcing on the katabatic wind.

4. Conclusions

An scenario of an eastward moving cyclone being typical for piteraq events in the Angmassalik area is investigated. The main features shown in observational and realistic modeling studies are reproduced, that is the development of the lee trough, its interaction with the katabatic wind and a development of a low-level mesoscale cyclone in the bay-like area southwest of Angmassalik. The orography of this area represents favourite conditions for channeled katabatic winds and mesocyclogenesis that are comparable to the Ross Sea region of Antarctica (Heinemann and Klein, 2003; Gallée, 1995). The present simulations with the coupled model show that an synoptically forced piteraq event of about 24 h duration can result in the formation of an initial coastal polynia, which in turn has a positive feedback on the katabatic wind intensity. The simulations of the present studies were performed for the Greenland area, but because of their idealized setup the results are also applicable for the conditions of the Antarctic.


Bromwich, D.H., Kurtz, D.D. (1984) Katabatic Wind Forcing of The Terra Nova Bay Polynya. Journal of Geophysical Research 89:  3561-3572.

Brücher, W. (1997) Numerische Studien zum Mehrfachnesting mit einem nicht-hydrostatischen Modell. Mitteilungen aus dem Institut für Geophysik und Meteorologie der Universität zu Köln 119, ISSN 0069-5882, 115pp.

Ekholm, S. (1996) A full coverage, high-resolution, topographic model of Greenland computed from a variety of digital elevation data. J. Geophys. Res. 101: 21961-21972. 

Gallée, H. (1995) Simulation of the mesocyclonic activity in the Ross Sea, Antarctica. Mon. Wea. Rev. 123: 2051-2069. 

Gallée, H. (1997) Air-sea interactions over Terra Nova Bay during winter: Simulation with a coupled atmosphere-polynya model. J. Geophys. Res. 102: 13835-13849.

Goodrick, S.L., McNider, R.T., Schroeder, W.W. (1998) On the interaction of the katabatic-land-sea wind system of Antarctica with the high latitude southern ocean. American Geophysical Union, Antarctic Research Series 75: 51-65.

Gordon, A. L., Comiso, J. C. (1988) Polynias in the Southern Ocean. Scientific American 258-6: 90-97.

Heinemann, G., Klein, T. (2003) Simulations of topographically forced mesocyclones in the Weddell Sea and the Ross Sea region of Antarctica. Mon. Wea. Rev. 131: 302-316.

Klein, T., Heinemann, G (2002) Interaction of katabatic winds and mesocyclones at the eastern coast of Greenland. Meteorological Applications 9: 407-422.

Shao, Y. Sogalla, M. Kerschgens, M.J., Brücher, W (2001) Treatment of land surface heterogeneity in a meso-scale atmospheric model. Meteorol. Atm Phys. 78: 157-181.

Sogalla, M., Kerschgens, M.J. (2001) Berechnung lokaler Niederschlagsfelder zur Parameterisierung der nassen Deposition auf der Basis größer-skaliger Vorhersagemodelle. Mitteilungen aus dem Institut für Geophysik und Meteorologie der Universität zu Köln, ISSN 0069-5882, 81pp

Figure 1: Map of Greenland with topography (isolines every 500 m) from high resolution (2 km) topography data (Ekholm, 1996). Triangles mark the radiosonde stations (abbreviations: Sco=Scoresbysund, Dan=Danmarkshavn, KG=Kangerdlugssuaq).
Figure 3: Initial profiles of potential temperature (solid lines) and relative humidity (dashed lines) for the idealized simulations (winter case, thin lines; summer case, thick line; see text)
Figure 4a: Simulations results for the coupled run for a synoptic forcing of an eastward moving synoptic cyclone, no polynia and 100% sea ice coverage at the initial stage after 12 h simulation time.Wind vectors at 15 m (vector scale in the lower left) and surface potential temperature (shaded)
Figure 4b: As part a, but for the frazil ice coverage (shaded and dotted isolines every 0.2) and ice velocity of the consolidated ice (vector scale in the lower left)
Figure 5: As Fig.4a, but after 22 h simulation time