Conclusions

Although this case was a very straightforward and clear example of a Polar Low which caused snow and wind problems over the North Sea and the adjacent coastal areas, it can hardly be placed in the category of catastrophic cases. Therefore, the number of casualties and damage costs were (luckily) low. The main reason being that Polar Lows are most active over sea and landfall of this low was during the very early morning, over a relatively sparsely populated area. Moreover, it decayed rapidly in the following hours.
Detecting the initial stage of a Polar Low is rather difficult because of its tendency to develop above Arctic waters in an observation sparse area. This is also the area where Meteosat images are of low quality due to the very low spatial resolution at that latitude. Polar satellite images are more appropriate for this region.
In this case some early developments had taken place at very high latitudes and were, therefore, impossible to detect with Meteosat. The required presence of characteristic meteorological parameters for development of such a phenomena was fulfilled over a large area in this case.
The investigated Polar Low, according to the Hirlam model, had not developed into a closed-isobar-low-pressure centre during the time period in question, but surface observations contradict that. It has to be concluded that the model provided a very clear and useful signal but was not able to completely describe these mesoscale phenomena.
In the initial stage the Polar Low was situated ahead of an isolated low-pressure area. In its later stages, the Hirlam model showed an upper trough with a surface trough ahead of it, and the Polar Low vortex near to the surface trough. The combination of widespread convection, the CISK mechanism (see introduction chapter) and PV as well as PVA were responsible for rapid fall in surface pressure and the development of a vortex. The Hirlam model calculated the surface pressure in the centre of the Polar Low 6 hPa too high, at the time it made landfall. Therefore, the forecast pressure gradient was also lower, which resulted in an underestimation of the wind speed by 3 to 6 m/s and, most important, by an underestimation of the wind gusts by 8 to 15 m/s.
The amount of precipitation and its exact location were also not well shown in model output.