Polar lows are not properly resolved in global re-analyses. In order to describe the year-to-year variability and decadal trends in the formation of such mesoscale storms, atmospheric limited area models, which post-process re-analysis data, may be an appropriate tool. In this study we demonstrate the merits and potential of this approach.
A series of 3-week long ensemble simulations of weather situations over the NE Atlantic with a limited area model/regional climate model (CLM) are examined. The model was driven with NCEP–NCAR re-analyses at the lateral and lower boundaries. Additionally, the spectral nudging technique was used to enforce the large-scale circulation, as given by the NCEP–NCAR reanalysis, on the simulation. The ensemble members differ by initial conditions taken from several consecutive days.
In most of the cases, a polar low developed after a simulated time of about 2 weeks, that is, long after the initialization of the model calculations.
The spectrally nudged version of the model is very insensitive to initial conditions. The observed polar lows were reproduced in all ensemble members. A reasonable correlation between the simulated polar low features and those derived from a satellite product (HOAPS-III) and operational high-resolution weather analyses (DWD) is found. The polar lows are considerably deepened compared to the driving NCEP–NCAR analysis, but the comparison with weather maps indicates some differences in detail.
When CLM is run without the large-scale constraint of spectral nudging, considerable variability emerges across the different ensemble members and the observed polar low often does not emerge.