The polar low in the Norwegian Sea on March 16-17 2008 has been studied using both weather analysis and the numerical weather prediction (NWP) model weather research and forecasting (WRF). This particular polar low was poorly forecasted by several operational models, and has therefore been subject to this study.
Weather analysis show that the low developed during a complex weather situation, in a confluence zone between polar and arctic air. It was one of three vortices found simultaneously in a wave-like cloudpattern along the confluence zone. There was a clear upper-level forcing by an advancing potential vorticity (PV) anomaly, along with strong convection during the cyclogenesis stage. The polar low developed in a region of the confluence zone with relatively high surface temperature and low static stability in the lower troposphere. It is suggested that this contributed to the rapid development of the vortex, and that the two other vortices might nothave experienced this.
Sensitivity studies with different initial times, high resolution runs and different parametrisation schemes representing microphysics, cumulus clouds and planetary boundary layer, were carried out. There were great deviations between all the simulations, particularly regarding time and location of development as well as trajectory and depth of the polar low, but all managed to produce at least one clear low. Most of the models produced several lows, which could be connected to the multiple vortices found in the confluencezone.The simulations initiated after cyclogenesis reproduced the depth of the polar low much better than the simulations initiated before cyclogenesis. It appears that the model have trouble simulating particularly the rapid development at the early stage. In contrast to previous studies, the high resolution runs did not improve the forecast much. The sensitivity study of the differentparametrisation schemes of microphysics, cumulus clouds and planetaryboundary layer, only showed improvements by changing the boundary layer scheme.
It is concluded that the WRF simulations produced a much improved forecast compared to the operational HIRLAM simulations, and alsothe UM simulations carried out by McInnes et al. (2011). It is suggested that this may be due to better lateral boundary conditions due to the nesting option in WRF. Further investigations could reveal if the WRF model, in general, is be better suited for polar low forecasts than the operational HIRLAM.