Simulations of a Polar Low Using the Canadian Regional Climate Model (CRCM6): Objective Verification against Conventional Observations

Polar lows (PLs) are maritime mesoscale weather systems associated with severe weather such as gale-force winds, which can reach hurricane force, and heavy snow showers. These extreme or intense storms form at high latitudes, in sub-arctic and arctic basins as well as in the Southern Ocean, near the seasonal sea-ice edge and shorelines during marine cold air outbreaks. Therefore, they pose a threat to coastal communities, shipping, and extractive activities. PL forecasting has long remained a challenge due to the small size (typically a diameter of 300 - 400 km) and short lifetime (less than two days) of these systems. Nevertheless, the representation of PLs in atmospheric models has notably improved with the advent of high-resolution models. In this work, we present a case study of an observed PL that made landfall in Norway in March 2019. First, we conducted simulations of the PL using the developmental version of the convection-permitting Canadian Regional Climate Model (CRCM6). The model has a horizontal grid cell size of 2.5 km and 62 vertical levels. The ERA5 reanalysis was used for the initial atmospheric conditions and to drive the simulations. An ensemble of short-range forecasts were initialized at different times to evaluate the impact of the initial conditions on the representation of the PL. Second, we objectively verified the simulation output against conventional observations to assess the skill of the model at reproducing the observed PL. We used conventional observations from surface stations located in Norway and from drifting buoys. For each simulation and variable, we computed the mean error, the root mean square error, the Spearman correlation coefficient, and the root mean square error skill score. In addition to computing aggregate statistics, we computed the statistics across time at a single location and across a spatial domain for a single time. The results show that the initial conditions have an important impact on the forecasting of the PL. The objective verification of the simulations using SLP observations shows that the simulations initialised on 24 March at 12:00 UTC and at 18:00 UTC are the ones that perform the best. However, the verification of the simulations using observations other than SLP are less conclusive.