Assessment of simulations of a polar low with the Canadian Regional Climate Model

PL, the accuracy of ERA5 and the correlation coefficient are notably higher than that of the simulations (not shown). 3.3.4 Relative humidity at 2 m. The simulations that have captured the PL show some- what better skill at representing the relative humidity than those that have not captured it (Table 5). The simulations 23d12h and 24d18h have the lowest mean error (0%), and 24d12h and 24d18h have the lowest RMSE (13%). The simulation 24d18h shows a weak correlation with the observations (0.41), and the other simulations have virtually no correlation with the observations. The time series of the ME (Fig 10C), the RMSE (not shown) and the Spearman correlation coefficient (not shown) differ between simulations, although 24d12h and 24d18h show a quite similar pattern. The ME of the simulations tends to converge with time, and the difference between them remains relatively small from around 1600 UTC on. The time series of the Spearman correlation coefficient shows correlation coefficients ranging from -1 to 1, and even the same simulation shows a wide range of correlation coefficients. Compared to the simulations, ERA5 has higher accuracy (RMSE of 11%) and notably higher correlation with the observations (r of 0.62) (Table 5). However, its ME (1%) is only lower than that of half of the simulations. ERA5 has lower RMSE and higher correlation coeffi- cient than the simulations from 1400 UTC until the PL has dissipated (not shown). 3.3.5 Wind at 10 m. The simulations that have not captured the PL show few values of

wind speed over 15 m s-1, whereas 24d12h and 24d18h show several values larger than 15 m s-1 (not shown). In general, all simulations show some large overestimations and underestima- tions of wind speed, but the observed wind speeds over 20 m s-1 are better captured by 24d12h and 24d18h. Most of the wind directions of the simulations that did not capture the PL are Table 4. Aggregate statistics computed to verify the simulated and ERA5 2-m dew point temperature against the observations from surface stations. ME [˚C] RMSE [˚C] r 23d00h 1.2 2.9 0.58 23d06h 1.4 3 0.53 23d12h 1.7 3.1 0.55 23d18h 1.9 3.1 0.47 24d00h 2.3 3.4 0.45 24d06h 1.9 3.3 0.44 24d12h 2.2 3.1 0.53 24d18h 2 3 0.56 ERA5 1.1 2.1 0.73 https://doi.org/10.1371/journal.pone.0292250.t004 Table 5. Aggregate statistics computed to verify the simulated and ERA5 2-m relative humidity against the obser- vations from surface stations. ME [%] RMSE [%] r 23d00h 1 16 0.2 23d06h 3 17 0.23 23d12h 0 14 0.27 23d18h 2 16 0.16 24d00h 5 19 0.03 24d06h 3 17 0.12 24d12h 1 13 0.31 24d18h 0 13 0.41 ERA5 1 11 0.62 https://doi.org/10.1371/journal.pone.0292250.t005 PLOS ONE Assessment of simulations of a polar low with the canadian regional climate model PLOS ONE | https://doi.org/10.1371/journal.pone.0292250 October 5, 2023 17 / 28