This paper constitutes the sensitivity study of application the Polar WRF

model to the Svalbard area with testing selected parameterizations, including planetary

boundary layer, radiation and microphysics schemes. The model was configured, using

three one-way nested domains with 27 km, 9 km and 3 km grid cell resolutions. Results

from the innermost domain were presented and compared against measured wind speed

and air temperature at 10 meteorological stations. The study period covers two months:

June 2008 and January 2009. Significant differences between simulations results occurred

for planetary boundary layer (PBL) schemes in January 2009. The Mellor-Yamada-Janjic

(MYJ) planetary boundary layer (PBL) scheme resulted in the lowest errors for air

temperature, according to mean error (ME), mean absolute error (MAE) and correlation

coefficient values, where for wind speed this scheme was the worst from all the PBL

schemes tested. In the case of June 2008, shortwave and longwave radiation schemes

influenced the results the most. Generally, higher correlations were obtained for January,

both for air temperature and wind speed. However, the model performs better for June

in terms of ME and MAE error statistics. The results were also analyzed spatially, to

summarize the uncertainty of the model results related to the analyzed parameterization

schemes groups. Significant variability among simulations was calculated for January

2009 over the northern part of Spitsbergen and fjords for the PBL schemes. Standard

deviations for monthly average simulated values were up to 3.5°C for air temperature

and around 1 m s-1 for wind speed.

Significant increasing trends in the air temperature were found both in the surface station of Svalbard Lufthavn and in the low-tropospheric temperature field over the Atlantic Arctic. The variability in temperature, as well as the multiannual trend, is at least three times bigger in the winter months than in summer. An attempt was made to explain the high day-to-day variability in the winter air temperature by the daily variability in the regional pressure field and circulation conditions. Six regional-scale circulation patterns were found by applying the principal component analysis to the mean daily sea level pressure (SLP) reanalysis data and their impact on the low-tropospheric air temperature variability was determined. A bipolar pattern, with a positive center over Greenland and a negative center over the White Sea, dominates in the region and strongly influences the air temperature field at 850 hPa geopotential height (correlation coefficients up to –0.65). The second pattern that impacts the temperature field in the Atlantic Arctic is the one with a center of action over Svalbard (mostly a low-pressure center in winter), strongly influencing the air temperature over the Barents Sea. The remaining circulation types, explaining only 5–8% of the total variance of the SLP field each, do not modify significantly the air temperature at 850 hPa geopotential level over the Atlantic Arctic, and none of the circulation types seems to influence the multiannual temperature trends.