Simulation of the ‘great’ 2012 Arctic cyclone with WRF

WRF (Weather, Research and Forecast) is a regional atmospheric model, which allows us to simulate synoptic-scale processes with an adequate resolution. In this study, we use two different reanalysis products, ERA-­‐Interim reanalysis and NCEP reanalysis, to simulate the great Arctic cyclone of 2012.
The Arctic cyclone that developed on August 2nd over Siberia and dissipated
around August 14th in the Canadian Archipelago was the most intense August
cyclone in terms of its sea level pressure (for more see Simmonds and Rudeva,
2013) observed in the Arctic since the begin of satellite remote sensing. This cyclone appeared during a period when the Arctic sea-­‐ice extent was on its way to a new record minimum and lasted for almost two weeks before it finally dissipated. As such, the cyclone got a lot of attention and its potential impact on the sea ice extent was investigated (e.g Parkinson and Comiso, 2013). In this study we investigated whether the sea ice cover, in turn, has an impact on the cyclone development.
At first, we tested whether WRF is able to simulate the cyclone in a control run with a spatial resolution of 47.5 x 47.5 km grid. The direct comparison shows that WRF simulated the cyclone reasonably well. The cyclone enters the Arctic around August 4 and intensifies on its way to the Arctic Ocean where it reaches a minimum pressure of 968 hPa on August 6. It stays over the Arctic Ocean for several days until it finally dissipates on August 13 over the Canadian Archipelago. Sensitivity simulations were run to test the effects of domain size, forcing conditions, time step, and sea ice concentration. For the domain size, a run with 60km resolution was compared with the 47.5km resolution of the control run.
The resultant larger domain produced a weaker, less organized cyclone. This may have resulted from an increased distance between the domain boundary and the area where the cyclone developed in the larger domain run. This leads to the conclusion that the placement of the domain boundary is extremely important for WRF to properly develop the cyclone. Two different reanalyses were used to supply initial and boundary conditions to the model. The ERA-­‐Interim dataset was used in this sensitivity test, while NCEP reanalysis was used in all other runs including the control run. While there were subtle differences between the trajectories of minimum sea level pressure in the two runs, the overall results were very similar.
While most runs had an integration time step of 600 seconds (10 minutes), an additional run was performed with a time step of 360 seconds (6 minutes). This 6 minute run was able to resolve atmospheric phenomenon with sharp temporal gradients a little better. The resulting trajectory remained in the ocean for a longer period than the control run, but had a slightly higher minimum pressure than the control run. However, the magnitudes of the differences were small, and we suggest that the time step does not play a major role in the simulation of the cyclone.
The final sensitivity test involved varying the sea-­‐ice conditions to initiate changes in the surface forcing. There were three runs performed: the control run, which had an evolving concentration of sea as determined by the NCEP reanalysis; the "fixed ice" case, where the sea ice concentration was held unvaried at the extent determined by NCEP from June 1st; and the "no ice" case where sea ice was removed from the simulation. The changes in cyclone development were drastic. The cyclone barely developed in the "no ice" case, while the "fixed ice" case had weaker SLP than the control run. One possible theory for this outcome is the different response on instability growth induced by the spatial inhomogeneity of sea ice and open water in the runs. For the control run, the cyclone tracked along and across many ice-­‐water boundaries. The "fixed ice" case had less of these boundaries in the cyclone path, while the "no ice" case had none. This leads us to conclude that there is a sizeable effect of the underlying surface conditions on cyclone development, specifically the inhomogeneity between open water and sea ice.
References
Simmonds, I., and I. Rudeva (2012), The great Arctic cyclone of August 2012,
Geophys. Res. Lett., 39, L23709, doi:10.1029/2012GL054259.
Parkinson, C. L., and J. C. Comiso (2013), On the 2012 record low Arctic sea ice cover: Combined impact of preconditioning and an August storm, Geophys. Res. Lett., 40, doi:10.1002/grl.50349.