Impact of Data Assimilation and Model Physics on the Predictability of the 2012 Great Arctic Cyclone

As human activity thrives in the Arctic, accurate prediction of Arctic weather is increasingly important. Compared to mid-latitude cyclones, studies on the predictability of Arctic Cyclones (ACs) are relatively limited. This study aims to understand both the impact of data assimilation and model physics on the predictability of the 2012 Great AC.

To understand and quantify the impact of model physics, we perform ensemble simulations with varying longwave and shortwave radiation, microphysics, cumulus and planetary boundary layer schemes. We initialize Weather Research and Forecasting Model (WRF) ensembles from Global Ensemble Forecast Systems (GEFS) analyses at 0600 UTC Aug 4 and conduct 72-hour prediction. Current results suggest varying longwave radiation schemes produces the largest spread while varying microphysics produces the smallest. A detailed look into diabatic source demonstrates that diabatic cooling patterns near tropopause differ noticeably depending on the choice of longwave radiation schemes. Such a difference in diabatic cooling subsequently affects the surface cyclone through the Tropopause Polar Vortex (TPV) and surface cyclone interaction. The multi-physics ensemble forecast envelopes the track and peak intensity of the AC in Global Forecast System (GFS) Reanalysis. However, during the cyclone initiation phase, ensembles collectively predict much faster deepening of the AC, indicating existence of other source of uncertain than model physics. Ensemble data assimilation experiments are ongoing to understand and quantify the impact of initial condition errors on the predictability of the same AC.