Improving the forecast of an extreme Arctic Cyclone in August 2016 with WRF 4DVAR

In the Arctic, it is important to provide timely and accurate predictions of Arctic cyclones for lead times of a day (or a few hours) to a week to avoid weather-related risk in advance. Multi-Resolution Incremental four-dimensional variational (MRI-4DVAR) data assimilation method is applied to study the forecast of an extreme Arctic cyclone in August 2016. 6-hourly analysis-forecast cycling experiments are performed for 20-days during August 2016 with 7-day free forecasts initialized at 0000 UTCs. MRI-4DVAR experiments had better forecasting skill than 3DVAR, especially for longer forecast ranges of 5 to 7 days for both 20-day statistics of 7-day forecasts and a single 7-day forecast of an extreme Arctic cyclone case. In the case study of the extreme Arctic cyclone on 16 August, 3DVAR distorted two troughs in the Kara Sea that over developed one coastal cyclone and subsequently displaced a second one to the west. These two coastal cyclones merged over the Arctic Ocean to form the intense Arctic cyclone by 16 August. Better trough analyses in MRI-4DVAR are obtained in dynamic and thermodynamic fields benefiting from using a forecast model as a constraint to impose dynamic balance on the assimilation. The Arctic cyclone forecasts by the MRI-4DVAR method are more similar to GFS analysis than those by the 3DVAR. There are clear advantages in MRI- 4DVAR for the forecast of cyclone position and intensity on lead times of a week over the Arctic Ocean.