Diagnosing Factors Influencing the Forecast Skill of Two Intense Arctic Cyclones in Early June 2018

Two unusually intense Arctic cyclones, AC1 and AC2, occurred in early June 2018 over the Arctic. AC1 formed northeast of the Caspian Sea within a frontal trough, and AC2 formed in the lee of Greenland along a moisture axis accompanying the remnants of Tropical Storm Alberto. Both AC1 and AC2 strengthened in a region of strong baroclinicity over western Eurasia ahead of respective high-amplitude upper-level troughs. AC1 attained a peak intensity of 968 hPa on 4 June in the ERA5, and AC2, which subsequently interacted with and absorbed AC1, attained a peak intensity of 962 hPa on 7 June in the ERA5. In a previous study, forecast skill of intensity and position for AC1 and AC2 at the time of peak intensity was examined using the 51-member ECMWF ensemble prediction system (EPS). Forecast skill of intensity for AC2 is much lower than that for AC1, with intensity forecasts being strongly underdispersive for AC2 and slightly underdispersive for AC1. Forecast skill of position for AC2 is higher than that for AC1 at 72–120-h lead time and lower than that for AC1 at other lead times, with position forecasts being somewhat underdispersive for AC2 and moderately underdispersive for AC1. The purpose of the present study is to diagnose factors that may influence the forecast skill of the intensity and position of AC1 and AC2.

At selected forecast lead times for each cyclone, ensemble forecasts from the ECMWF EPS will be separated in two groups: the most and least accurate members in terms of intensity and position errors of the cyclone at the time of peak intensity. Composite differences between the two groups will be computed for selected dynamical and thermodynamic quantities to diagnose factors that may influence the forecast skill of each cyclone. Then, ensemble sensitivity analysis will be performed using the ECWMF EPS to determine the sensitivity of the intensity and position of each cyclone at the time of peak intensity to the selected dynamical and thermodynamic quantities at earlier times. For example, it is hypothesized that there is sensitivity of the intensity and position of AC2 at the time of peak intensity to the structure of the sea level pressure field associated with AC1 at earlier times because AC2 interacts with and absorbs AC1. Thus, the aforementioned sensitivity, among others, will be examined in this study.