A Comparison of Arctic and Atlantic Cyclone Predictability

Abstract

Abstract Arctic cyclones (ACs) are synoptic-scale features that can be associated with strong, intense winds over the Arctic Ocean region for long periods of time, potentially leading to rapid declines of sea ice during the summer. As a consequence, sea ice predictions may rely on the predictability of cyclone-related wind speed and direction, which critically depends on the cyclone track and intensity. Despite this, there are relatively few studies that have documented the predictability of ACs during the summer, beyond a few case studies, nor has there been an extensive comparison of whether these cyclones are more or less predictable relative to comparable midlatitude cyclones, which have been studied in greater detail. The goal of this study is to document the practical predictability of AC position and intensity forecasts over 100 cases and compare it with 89 Atlantic Ocean basin midlatitude cyclones using the Global Ensemble Forecast System (GEFS) Reforecast V2. This dataset contains 11-member ensemble forecasts initialized daily from 1985 to the present using a fixed model. In this study, forecasts initialized 1 and 3 days prior to the cyclone development time are compared, where predictability is defined as the ensemble mean root-mean-square error and ensemble standard deviation (SD). Although Atlantic basin cyclone tracks are characterized by higher predictability relative to comparable ACs, intensity predictability is higher for ACs. In addition, storms characterized by low ensemble SD and predictability are found in regions of higher baroclinic instability than storms characterized by high predictability. There appears to be little, if any, relationship between latent heat release and precipitable water and predictability.