Initial Condition Sensitivity of Polar Low Intensity and Location Determined Using Ensemble Sensitivity Analysis

Abstract The impacts of initial condition errors on the forecasted intensity and location of two polar lows (PLs) occurring over the Norwegian–Barents Seas region are evaluated by applying the ensemble sensitivity analysis (ESA) to 48-h ECMWF ensemble forecasts valid at the time of the PL maximum intensity. The two PL cases represent opposite ends of the so-called PL spectrum and are chosen based on the observed cloud morphology and large-scale atmospheric conditions. The comma-cloud case is characterized by strong, deep-tropospheric baroclinicity, while the spiraliform case exhibits weaker, low-level baroclinicity. Both cases develop in connection with a marine cold air outbreak and an associated upper-level trough. Our analysis demonstrates the importance of the marine cold air outbreak (MCAO) temperature (based on the upper-level geopotential height) to the intensity of both systems, but the comma-cloud case is more sensitive to the strength of environmental baroclinicity. The forecasted location of both systems is sensitive to the representation of the synoptic-scale circulation, with the positioning of the upper-level trough ultimately determining where the PL forms. Our analysis suggests that the PL location is more predictable than PL intensity, and we show how ensemble prediction might be utilized to understand uncertainty in real-time forecasts of PLs. Significance Statement Polar lows are intense high-latitude cyclones with small spatial scales (∼300 km) and lifetimes on the order of a day. They intensify rapidly and produce hazardous weather conditions that threaten communities and operations in the Arctic, which motivates skillful polar low forecasts. Accurate initial conditions for polar low forecasting are challenging due to sparse observations in the Arctic. In this study, we investigate the relationship between initial condition uncertainty and short-term forecasts of polar low intensity and location. We find strong relationships between the representation of key environmental conditions and forecasts of polar low intensity and location, which may be used to better assess uncertainty in the real-time forecasting of polar lows.