Seasonal climate predictions for marine risk assessment in the Barents Sea

Marine risk embraces an assessment of likelihoods and consequences of impacts from climate fluctuations in order to identify time and regions vulnerable to climate hazards. This information can support sustainable and safe marine activities. The marine risk assessment is a part of the marine service provided by the DNV GL (short for Det Norske Veritas and Germanischer Lloyd). In their current risk application, likelihoods of extreme conditions on the sea are based on historical observations and atmospheric reanalyses. We assess predicted likelihoods of extreme conditions over 1990-2017 in the boreal summer (prediction months 2-4) from the seasonal forecast system provided by the German Meteorological Service (DWD). We chose summer as it represents the time of the open-water season, when the highest marine activity in the Barents Sea takes place. We selected three indicators from the marine risk assessment. Two of them represent meteorological properties such as wind speed and 2-meter temperature (T2m). The third indicator – the wind chill index (WCI) is a combination of the previous two and represents heat loss from the human body to its surroundings during cold and windy weather. As expected, the prediction skill assessment suggests different levels of predictability for the three indicators, with T2m having the highest skill followed by WCI and wind speed. The prediction skill represents the "trust layer" superimposed on the predicted likelihoods and used as input fields for marine risk assessment. From the likelihood maps for the test period of summer 2020 follows that large areas of the Barents Sea represent favorable conditions for marine operations considering high prediction skill and low likelihood for extreme WCI (>1000 W/m²) and T2m (<0 °C) conditions in July and August. The wind speed (>13.9 m/s) is poorly predictable beyond the first lead month. Thus, if risk assessment is based on a suite of climate indicators with the heterogeneous prediction skill, the total risk assessment might be limited by the skill of the indicator with the lowest prediction skill. However, not all climate indicators are equally contributing to the risk assessment. The study describes a workflow for application of seasonal climate predictions and points to a few lessons learned, which can be useful to future climate services.