Method for predicting the ice cover parameters in the waters of the Far-Eastern Seas with a long lead time

Abstract

New method for long-term forecasting of mean month and mean 10-days values of the ice cover and position of the ice edge in the Far-Eastern Seas is presented. The sea ice regime is formed under influence of thermal and dynamic patterns in the atmosphere and hydrosphere, though mechanisms of its forming and evolution are not yet completely clear, so the sea ice forecasting is based mainly on statistical methods. The new method is developed for the ice parameters prediction for the period with stable ice cover. It uses a physical-statistical model with ensemble approach. The minimum lead time of this method is 7 months. The model assimilates the data on absolute topography of 500 GPa surface, atmospheric pressure at the sea level, air temperature at 850 GPa surface and at the sea surface, relative topography of 500/1000 GPa surfaces, and the South Oscillation index. Archives of these fields for the Northern Hemisphere from 1961 to 2017 are loaded. The ensemble of predictions is formed using the criterion of their maximum accuracy on independent data sets. The method is tested for the winter seasons of 2015/2016 and 2016/2017. The most accurate by 3 parameters are the forecasts for the Okhotsk Sea with the average accuracy 75–83 % that is much better than the accuracy of climatic forecasts (61–67 %). The forecast of the mean month ice cover only is satisfactory for the Japan Sea, and the forecast of the ice edge position only (65 % accuracy) exceeds the climate forecasting accuracy for the Bering Sea, while the climatic forecasting shows better results for the ice cover. The average accuracy of forecasting with new method (all parameters for all seas) exceeds 70 %, that allows to recommend the method for practical using. A prognostic product could be proposed as charts of the sea ice edge for future winter with estimations of the ice cover for each sea by months and 10-days.