The article is concerned with results of probabilistic interpretation for forecast dates when ice thickness growth up to 20-25 cm in the Kara, Laptev, East-Siberian and Chukchi Seas. The method of the probabilistic interpretation is based on the condition forecast errors distribution's approach formulated by the author (Dmitriev, 1997). Its goal is to find a way to estimate categorical (binary) forecast uncertainty for further translating into uncertainty in the quantity of interest to the user ac-cording to "end-to-end" forecasting concept. The results of calculating forecast errors distribution density parameters are given for combining data from the areas mentioned above. It provides a principal possibility to use binary forecasts in continuous optimization during planning sea ice-depended actions.
At present new techniques are developed for economic situations (which have weather dependent component) where the outcome is a continuous variable and where the decision is also a continuous variable (Langland et al., 2001; Smith et al., 2001). In accordance with "end-to-end" forecasting concept (Langland et al., 2001; Smith et al., 2001) it is necessary to translate uncertainty in the weather into uncertainty in the quantity of interest to the user that is clear for probabilistic forecasts but is a matter of some difficulty with categorical (binary) forecasts in case of continuous (or conditionally continuous) variable. The problem is well known and there are two basic ways to get a solution. A "classic" approach is based on Model Output Statistics (MOS) and it consists of "matrixes of contingency" calculation in accordance to traditional forecast procedures (Brooks et al., 1996; Murphy, 1985; Vorob'ev et al., 2002). A modern approach is based on an ensemble forecasting technique (see, for example, (Gustafsson., 2002) and its references) with a statistics modelling as a main idea.