Вероятностный подход к прогнозу сроков появления ледовых явлений на северо-востоке Каспия

In order to minimise damage to various sectors of maritime activities in the northern part of the Caspian Sea, the development of timely and accurate forecasts for the appearance and destruction of ice cover is required. In the conditions of the climate changing, synoptical and statistical methods typically used for long range forecasts are losing effectiveness. The most promising method for hydrometeorological characteristics forecasting is the calculation of a numerical hydrodynamic model using an ensemble approach. This study justifies the use of reanalysis data of meteorological parameters to identify thir predictive relationships with ice formation processes. The research was conducted based observational data of the Caspian ice conditions from 2000 to 2021. For predicting ice appearance with a lead time of over a month, an ensemble forecast developed using the CFSv2 model was proposed. The study suggests calculating the sums of daily degrees below freezing based on the forecast ensemble of the model and determining the probability of achieving forecasted calendar dates of ice appearance. To verify the proposed approach, an experimental forecast of ice formation processes in the northern Caspian Sea area for the winter of 2020–2021 was developed. The 2-month lead-time forecast showed an error of three days for predicting the first ice appearance relative to the actual dates. The ice cover appearance forecast was developed with an 80 % probability of realization, consistent with observations obtained from the MODIS satellite.