Factors determining the timing of ice appearance on the Kama Reservoir

The analysis and assessment of the correlations of the ice dates vice the zero-isotherm dates, dates of water temperatures transition through 2 and 3 °C were carried out. The initial data were long-term series of observations of air temperature at weather stations, water temperature and ice dates at the gauge stations of Roshydromet for the whole period of the reservoir existence (1956–2021). As the characteristic under study, the dates of ice formation in autumn were used. Main results: verification of previously obtained correlations on independent data (1996–2021) showed that the accuracy of issued forecasts (P) was decreased by 20% compared with the accuracy on dependent data (1956–1995); an analysis of the whole period applicability (1956–2021) for constructing predictive correlations with an assessment of their justification showed that there was no noticeable refinement. The accuracy of the forecasts based on the dates of water temperatures transition through 3 °C has not changed, while forecasts based on the zero-isotherm dates and the dates of water temperatures transition through 2 °C, has noticeably decreased at all stations; the use of data only for the period of modern climate changes (1996–2021) led to the following: correlations of the ice dates vice the zero-isotherm dates and water temperature through 2 °C did not pass the criteria S and P. This period is characterized by a decrease in the frequency of small forecast errors (< 4–5 days) and an increase in large errors (> 6–11 days) by 2–3 times; a forecast technique is proposed based on the separation of the studied correlations for years with different meteorological conditions in autumn (years with stable cooling and with heat returns). The criterion for using this or that dependence in a particular year is the weather forecast. Forecasts issued according to such dependencies have a higher accuracy.