Forecasting Catastrophic Floods in Crimean Territory

The catastrophic situations of recent years—in June 2021 near Yalta and in January 2024 in Sevastopol—associated with abundant precipitation, water level rise in rivers, and the formation of mudflows—once again showed the need for advance forecasting of events with extreme precipitation in Crimean territory for prompt response and minimization of economic losses. The region of the mountain Crimea with its complicated relief and considerable slopes is especially susceptible to the formation of dangerous situations after heavy (often multi-day) rains. Daily data on precipitation from the Ai-Petri weather station were used to calculate and analyze the cases with total precipitation ≥40 mm within three consecutive days. Such conditions were used in the analysis as a threshold of extreme precipitation leading to channel erosion in mountainous Crimea rivers and the formation of debris flows. The catastrophic flood on the Chernaya River in January 2024, which was due to three days of extreme precipitation in the Sevastopol region, is considered. This situation was analyzed to determine the possibility to forecast it up to 3 months in advance with the use of the developed artificial neural network model. The obtained results showed that the quality of the developed neural network is satisfactory to forecast with a lead time of 3 months 2–3-day long extreme precipitation, which intensifies the erosion processes in the mountainous Crimea.