Breakup Ice Jam Forecasting Based on Neural Network Theory and Formation Factor

Cold regions suffer from severe ice jam flooding in the spring. Forecasting of ice jam and its break-up is crucial to prevent or reduce flooding risk in cold region. This paper analyzes the formation factors governing the formation of breakup jams. A neural network model based on formation factors has been developed with air temperatures and precipitation as inputs and applied for ice jam forecasting in a given year in the upper reaches of the Heilongjiang River (Amur River), where ice flooding occurs frequently during spring. The model based on the neural network clustering method had an accuracy rate of 85%, which was significantly higher than the 62% accuracy rate of the conventional statistical method for breakup ice jam forecasting. The model had a forecast period of 10 days with a maximum error of 2 days and forecast qualified rate of 100% for breakup date forecasting. The forecast on the breakup ice jam, which was released 24 days ahead, provides the accurate results for the breakup date and the occurrence of breakup ice jams in the spring of 2017.