Study On The Break-up Characteristics Analysis And Simalation Forecast Of Wanjiazhai Reservoir Upstream Of The Yellow River

Before Wanjiazhai reservoir was constructed, water level dropped greatly from Lamawan in the upper reach of the dam to Mazha in the downstream of the dam. The section of the Yellow River from Lamawan to Toudaoguai was frozen up in the winter in some years in the upper reaches of the dam. However, the water level was raised maximally by 60m in the reservoir since the dam was constructed in 1998. The freezing up and thawing characteristics were changed due to the reduce of hydraulic gradient and flow velocity. As a result of this, the section of the Yellow River from Lamawan to Toudaoguai was stable frozen up in each winter. Ice dams were frequently occurred in the lower reach of Lamawan because of the effect of the reservoir backwater, which lead to serious ice flood disasters some times.To get better understanding in the formation mechanisms and formulate corresponding preventive measures of the ice flood disasters, field experiments were carried out in 2014-2016 including the observations of icing up and thawing initial time, processes of ice development, processes of ice flow during both the freezing up and thawing periods etc. BP neural network and GM were applied to model the initial time and duration of both the icing up and thawing. The historical ice data in 1998-2015 was used to build up and verify the models.The results showed that the place where the ice firstly covered the river cross section was occurred at Guandi in the winter in 2014～2015 and at Sifenzi in 2015-2016 respectively. The historical records in 1998～2015 showed that characteristics of freeze up upstream of the dam were:stable freeze up occurred at 32km upstream to the dam, stable and turbulent alternative freeze up occurred at 32km～52.1km upstream of the dam, turbulent freeze up occurred at 52.1km～67.6km upstream of the dam, and again stable freeze up occurred at 67.6km upstream of the dam. The turbulent freeze up usually occurred at "S" shaped curve river.The correlations between the freeze up and thawing dates and flow rate, air temperature, and ice thickness were analyzed. The results showed the air temperature was the most significant factor together with the flow rate determined the location and initial time of freeze up. The freeze up types mainly depended on the hydraulic gradient and river shapes. The ice formed the ice dam at the end of the reservoir near the cement plant was mainly from Toudaoguai reach 67～100km upstream of the dam. Thawing might start when the air temperature raised to 0～5℃ with a flow rate of 230m3/s. Although there were many effect factors that affected the freeze up and thawing processes, the air temperature, the number of frozen days, the ice thickness, and the flow rate were the main control factors. GM (1, N) modeled results were more accurate to the observed data compared to that of BP neural network, meaning that GM (1, N) might be more suitable for the simulation of ice conditions.