Development And Application Of Similarity Analysis Of Rainstorms And Floods Extension Model At The Basin Scale

During every stage of flood control operation, finding out the similarity degree between a current rainstorm flood and historical rainstorm floods, and making effective use of the generating and developing information of similar rainstorms and floods in history, are a fresh method to improve real-time flood-control forecasting system. At present, similarity analysis of rainstorms and floods has not been carried out effectively in large river basins of China. An extension model for similarity analysis of rainstorms and floods is established based on the research area of Songhuajiang River Basin, which is a useful trial for developing similar analysis in large river basins.Firstly, similarity analysis index system of rainstorms and floods is studied and summarized through the three levels, which are atmospheric process, generating and developing processes of rainstorm and flood. Secondly, the extension theory is introduced and the relevant methods are applied to resolve the similarity analysis problems, which are difficult to identify subjectively from multi-angles and multi-directions. The idea of searching out the similarity degree through comparing the amount of rainfall in the whole river basin is brought forward. For realizing it, cumulative average rainfall and the number of rain gauge stations involved in a time interval of rainstorms are selected as two characteristic indices to estimate the proximity of the amount of rainfall between sub-basins. Then by evaluating the relative magnitude of rainstorms in the whole river basin, the analysis of similar rainstorms is made, and the similarity analysis extension model is established. In the similarity analysis of floods, the extension model is developed which takes the average water level and average flow of key stations as two characteristic indices. Moreover, the novel extension neural network is introduced, and similarity analysis extension neural network model is built on the characteristic indices used previously. Finally, the two kinds of models are applied for similarity analysis in Songhuajiang River Basin, and the results and methods of both models are studied and compared.The results show that both kinds of extension model can be used to realize similarity analysis of rainstorms and floods, and the process of establishing the models can be referred to by researchers of other river basins. The comparison suggests that the similarity degree between different rainstorms and floods can be indicated by specific quantitative value with extension evaluating model, while the grade of rainstorms and floods can be distinguished easily with extension neural network model.