Similar pattern mining has become an important research direction in the field of Hydrology. It is a significant work to process similarity mining in historical data that can be conducive to recognize the trend pattern of hydrological data and provide technical support for the flood forecasting and flood control. Thus, this paper proposed a quick similarity search model according to hydrological sequence features. This model employed series segment, serial cluster and reference training method to generate reference set, and transferred similarity search to European vector space search with indexed by reference set so as to improve the searching efficiency.