Research And Application Of Hydrological Time Series Similarity Pattern

Hydrological time series similarity search can be used to rain and floods forecast, and the analysis of environment evolvement and hydrological process, etc. The direct application is to answer such questions as "which hydrological process in history does the ongoing one correspond to?" The quantity of hydrological data is huge and its type is complexity. It is hard to analysis using only the methods of hydrological theories and traditional statistics, because of the limitation of computing ability, storage ability and algorithms. The introduction of data mining theories and technologies to hydrological area provides a new way to settle the problems that the hydrological science confronts with.Based on the pattern representation and similarity measure of time series, this paper explores the similarity search model, which is adaptive to the characteristics of hydrological data. The work of the paper mainly includes:(1) Pattern Representation of Time Series.Use a method of time series piecewise linear representation based on feature points as a way for pattern representation. This method is simple and intuitive, and can achieve data compression effectively for time series with obvious periodicity and violent pattern fluctuation during a specific period, so as to well hold the change characteristics of the whole pattern.(2) Similarity Measure.Based on the pattern representation of time series, this paper introduces the Dynamic Pattern Matching (DPM) distance, which supports time warping of series and whose time complexity approaches linearity along with the increase of the length of patterns.(3) Similarity Search Model of Hydrological Time Series.We design the similarity search model of time series, which is adaptive to the characteristics of hydrological data. Then using the method of PLR based on feature points as means of dimension reduction and DPM distance as similarity measure, combining the motif of hydrological time series similarity mining, this paper implements a concrete application.Experiment shows that the combination of PLR based on feature points and DPM is an effective way to deal with the common issue of hydrological similarity.