The Analysis And Prediction Of Hydrological Characteristics Based On Wavelet Analysis In Handan Mountain Area

Hydrology is a subject area which is based on studies of large quantities of data, These data are accumulated through long-term observation and practice by the hydrologists, The data includes climate, geography and other natural change rule and the interaction mechanism and the corresponding hydrological factors to human activity in response to a variety of information. This makes the hydrological sequences has generally nonlinear, variability, randomness, and other characteristics, while humans will applied wavelet analysis to the hydrologic system in order to be able to accurate analysis and forecast of hydrological sequences.Since enter the21st Century, Wavelet analysis, fractal theory and artificial neural network technology and various nonlinear method of gradual development and increasingly mature. Because the wavelet with automatic variable "time-frequency" window, thus it has the multi-time scale resolution function; Fractal is called the real geometry and is the quantitative index which determine the appropriate of sequence complexity; Artificial neural network has adaptive, self-learning function, so it can make up for a series of defects of the traditional linear statistical model; The above-mentioned nonlinear method has been widely applied in hydrology and water resources subject.This thesis adopt wavelet analysis, fractal theory and artificial neural network technology as the basic techniques, analysis variation characteristics of Multi-scale periodic and complexity for the mountainous area of HanDan area nearly45years of rainfall sequence at the first time, and build forecasting model about a typical area of Qing Zhang River Basin Liu Kuang range runoff through a combination and coupling between various nonlinear method.Through the Morlet wavelet analysis and calculation concluded that the annual precipitation process of HanDan mountain exists5,9years or so of the main cycle changes, and there exists about3years of scale variation.Through the analysis we can know that wavelet coefficient variation of annual precipitation change,and it can reacte the characteristics of annual rainfall abundance in the range of time and areas change and evolution of the trend.On the foundation of Morlet wavelet transformation, this thesis using correlation analysis and significance test method reveal the relationship between regional hydrological time series which study in different scales and the change of wet and low water. The result show that the rainfall sequence of Zhang He mountain and the runoff series of typical regional Qing Zhang He Liu Kuang interval have a good correlationTaking the coefficient combination of Daubechies7discrete wavelet and information quantity and the index portfolio of Daubechies7continuous wavelet and spectral as evaluation index respectively, then estimate various evaluation index of rainfall process of the Han Dan’s mountains. Due to the different calculate index, each hydrological series complexity also have its corresponding characterization. The two groups of indicators calculated results finally show that Fu Yang mountain rainfall process have more complexity than Zhang He mountain.Based on the wavelet analysis method and artificial neural network theory, using the Daubechies7wavelet BP neural network assemble model to construct the area of wavelet neural network combination model. To take the rainfall of Zhang He mountains as input, simulating the interval annual runoff of Zhang He Liu Kuang, the results of forecast is good.