Improvement On Wavelet Analysis Methodology And Its Application In Hydrologic Time Series Analysis And Forecasting

Hydrologic series analysis is a typical and substantial topic in the field of stochastic hydrology. The wavelet analysis (WA) method has the ability of simultaneously elaborating the localized characteristics of time series both in temporal and frequency domains, thus it is suitable for hydrologic series analysis. Presently, wavelet analysis has been widely applied in the field of hydrology. Various analyses and researches have manifested the effectiveness and applicability of WA. However, in practical applications, the results of hydrologic series analysis by wavelet analysis are usually influenced by many key but difficult factors.In the present thesis, in order to improve and develop the wavelet analysis methodology, and further to improve the results of hydrologic series analysis, the topic of wavelet-based hydrologic series analysis was focused on. Firstly, present researches and applications of wavelet analysis in hydrology were reviewed, and their shortcomings and several open issues were pointed out. Then, two fundamental issues, namely wavelet function choice and wavelet decomposition level choice (called temporal scale choice in continuous wavelet analysis), were studied, and the methods for solving them were put forward. Based on them, three key but difficult issues concerning wavelet analysis were studied, and the corresponding suggestions and approaches to solve them were proposed; they include the wavelet threshold de-noising, discrete wavelet decomposition, as well as wavelet cross-correlation analysis; various cases studies (including both synthetic series and observed hydrologic series) verified the performances of these proposed approaches. Next, the improved wavelet analysis method was applied to study several typical issues in the process of hydrologic series analysis, including the quantitative characterization of hydrologic series’ complexity, periods’ identification of hydrologic series, and the phenomenon of periodic varying of hydrologic series; some new understandings and findings about the complicated composition and characteristics of hydrologic time series have been obtained based on these analytic results, and also two new methods of periods’ identification were put forward. Afterwards, the improved wavelet analysis method was applied to the middle and lower reaches of the Yellow River to analyze the variations of runoff and water-sediments during the last five decades, and applied to the Yangtze River Delta to analyze the climate variability in this zone during the last four decades. By comprehensive analyses, a set of new understandings and findings about the variations of climatic and hydrologic processes in these basins have been gained.After that, present studies on hydrologic time series simulation and forecasting were reviewed. By discussing several key issues concerning wavelet-based hydrologic forecasting, an improved modeling framework for hydrologic time series simulation and forecasting was put forward; analytic results of various cases have verified its effectiveness and applicability; followed by studying two key issues on hydrologic frequency analysis, that is, parameters’ estimation and Bayesian-based hydrologic frequency analysis, random characters of noises can be described more accurately, based on which the uncertainty of hydrologic time series simulation and forecasting results can be quantitatively evaluated.At last, analytic results throughout the thesis were summarized and concluded. The improved wavelet analysis methodology was explained systematically, and all the new understandings about the complicated compositions and characteristics of hydrologic time series gained in this study were summarized and elaborated. In addition, a set of suggestions on the future studies of wavelet analysis and hydrologic series analysis were given.