Study On Theory And Methods Of Watershed Hydrologic Modeling

Water is the life source of the earth-biological system. With global climate changeand human activities, in water resources engineering, the major concern is given to twotopics, flood and drought. And hence, it extends to water pollution and waterenvironment. Hydrological simulation of watershed is one of the major theories forstudying hydrological natures and characteristics and to solve the practical problems ofhydrological engineering. Because of the spatial and temporal variability of naturalconditions, these mechanisms understanding is still far from complete. To pursue theman-nature harmonious and sustainable society and economy, and to open out themystery of hydrological phenomena, a series of novel theories and correspondingmethods should be researched and created in the present and the future. This dissertation, has studied on watershed rainfall-runoff system, focused onsystem distributional and nonlinear characteristics, combining artificial neural network(ANN) with Volterra functional series, combining genetic algorithm with simulatedannealing, applying geographic information system (GIS), remote sensing (RS) anddigital elevation model (DEM) in digital watershed. The main studies that have resultedare the following: The dissertation consists of three parts and eight chapters. Part 1 (Chapter 2 andChapter 3) studies on digitizing watershed and the method for information managementfor watershed. Part 2 (Chapter 4 and Chapter 5) studies on ANN theory and itsapplication of rainfall-runoff model. Part 3 (Chapter 6 and Chapter 7) researches onnovel methods for flood routing in channel. Chapter 1 provides an overview of watershed hydrologic modeling, introduces theproposed, objective and significance of the project, and sum up the related theory,existing problems and development of watershed hydrologic modeling. Chapter 2 researches a technical foundation for digital hydrological modeling basedon DEM. The spatial distribution of land surface characteristics, such as watershed divideboundary, drainage network, slope, aspect, topography, catchment area, could beexpressed digitally. The automatic extraction of stream network from DEM represents aconvenient and rapid way to parameterize a watershed. Chapter 3 explores a new method based on GIS for information system for water IIIresources. The Qingjiang watershed information management and analysis system isdeveloped with GIS software platform. This system realizes visual inquiring,management, analysis, calculation about hydrological information. It suppliesinformation support to assess prevent flooding, understand environmental issues, andmanage water resources. Chapter 4 solves the difficulties in determining the number of hidden units of ANN,a self-organized neural network model has been established. The theory is the basis ofconnected analysis, this self- organized neural network algorithm can automatic unite thenodes of correlation, and adjust the number of hidden nodes. Chapter 5 proposes Volterra neural network hydrology model (VNNH) based onstudying in consistency of Volterra model and ANN model. VNNH model relieved thelimitation of Volterra method. The number of hidden units of VNNH model wasestimated by the self-organized neural network algorithm. VNNH model has beendesigned with a kind of polynomial activation function. It is found useful and effectivefor high-order nonlinear systems via equivalent VNNH training. The initial weights ofVNNH equal to the value of unit hydrograph of watershed by the structure of Volterramodel. Numerical simulations in Qingjiang watershed have been made and used to testthe VNNH model. Comparisons of VNNH model to Volterra and ANN modelrespectively show that the proposed VNNH presented much more effective and valuablefor the applications to hydrological nonlinear system. Chapter 6 identifies parameters of flood diffusion wave equation based on ANN.Flood...