Research On Model Of Runoff And Sediment Yielding In A Watershed Based On Cellular Automata

Model runoff and sediment yielding is the key point to understand and simulate the process of the yielding profoundly, is an effective way to research the yield in a watershed quantitatively, and is also the requirement of watershed management and planning. The yield lies on many factors of a watershed, such as topography, physiognomy, soil, vegetation, weather, climate, human activities and so on. Because of the spatial-temporal variability and inhomogeneity of natural factors and the uncertainty of human activities, runoff and sediment yielding is a complicated nonlinear process. Cellular Automata (CA) is a typical method to study intricate phenomenon. CA has "bottom-up" research method, powerful complex computational function, inherent parallel calculating ability and lively feature of coupling time and space, so it is especially fit for dynamical simulation research of geographical spatial system. CA provides a new study method for building distributed model of runoff and sediment yielding in a watershed, which has significant theoretical and practical meaning.Models of runoff and sediment yielding based on geographical information system (GIS) and applications of CA on hydrological models were reviewed. The factors influencing runoff and sediment yield were analyzed deeply, and the Runoff and Sediment Yielding Cellular Automata was built based on runoff and sediment yielding model and basic theories of CA. An information system was developed. Chabagou watershed on the Loess Plateau was selected as an experimental area, and the model was applied to the watershed. The contents of the research were as follows. (1) The model was built. Using object-oriented method, the model was defined, extending the structure of traditional CA, including the set of sates, transition functions, time step, etc.; (2) The model and GIS were integrated. The data that the model needed was read, stored, processed, and exported by GIS. Supported by component of ArcEngine, the coupling between the model and GIS was implemented. (3) The model was applied. Topography, land use, vegetation coverage, hydrology and sediment of the watershed were selected, the model parameters were celebrated, and the model was validated and analyzed. Simulation scenes were designed artificially to analyze the influence of land use and vegetation coverage changes on the runoff and sediment yield of the watershed.The conclusions were drawn by the application of the model to experimental area. (1) Runoff and sediment yielding in a watershed had obvious nonlinear characters, and CA was an effective nonlinear way to simulate the complicated process of runoff and sediment yielding. (2) The traditional distributed models were extremely intricate and had many parameters, but the model had fewer parameters, alleviating the various task load of parameter calibration, and the structure of the model was simple, reducing the difficulty of design of the model. (3) Combining the principles of CA and calculation of runoff and sediment yield, building a model, and developing a information system of the model not only could calculate the runoff and sediment yield in a rainfall, but also could present the process of spatial variation of runoff and sediment dynamically. A relatively satisfactory result could be simulated by loop operations of the model.