An Intelligent Method For Extracting Spatial Parameters Of Hydrological Models

Hydrological model is a mathematical model describing hydrological process. It plays a vital role in scientific research and integrated watershed management. Spatial parameterization for hydrological modeling is an indispensable step. The main inputs of hydrological model are generally extracted from the land surface data (such as DEM,soil data, landuse data, etc.). In general, the process of spatial parameterization involves complex software operations and professional tools, and it is often difficult for an amateur to conduct this work. With the extensive use of hydrological models, there are more and more users from other professional fields. These users usually lack the professional knowledge of hydrology, and have difficulty in extracting hydrologic model spatial parameters using existing GIS methods and tools.To help users without professional knowledge extract hydrologic model spatial parameters, this thesis proposed an intellectualized method of spatial parameterization based on formal knowledge representation and computer inference method, which integrates knowledge-driven modeling and visual modeling techniques. Moreover, a prototype system was developed to showcase this intellectualized method. The main research contents and results are as follows:1. Knowledge formalization and inference for spatial parameter extractionFirstly,obtain and classify the knowledge about spatial parameterization from user's manual or documentations. Then implement the knowledge formalization based on resource description framework and JavaScript object notation methods. Depending on the application purpose and parameter settings, according to graph theory, design a computer inference system and determine hydrological model spatial parameterization workflow.2. Design and construction of hydrological model spatial parameterization prototypeBased on the knowledge formalization for spatial parameter extraction, this thesis builds a knowledge library and algorithm library of hydrological model spatial parameterization. Java and JavaScript languages are used to develop the automatic inference engine, semi-automatic inference engine, and workflow engine. Additionally,using visual modeling technology, this thesis achieves a visual modeling of hydrological model extraction workflow. Finally, a hydrological model spatial parameterization prototype is developed by Java Web technology and B/S architecture.3. The application and evaluation of the methodTake SEIMS hydrological model as an example, this thesis gives a comparison of two different extraction methods: one is traditional spatial parameter extraction method, the other is the prototype proposed in this thesis. Through the comparison of these two methods, the prototype in this thesis is verified to be an effective one, because it uses hydrological model spatial parameterization knowledge comprehensively and it is much easier for an amateur user to conduct a hydrological study. Also, with the help of computer inference system and visualization technology, the thesis achieves an intelligent construction of spatial parameterization workflow, which shows a good intuitiveness and easy-using.In order to benefit the non-expert users,this thesis proposed an intellectualized method to extract hydrological model spatial parameters based on knowledge-driven modeling and visual modeling technologies. This is to show scholars a much easier and more unsophisticated hydrological model method, and provide them a quick way of hydrological modeling.