Application And Study Of Improved Partheno-Genetic Algorithm To Catchment Basin's 3-D Modeling

Catchment basin is quite important in geochemistry and other domains.We often findand study regional abnormity with catchment basin In geochemical exploration. Whenselecting the site of reservoirs and planning the municipal construction, the influence ofcatchment basin should be considered. Some time ago, catchment basin is mostly extractedfrom topography contour map with manual work. However, the precision of 2-D map thatdrawn by manual work is quite low and it is also time-consuming and not visual. With the deeper and wider study of GIS technology, extracting catchment basinautomatically with computer had already applied in practical geologic reconnaissance. Inearly stage, when modeling catchment basin with hill-climbing algorithm, it often falls intolocal minimum because initial choice is in the local minimum area. Sequentially, the entirecatchment basin can not be obtained. In this condition, genetic algorithms'globaloptimization ability indicates the solution to the problem. Genetic algorithms simulate the parents reproduction processing of the most of life anduse genetic operators such as crossover and mutation,etc.Genetic algorithms have lots ofadvantage, such as high convergence speed, simpleness, robustness and adaptability. Inaddition, GA neither have special requirement for problem space nor calculate differentialcoefficient. GA have been applied extensively in every field. Now, to enhance the efficiencyand the capacity of GA becomes the hot study point. Partheno-Genetic Algorithm is a new genetic algorithms. Its genetic code is sequencenumber, not the exchange operator that the traditional genetic algorithms use frequently.And its all genetic operations go on with one chromosome. Partheno-Genetic Algorithm canachieve the function of traditional genetic operators through new operator such as geneexchange and gene shift. The algorithm can run successfully when all individuals inpopulation are the same without crossover operator. Diversity of initial population is notnecessary any more and "premature convergence" problem can be avoided. In practice, it is rather difficult to auto-detect catchment basin based on its traditionaldefinition when modeling with computer. Because the traditional catchment basin'sdefinition is lack of quantitative describing. Then, another definition of Catchment Basinbased on dot is presented. In this article, an improved Partheno-Genetic algorithm is broughtforward according to the new definition. The algorithm brings "tolerant choice" and other improved method to simplify geneticoperation and increases the computing efficiency. Diversity of initial population is notnecessary any more and "premature convergence" problem is overcome well. Thisalgorithm can work out entire basin in catchment basin's 3-D modeling, while thehill-climbing algorithm falls into local minimum. Since data in the current GIS are organized in 2-D maps, they can only process theinformation on the surface of the earth. 3-D GIS is needed when value change and propertyof 3-D space should be handled. So the development of 2-D GIS into 3-D GIS is inevitable. A prototype program is written using OpenGL and VC++.NET for the practicalcontrastive test in the study. And it can describe landform of the real world. On this platform,the implement of the new algorithm is tested by the DEM data from real world. The testresult testified the efficiency of this new algorithm and shows better third dimension. Inaddition, the prototype program and the experiment are very instructive in theory andpractice to the further research in GIS field.