A Cellular Automata based Distributed Model for Simulating Runoff in LeAnhe Watershed

The effective management of the water related resources-environment-disasters system (or as traditionally and professionally so called as hydrological system) is one of the vital subjects for social and economic development and people's living and wealth. It's an essential solution for water related issues to comprehensively manage the system on the base of natural drainage basin by starting from the rainfall-runoff simulating, routing and forecasting, and exploring the physical mechanism and system control measurements. Watershed-based rainfall-runoff model and routing model are always the research highlights in the field of hydrology as well as environmental sciences and calamity. With the development of geographical information system (GIS), remote sensing, the theory of computing analysis and GPS-meteorology, more powerful tools can be used to perform watershed spatial data obtaining, storing, processing and analysis and distributed hydrological model has been attracting more and more attention from both engineering and academic communities. In the meantime, Cellular Automata as an inherent distributed computer might be integrated with distributed hydrological model, which may create a new way to improve watershed hydrological model development. Based on the analysis of the general disciplinarian and the particularity for the application of Cellular Automata in geography and hydrology respectively, according to the basic principle of hydrological model and Cellular Automata, a watershed routing model based on GIS and Cellular Automata, HydroCA is proposed in this paper. The local updating rule is the core of HydroCA, and is also the key in the process of uniting cellular automata with distributed hydrology models. The local rules are close interrelated to the model precision. The key of establishment of local rules in HydroCA is to determine flow direction and flux arrangement among the site and sites in the neighborhood. And the rules should be applied in whole the lattice space. Flow direction of the site determines flow lines and flow distribution among its Moore neighborhood. For the moment, usual methods to determine flow direction, are single steepest descent flow direction method, multiple flow direction algorithm, resultant vector methods and etc. Inputs of the model is divided into two classes, one includes core variables of lattice site, the other consists of assistant variables including DEM, topographic index(TOPMODEL), water supply ratio, M and etc. Core variables are computed using runoff production model. As for assistant variables, DEM is provided by geographic information system software and the other variables are calculated by special modules. Local rules are based on multiple flow direction algorithm, and Runoff flow course is the Output. A simulation software platform, Hydro_CA, by combining the HydroCA model for runoff routing with TOPMODEL / SCS model for rainfall-runoff simulation operated on the watershed DEM with 100* 100 m grid size was developed and used to simulate runoff process from 1 April 1998 to 31 August 1998 in the study area LeAnhe Watershed in the northeast of JiangXi Province, China. The results of the simulation prove that the precision of TOPMODEL simulation can be improved by using HydroCA routing model from 64.7% to 71.0%, while the precision of combining SCS simulation with HydroCA routing is slightly worse than that of combining SCS simulation with Unit Hydrograph routing (from 54.7% to 57.5%). It is shown that HydroCA model is feasible to simulate the watershed runoff routing. Cellular Automata model will contribute to the distributed watershed hydrological model research and development by providing a new method in calculating runoff forming and converging. It can simulate routing process depending on a mechanism from bottom to top or from local to global in the research area compositing the movement of water of the land-slope and channel. The runoff routing model on the basis of cellular automata is calculated in distributed way considering heterogeneity in both space and time. Therefore, a task left for us is to build a reliable distributed runoff generation model, which can provide the values of water condition variables as the temporal values for all cells in the lattice space during time intervals.