Optimal Inverse Method Applied For Determination Of Soil Infiltration Parameter And Manning's Roughness Coefficient

The surface irrigation simulation model is widely used for evaluation of the performances of the on-farm surface irrigation system, which is the important measure for improvement of the conventional surface irrigation technology. The application of the simulation model needs to know the feasible soil infiltration parameters and the Manning's roughness coefficient. The main aim of this dissertation is to build up an inverse system that can optimize the soil infiltration parameters and the Manning's roughness coefficient through the comparison of the data from the simulation and the observation for given irrigation event. Thus, the following works have been done:(1) Three types of the object function for the inverse method are set up according to the characteristic of water flowing under the surface irrigation. Only considering the advance course, the advance data from the simulation and observation are used to calculate the objective value. Only considering the recession course, the recession data from the simulation and observation are used to calculate the objective value. Considering the advance and the recession courses, both the advance data and recession data from the simulation and observation are used to calculate the objective value.(2) Since there is no concrete mathematical relationship between the object function value and the optimized parameters, the Genetic Algorithm has been applied in this work.(3) The Computer Analysis Technique is used to analyze the structure of surface irrigation simulation model srfr firstly, and then the srfr is connected with the Genetic Algorithm model, to consist of self-contained parameter optimization inverse program.(4) Examples have been calculated by use of parameter optimization inverse program. The results from different types of the object function are compared. Then the optimization results are compared again with the results from optimization model OASP and trial and error method. It is shown that the relatively better result can be obtainedwith the third object function namely considering both the advance and recession course. The results of the best fitting from the different methods indicate that differences are not obvious.