Study On Parameter Estimation And Sensitivity Analysis Algorithms Based On Hydrological Model And Ecological Model

In the past half century, the mathematical modeling theory obtains a rapid development with the development of the computer. Mathematical modeling methods become important means to solve practical problems. However, with the development of various fields, the application of mathematical models show strong trend of complicated, such as multi-dimension, multi-apex,(lots of) nonlinear, discontinuity and non-protruding in mathematical modeling, which brings great challenges to the modeling work. In this paper, we combine mathematical modeling, mathematics, statistics, optimization calculation and computer science to the investigation of parameter estimation and sensitivity analysis associated with the hydrological model, using both theoretic arguments and numerical experiments. This paper also elaborates the parameter estimation algorithm and sensitivity analysis algorithm, as well asthe results of the simulation.Among the existing meta-heuristic algorithms, a relatively popular optimization method is the Shuffled Complex Evolution at the University of Arizona (SCE-UA), The paper describes the process of the SCE-UA algorithm in detail, proposes the theoretical support for SCE-UA which can be applied in Hydrus-1D. Then the SCE-UA algorithm is applied to the Hydrus-ID model. We can get that1.parameter estimation is a critical step in mathematical modeling and simulation. Some parameters of mathematical models can’t be determined by observations directly, only can be estimated indirectly.2. The application of the algorithm improves the confidence of the Hydrus-1D model parameters.3.SCE-UA is proved to be a feasiblestable and efficient method by the successful application in the modeling.In addition, the algorithms with theoretical support are applied to the coupling model of Hydrus-1D and Wofost. The Morris algorithm and Sobol algorithm are proved to be versatile, practical and feasible by successfully applying the Morris algorithm and Sobol algorithm to the model coupling, then reasonablyget a conclusion thatsensitivity analysis algorithm can improve the reliability of the model prediction and consummated the model theory as well. It forecast the distribution of the mathematical model results by rational design. The sensitivity analysis not only can test the coupled model behavior but also shows the effect of the coupled model parameters, climatic and environment scenarios on crop output.