Review of the current methods used to assess the values of coefficients, sensitivity, and adequacy of simulation models of aquatic ecosystems
摘要
The major principles, current methods, and approaches to solving optimization problems in assessing parameters for simulation models of aquatic ecosystems and their sensitivity to changes in the values of model parameters are considered. Analysis of the practical application of different methods to model calibration shows that no method of global optimization for complex hydrological models and models of aquatic ecosystems can guarantee obtaining a unique and best solution. Some results are shown, which have been obtained in solving the problem of the choice of parameters for the biohydrochemical block of a simulation model of Neva Bay (NB), the Gulf of Finland. The process of solving the optimization problem is formalized in the form of a two-stage direct-search algorithm: a random search algorithm (global search in the entire parameter space), followed by a local search by a modified Nelder-Mead simplex method.