Application of the Genetic Algorithm to the Problem of Reconstruction of Missing Data

We study the problem of reconstruction (interpolation and extrapolation) of the vertical profiles of hydrochemical and hydrobiological elements according to incomplete sets of data with simultaneous filtration of short-period components based on expansions in empirical orthogonal functions. The genetic algorithm is used to compute the coefficients of expansion of profiles with missing data. We present the results of processing the data arrays on oxygen, chlorophyll A, and biogenic elements collected in the Black Sea in 1982–1993. The mean error of reconstruction of the profiles enables us to conclude that the proposed method has considerable advantages over the conventional approaches.     

Prediction of hydrological phenomena using self-organizing mathematical models

The paper considers the application of self-organizing models, specifically, the method of grouped arguments consideration (MGAC), to forecast short and non-stationary time series of observations in the ocean. A sequence of operations for the treatment of observational series is suggested. To assess its efficiency, we have used mean monthly oxygen concentration data collected in the surface and near-bottom layers of the Taganrog Bay. It is shown that the application of the MGAC model allows one to reduce by two times the root-mean-square error of that of the series prediction by five points, in comparison with the Jenkins-Box regressional model. It has been concluded that the predictors' non-linear functions may be effectively used in the treatment of short samplings. Translated by Vladimir A. Puchkin.     

Comparative analysis of the experiments aimed at modeling marine ecosystems by using the object-oriented approach

We describe the results of comparative analysis of the numerical experiments carried out by using two different approaches to modeling the biological components of marine ecosystems: “ordinary” and object-oriented. These results demonstrate that the character of the space distribution (patchiness) of the food resource on the lowest level of the trophic chain strongly affects the growth rate of the density of population on the next level and forms specific features of its space distribution which, in fact, affect the productivity of all levels of the trophic chain. The representation of phytoplankton (lowest level) in the model in the form a continuous field of passive admixtures creates more favorable conditions for the nutrition of the organisms consuming this resource. Therefore, in our calculations based on the “ordinary” procedure of modeling of marine ecosystems, the productivity of all its living components is higher than in the case of application of the object-oriented method. The indicated distinctions are especially well visible on the scales of about ten kilometers. Translated from Morskoi Gidrofizicheskii Zhurnal, No. 1, pp. 26–35, January–February, 2009.     

Object-Oriented Model of Functioning of the Plankton Community of the Shelf

We propose to study the ecosystem of the plankton community of the euphotic shelf zone as a complex system with five hierarchical levels. To construct the computer image of the ecosystem, we use the methods of object-oriented modeling. The vital activity of the organisms of each species inside a certain conventional volume of water (basic object of the model) is described in detail. As a result, the entire system simulates the interaction of species in the analyzed ecosystem. The system combines four groups of biological elements (phytoplankton, bacteria, protozoa, and zooplankton) and biomineral elements (organic and mineral phosphorus) and includes the equations for the temperature of water and dissolved oxygen. The proposed numerical model reproduces the observed spotted character of the space and time distributions of the density of aquatic life and its activity. The data of the performed numerical experiments reveal the correspondence between the integral quantitative parameters computed in the model and the average characteristics obtained under natural conditions.__________Translated from Morskoi Gidrofizicheskii Zhurnal, No. 6, pp. 43–61, November–December, 2004.     

A model of the dynamics of the ecosystem of the Sea of Azov

We use many-year archival observations of hydrologic and hydrobiologic state of the Sea of Azov for the identification of the structure of a model of its ecosystem. The ecosystem model, supplemented with expert estimates of bioresource consumption (fish reserve), contamination level, and possible ecologic fines for violation of the sea natural state, is formalized by the method of system dynamics. Hereat, the major part of influence functions is found in terms of observational data with application of self-organization algorithms. We also present the results of simulated experiments with the model of the ecosystem, which enable us to analyse scenarios of its behaviour under the influence of various external factors (wind, river discharge, water exchange with the Black Sea, etc.). Translated by Peter V. Malyshev and Dmitry V. Malyshev     

Analysis of the response of a non-linear data assimilation algorithm in a one-dimensional model of the upper ocean

A one-dimensional (1D) integral dynamico-stochastic model of the upper ocean with a non-linear assimilation algorithm is considered. The accuracy of computing the characteristics of the upper layer depends essentially on the values of the empirical coefficients. A numerical experiment was carried out which verified the efficiency of the model's adaptive mechanism operation when different values of the empirical coefficients and their variances were preset. Recommendations on assigment of the model's initial parameters are derived.Translated by Mikhail M. Trufanov.     

Prediction of time series by the method of analogs

We consider an algorithm of prediction of nonstationary time series based on the method of analogs. Since the exhaustion of a great number of versions is required for the adjustment of the parameters of the optimal prognostic model, we describe a genetic algorithm used in this case. We consider several procedures of construction of prognostic models. The numerical results are used to choose the procedure guaranteeing the minimum mean square error. The parameters of the model affecting the quality of predictions are determined. The proposed method is tested by using the reanalysis data (NCEP/NCAR project) on the anomalies of the monthly average surface air temperature for 58 yr. The results of predictions are compared with the estimates obtained by the linear regression method. It is shown that the method of analogs gives satisfactory results even in the cases where the regression methods lead to errors equal to the variance of predicted series. __________ Translated from Morskoi Gidrofizicheskii Zhurnal, No. 4, pp. 70–80, July–August, 2007.     

Dynamic-stochastic model for assimilation of remotely-sensed sea surface temperature data

We consider the possibility of assimilating sea surface temperature data by a combined ocean dynamics model incorporating a three-dimensional hydrodynamic model for synoptic variability for the considered ocean area and an integral model block describing the upper mixed layer. Model numerical simulations and experiments involving satellite-derived data are analysed.Translated by V. Puchkin.     

Reconstruction of the surface temperature fields according to the fragmentary data of remote sensing

We propose a gap-filling method for the data of remote sensing of the hydrophysical and biological characteristics of the water surface. The proposed method of reconstruction is based on the representation of the fields of surface characteristics as the sums of certain numbers of empirical orthogonal functions (EOF) making the largest contributions to the total variance of the field. According to the fragmentary data obtained as a result of processing of the satellite images for the summer season, we construct estimates of the mean field and of the four-dimensional space covariance function of the surface temperature of the Black Sea. The coefficients of expansion are computed by the method of least squares or determined with the help of a genetic searching algorithm. The results of numerical experiments show that the proposed method is quite promising for applications in the problems of gap filling in the available satellite data.