Technical capacity information from satellites about the state of the surface of the ocean and seas, the development of methods and algorithms of information processing and mathematical modelling of water circulation, leads to the creation of algorithms for assimilation of such information in dynamic models. The parameters found in this case make it possible to obtain a model solution that best matches the measurements due to the minimization of the forecast quality functional used. Variational algorithms for solving various problems of identifying certain parameters of numerical simulation of passive impurity transfer are based on solving adjoint problems and implementing iterative procedures for finding optimal parameters from measurement data. Such parameters in the passive impurity transfer problem can be the initial data, the flows of matter at the bottom and surface, the power of point sources, the coefficients of turbulent diffusion, and the velocity fields. In this paper, this parameter is the particle sedimentation rate. they are deposited by gravity, which naturally depends on the particle size. However, the task of estimating some averaged values is possible on the basis of assimilation of operational information obtained from the satellite. The algorithm is based on solving the conjugate problem and the problem in variations. The problem is solved by searching for the minimum of the quadratic functional of the forecast quality, and the model acts as constraints when minimizing it. To construct the functional gradient, the solution of the corresponding adjoint problem is used. Expressions for determining the desired functional gradient in the parameter space are given.

В работе на основе вариационного алгоритма ассимиляции данных измерений построена процедура поиска скорости седиментации взвешенного вещества в море. Процедура реализуется для модели переноса пассивной примеси по данным измерений, распределенных по времени и пространству и поступающих с поверхности моря. В основе алгоритма лежит решение сопряженной задачи и задачи в вариациях. Задача решается за счет поиска минимума квадратичного функционала качества прогноза, а модель выступает в роли ограничений при его минимизации. Для построения градиента функционала используется решение соответствующей сопряженной задачи. Приведены выражения для определения искомого градиента функционала в пространстве параметров.