DETERMINATION OF BATHYMETRIC AND HYDROGRAPHIC PARAMETERS OF WATER BODIES THROUGH NEURAL NETWORKS.

Abstract

Abstract. Currently, there is an increase in the active use of artificial intelligence systems for solving urgent, scientific and technical problems of various types. Such tasks include the field of remote sensing of the Earth.

As applied to remote sensing problems, approaches based on the use of neural networks can be conditionally classifiedas algorithms that implicitly use physical and non-physicalquantities, i.e. measurement data from satellite instruments.In other words, a function that connects input values and an output result (for example, the probability of a certain class in a classification problem) is approximated by these algorithms, unlike other approaches in which this function is usually set explicitly. An example of an explicit physical approach is the widespread use алгоритмов of classification algorithms based on spectral analysis and threshold methods, when physical quantities (for example, the spectral brightness ratio, brightness temperature, etc.) or spectral indices are estimated using threshold values set for each class [1]. This approach requires taking into account complex dependencies to obtain more accurate results withan unknown analytical form. However, with the use of neural networks, such difficulties can be overcome.

The purpose of the article is to highlight the primary results of using a neural network in determining bathymetric and hydrographic parameters of water bodies.