Application of information divergences to the analysis of geosystems and processes using remote sensing data

• Main Authors: Artiushenko, Mykhailo, Porushkevych, Anatolii
• Format: Article
• Language: Ukrainian
• Published: Scientific Centre for Aerospace Research of the Earth Institute of Geological Science National Academy of Sciences of Ukraine, Kyiv, Ukraine 2023
• Subjects: aerospace monitoring; images of geofield dynamics; difference characteristics; probability measures; entropy; informational divergences; analysis of peatland temperature field; аерокосмічний моніторинг; зображення динаміки геополів; характеристики відмінності; ймовірність; ентропія; інформаційні дивергенції; температурне поле торфовища
• Online Access: https://ujrs.org.ua/ujrs/article/view/232

Visual observation of changes occurring in the states of objects and processes on the Earth's surface is successfully solved through aerospace monitoring. Further improvement of information technologies of monitoring is associated with automating processing and interpretation of dynamic data represented by digital images. The article substantiates and demonstrates using of information characteristics of differences in images of geofield: a probabilistic measure, Gibbs-Shannon and Renyi entropy. Examples of calculation of various functionals of distributions of physical quantities represented by digital images and characterizing the degree of their proximity are given: Kullback divergence, alpha-divergence or Renyi divergence. The considered approach is illustrated by the example of calculating the informational divergences of the peatland temperature field obtained as a result of data processing using images taken by Landsat 8 satellite. The results of computer simulation of the above example show a significant dependence of the considered divergence measures on the spatial resolution of the field images. For the correct calculation of informational divergences, it is necessary to use images obtained with the same spatial resolution. Further development of information divergence methods is associated with introducing of scale-invariant measures. This will make it possible to use sensors with different spatial resolutions in the aerospace monitoring system to determine dynamic geosystem changes and processes.