DOI: 10.5593/sgem2017/23/S10.006


A.D. Rukhovich, D.D. Rukhovich, D.I. Rukhovich, D.A. Shapovalov, A.L. Kulyanitsa
Monday 11 September 2017 by Libadmin2017

References: 17th International Multidisciplinary Scientific GeoConference SGEM 2017, www.sgem.org, SGEM2017 Conference Proceedings, ISBN 978-619-7408-03-4 / ISSN 1314-2704, 29 June - 5 July, 2017, Vol. 17, Issue 23, 41-46 pp, DOI: 10.5593/sgem2017/23/S10.006


For the analysis of various characteristics of natural objects distributions cluster analysis, principal component analysis and regression analysis are often used. There is a large number of distributions of the parameters of natural objects (e.g. spectral characteristics of soil), where the function is unknown, and discreteness is not visible, and contrary to the basic properties of described object. The soil cover is continuous in space and has no clear boundaries between spectral characteristics, such as soil lines coefficients. The task of this work was the development of methods for soil maps creation based on the multitemporal remote sensing data using the mathematics of elastic approximation. This mathematical algorithm was not previously used in soil mapping and for remote sensing data analysis. In the future, the method allow to create soil cover maps by teaching without a teacher. The obsolescence of archival maps requires the development of soil mapping methods for the reduction of ground surveys and those with predictive effects. Building of soil lines for the 34 Landsat images for the period from 1985 to 2015, allowed to obtain three coefficients for each element of image resolution. In the graphical representation, the three coefficients form a cloud of values that has not expressed nuclei and not represents a known mathematical function. But, this cloud has an elongated shape and, as stated earlier, different parts of the cloud characterize different soil types. As a result, a map of the spectral characteristics of the soil line consistent with large-scale soil maps was obtained.

Keywords: soil line, remote sensing data, elastic map method, soil map