IDENTIFICATION OF PHYTOHORES WITH TRAINING ON MIXED FIELD AND SPACE DATA (USING THE EXAMPLE OF THE NATURAL TERRITORY OF THE BOTANICAL GARDEN OF PetrSU)

DOI: 10.24057/2414-9179-2018-2-24-186-194

View or download the article (Rus)

About the Authors

Andrey V. Korosov

Petrozavodsk State University,
Lenina str., 33, 185940, Petrozavodsk, Russia,
E-mail: korosov@psu.karelia.ru

Elena A. Platonova

Petrozavodsk State University,
Lenina str., 33, 185940, Petrozavodsk, Russia,
E-mail: meles@sampo.ru

Abstract

Joint quantitative processing of space and field data method has been proposed in order to perform thematic classification of satellite image pixels. Traditional decoding algorithm classification is carried out separately from the field verification of obtained clusters. The proposed algorithm combines these procedures in component analysis framework. The task is to identify phytohor in the vegetation cover of Petrozavodsk state University Botanical garden natural territory. The parameters of 157 geobotanical descriptions of forest vegetation of this territory were analyzed: the number of species, the number of specimens of four tree species (Pinus sylvestris, Picea fennica, Populus tremula, Betula pubescens and Betula pendula), heat supply index, shade index and brightness values of satellite images. Given parameters has quantitative form. The analysis algorithm includes the step of combining geobotanical descriptions and brightness characteristics of the image pixels that cover the description area. Jointed component analysis of brightness value and geobotanical characteristics allows evaluating the contribution of these parameters to the main components. By the size of the loads, phytohor variant is selected according to certain component. Using the resulting loads, component values are calculated for the entire territory. Point selection corresponding to the selected phytohors is allows to define a narrow range of pixel brightness and perform pixels selection corresponding to the desired areas. Phytohor “monodominant green-skinned, mainly cowberry-blueberry pines” occupies about 17 % of the entire range of component values. Using the classification procedure with training on the basis of the principal components method for the territory of PetrSU Botanical garden, a contour map of pine tree type distribution was obtained.

Keywords

cosmonaut, classification with training, component analysis, phytochores, botanical garden PetrSU

References

  1. Cyganov D.N. Phytoindication of ecological regimes in the subzone of coniferous-broadleaf forests. M., 1983. 196 p. (in Russian).
  2. Efimov V.M., Galaktionov Yu.K., Shushpanova N.F. Analysis and prediction of time series by the principal component method. Novosibirsk: Nauka, 1988. 70 p. (in Russian).
  3. Phillips S.J. A Brief Tutorial on Maxent, Network of Conservation Educators and Practitioners. Center for Biodiversity and Conservation, American Museum of Natural History. Lessons in Conservation. 2009. V. 3. P. 108–135 (in Russian).
  4. Platonova E.A., Lantratova A.S. Variety of vegetative cover of the protected area of the Botanical Garden of PetrSU, Hortus botanicus. 2001. T. 1. P. 42–51 (in Russian).
  5. Program and methodology of biogeocenological research. Otv. red. N.V. Dylip. M.: Nauka, 1974. 403 p. (in Russian).
  6. Shitikov V.K., Rozenberg G.S. Randomization and bootstrap: statistical analysis in biology and ecology using R. Tol’yatti: Kassandra, 2013. 314 p. (in Russian).

For citation: Korosov A.V., Platonova E.A. IDENTIFICATION OF PHYTOHORES WITH TRAINING ON MIXED FIELD AND SPACE DATA (USING THE EXAMPLE OF THE NATURAL TERRITORY OF THE BOTANICAL GARDEN OF PetrSU). Proceedings of the International conference “InterCarto. InterGIS”. 2018;24(2):186–194 DOI: 10.24057/2414-9179-2018-2-24-186-194 (in Russian)