View or download the article (Rus)

About the Author

Peter Yu. Litinsky

Karelian Research Centre of the Russian Academy of Sciences,
Pushkinskaya str., 11, 185910, Petrozavodsk, Russia,
E-mail: litinsky@sampo.ru


A new approach to the use of the Landsat images based on 3D modeling of their spectral space is described. The model is built in the x-y-z axes, where x and y are the two first principal components of the image matrix (bands R, NIR, SWIR2) in logarithmic form and z is moisture stress index MSI (SWIR1 / NIR). Empirically found transformation is more suitable for boreal ecosystems than Tasseled Cap. Segmentation of the spectral space by the method of ellipsoids provides more accurate, in comparison with traditional methods, the delineation of the ecosystem contours. It is shown that in the spectral space of images of the northern taiga subzone of Eastern Fennoscandia all the main types of biogeocenoses (complexes of Quaternary deposits + vegetation) are represented, and their localization corresponds to ecological typology. The ecological series of forest automorphic and hydromorphic ecosystems are clearly visible, as well as the trajectories of reforestation after cutting, from the appearance of vegetation to young growth, middle-aged and mature forests. For mires, the localization of spectral segments corresponds to the type of water and mineral nutrition (oligotrophic or mesotrophic). The fundamental difference between this model and those created by traditional methods is that the result is determined not by the layout of training sites, but by an objective biophysical parameter, i.e. the position of the ecosystem in the spectral space. The spectral model is a mathematically formalized object which describes the quantitative and qualitative characteristics of the ecosystems. Being deployed in geographic space, it becomes an optimal cartographic basis for planning ecologically balanced nature management for sustainable development.


geomatic modeling, boreal ecosystems, Landsat images classification.


  1. Biotic diversity of Karelia: conditions of formation, communities and species. Petrozavodsk: Karelian Research Centre of RAS. 2003. 244 p.
  2. Cohen W.B., Spies T.A., Fiorella M. Estimating the age and structure of forests in a multiownership landscape of western Oregon, U.S.A. Int. J. Remote Sensing. 1995. V. 16, No 4. P. 721–746.
  3. Hirata Y., Takahashi T. Image segmentation and classification of Landsat Thematic Mapper data using a sampling approach for forest cover assessment. Can. J. For. Res. 41(1). P. 35–43. DOI: 10.1139/X10-130.
  4. Huang C., Yang Wylie L., Homer Collin, Zylstra G. Derivation of a tasselled cap transformation based on Landsat 7 at-satellite reflectance: USGS Staff—Published Research. 2002. Paper 621. http://digitalcommons.unl.edu/usgsstaffpub/6
  5. Kanellopoulos I., Wilkinson G.G. Strategies and best practice for neural network image classification. International Journal of Remote Sensing. 1997. 18(4). P. 711–725.
  6. Kauth R.J., Thomas G.S. The Tasseled Cap a graphic description of the spectral-temporal development of agricultural crops as seen by Landsat. In: Proceedings on the Symposium on Machine Processing of Remotely Sensed Data, 4b: 41–51, 6 June 2 July 1976 (West Lafayette, Indiana: LARS, Purdue University).
  7. Krankina O.N., Pflugmacher D., Friedl M., Cohen W.B., Nelson P., Bacini A. Meeting the challenge of mapping peatlands with remotely sensed data, Biogeosciences Discuss. 5. 2075–2101. DOI: 10.5194/bgd-5-2075-2008.
  8. Kryshen A., Litinsky P. Comparison and mutual verification of the geoinformation and the ecological dynamics models of forest ecosystems diversity. Trudy KarNTS RAN. 2013. No 2. P. 86–91. http://forestry.krc.karelia.ru/publ.php?id=10572&plang=e (in Russian).
  9. Litinsky P. Multispectral imagery classification method based on spectral space modeling. Trudy KarNTS RAN. 2011. No 5. P. 45–54. http://forestry.krc.karelia.ru/publ.php?id=8809 (in Russian).
  10. Litinsky P. Geoinformation model of Eastern Fennoscandia northern taiga ecosystems. Trudy KarNTS RAN. 2012. No 1. P. 3–15. http://forestry.krc.karelia.ru/publ.php?id=9352&plang=e (in Russian).
  11. Litinsky P. Geoinformation Model of Terrestrial Ecosystems of the White Sea Lowland. Trudy KarNTS RAN. 2016. No 3. C. 3–9. DOI: 10.17076/bg221 (in Russian).
  12. Pignatti S., Box E.O., Fujiwara K. A new paradigm for the XXIth century. Ann. Bot. 2002. V. 2. P. 3057.
  13. Puzachenko Yu.G., Aleshchenko G.M., Molchanov G.S., Puzachenko A.Yu. Analysis of aerophoto images for distinguishing the types of territorial structures. Proc of 2nd all-Russian conference “Aerospace methods and geoinformation systems in forest science and forestry economy”. M., 1998. P. 156–159.
  14. Richards J.A., Xiuping Jia. Remote Sensing Digital Image Analysis. Berlin, Springer, 1999. 400 p.
  15. Shatalov A.V., Zhirin V. M., Sukhikh V.I. et al. Analysis of the information content of highresolution QuickBird images. Int Conf “Aerospace methods and geoinformation technologies in forest science and forestry”. M., 2007. P. 168–174.
  16. Volkov A.D., Gromtzev A.N., Erukov G.V. et al. Ecosystems of north-west taiga landscapes (structure, dynamics). Petrozavodsk: KarNTS RAN, 1995. 194 p. (in Russian).
  17. Zamyatin A.V. Analysis of landscape cover dynamics on the basis of data of remote sensing of the Earth. Research of the Earth from Space. 2006. No 6. P. 50–64.
  18. Zhou L., Yang X. Use of neural networks for land cover classification from remotely sensed imagery, The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. V. XXXVII, part B7. Beijing. 2008. P. 575–578.

For citation: Litinsky P.Yu. 3D MODEL OF THE SPECTRAL SPACE OF LANDSAT IMAGES AS THE BASIS OF THE BOREAL ECOSYSTEMS GEOMATIC MODEL Proceedings of the International conference “InterCarto. InterGIS”. 2018;24(2):116–128 http://doi.org/10.24057/2414-9179-2018-2-24-116-128