Predictive soil cartographic materials as elements of modern large-scale surveys

new site

Даний сайт більше не оновлюється!
Новий сайт журналу знаходиться за адресою https://visnyk.lnup.edu.ua/

 

Visnyk LNAU: Agronomy 2019 №23: 202-206

Predictive soil cartographic materials as elements of modern large-scale surveys

Dmytruk Yu., Doctor of Biological Sciences
ORCID ID: 0000-0002-3157-0503
Cherlinka V., Doctor of Biological Sciences
ORCID ID: 0000-0001-5354-4851
Demyd I., researcher
ORCID ID: 0000-0001-5819-6390
Yuriy Fedkovych Chernivtsi National University,
Department of Agrotechnology and Soil Science

https://doi.org/10.31734/agronomy2019.01.202

Annotation

The current state of soil map data in Ukraine is considered and it is shown that significant territories (up to 33%) are not covered by large-scale soil surveys. A way is shown for obtaining maps of soil cover or cartograms of agro-industrial groups, the essence of which is the morphometric analysis of digital relief models, the selection of a number of predictors from them and the association of existing cartographic soil materials with these data by creating a mathematical predicative model using the reference points of the landscapes and soil taxons linked to them, which allows you to fill in unexplored areas with forecast data.

Comparison of the map data obtained by different algorithmic methods shows that Random Forest has the highest predictive power from all present in this study – 89,3 % vs. 54,8 % of the neural network and 76,8% in the k-nearest neighbor. The high degree of coincidence of forecast and cartographic data of the random forests’ algorithm within the research boundaries of the plotted areas gives rise to the use of the obtained materials in the areas of "white spots". In the absence of any variants of cartographic information, predicative variants can be used in the practice of agricultural production, the planning of fertilizer system, the development of a system of anti-erosion measures, etc. In the probable case of upcoming regular large-scale soil surveys, the maps-version obtained is an element that will allow a significant reduction and minimization of financial, time and labor costs for field surveys by introducing half-profile and cuts instead of full-profile variants in cases of coincidence of predicted and actually observed soil.

Key words

soil map, cartogram of agro-industrial groups of soils, predictive algorithms, simulation, morphometric parameters, DEM

Full text

pdf

Link

  1. Cherlinka V. R. Adaptation of large-scale soil maps to their practical use in GIS. Agrochemistry and soil science. Kharkiv, 2015. Iss. 84. P. 20–28.
  2. Cherlinka V. R. Variations in the predictive efficiency of soil maps depending on the methods of constructing training samples of predicative algorithms. Ecology and Noosphereology. 2017. Vol. 28, N 3–4. P. 55–71.
  3. Cherlinka V. R., Lobova O. V. Methodical approaches to the coordination of soil cartographic materials at the borders of administrative-territorial units of Ukraine. Scientific reports of NULES of Ukraine. 2018. N 6 (76). P. 1–15.
  4. Cherlinka V. R. Morphometric parameters of relief as a basis for predicative modeling of spatial distribution of soil variations. Agrochemistry and soil science. Kharkiv, 2017. Iss. 86. P. 5–16.
  5. Cherlinka V. R. The influence of resolution of digital relief models on the quality of predicative simulation of soil cover. Soil science. 2017. Vol. 18, N 1–2. P. 79–95.
  6. Cherlіnka V. Usіng Geostatіstіcs, DEM and Remote Sensіng to Clarіfy Soіl Cover Maps of Ukraіne. Soіl Scіence Workіng for a Lіvіng: Applіcatіons of soіl scіence to present-day problems / Ed. by Davіd Dent, Yurіy Dmytruk. Cham, Swіtzerland: Sprіnger-Verlag GmbH, 2017. Part 2, chapt. 7. P. 89–100.
  7. Cherlіnka, V. Usіng Geostatіstіcs, DEM and Remote Sensіng to Clarіfy Soіl Cover Maps of Ukraіne. Soіl Scіence Workіng for a Lіvіng: Applіcatіons of soіl scіence to present-day problems / Ed. by Davіd Dent, Yurіy Dmytruk. Cham, Swіtzerland: Sprіnger-Verlag GmbH, 2017. Part 2, chapt. 7. P. 89–100.
  8. EasyTrace group. Easy Trace 7.99. 2015. Dіgіtіzіng software. URL: http://www.easytrace.com (Last accessed: 01.05.2019).
  9. GRASS Development Team. Geographіc Resources Analysіs Support System (GRASS GІS) Software. Versіon 7.2. 2017. URL: http://grass.osgeo.org (Last accessed: 01.05.2019).
  10. Landіs J. R., Koch G. G. The Measurement of Observer Agreement for Categorical Data. Bіometrіcs. 1977. Vol. 33, N 1. P. 159–174.
  11. Lіu B. Web Data Mіnіng: Explorіng Hyperlіnks, Contents and Usage Data. London; New York; Dordrecht: Sprіnger-Verlag GmbH, 2011. 622 p.
  12. QGІS Development Team. QGІS Geographіc Іnformatіon System. 2015. URL: http://qgіs.osgeo.org (Last accessed: 01.05.2019).
  13. R Development Core Team. R: A language and envіronment for statіstіcal computіng. R Foundatіon for Statіstіcal Computіng. 2017. URL: http://www.r-project.org (Last accessed: 01.05.2019).
  14. Rіpley B., Venables W. R-package nnet: Feed-forward neural networks and multіnomіal log-lіnear models. v.7.3-12. 2016. URL: https://cran.r-project.org/package=nnet (Last accessed: 01.05.2019).
  15. The Shuttle Radar Topography Mission (SRTM), NASA. U.S. Releases Enhanced Shuttle Land Elevation Data. 2015. URL: http://www2.jpl.nasa.gov/srtm (Last accessed: 01.05.2019).
  16. Wright M. N., Ziegler A. Ranger: A fast implementation of random forests for high dimensional data in C++ and R. Journal of Statistical Software. 2017. Vol. 77. № 1. P. 1–17.
титулка Агро