Semi-Automated Classification of Landform Elements in Armenia Based on SRTM DEM using K-Means Unsupervised Classification

Artak Piloyan,Milan Konečný

doi:10.1515/quageo-2017-0007

Artak Piloyan, Milan Konečný

Open Access

https://doi.org/10.1515/quageo-2017-0007

Copy DOI

Abstract

Abstract Land elements have been used as basic landform descriptors in many science disciplines, including soil mapping, vegetation mapping, and landscape ecology. This paper presents a semi-automatic method based on k-means unsupervised classification to analyze geomorphometric features as landform elements in Armenia. First, several data layers were derived from DEM: elevation, slope, profile curvature, plan curvature and flow path length. Then, k-means algorithm has been used for classifying landform elements based on these morphomertic parameters. The classification has seven landform classes. Overall, landform classification is performed in the form of a three-level hierarchical scheme. The resulting map reflects the general topography and landform character of Armenia.

Highlights

Information about landforms is crucial, for instance, for studies of landscape evaluation and, assessment of potential erosion, natural hazards and risk prediction, as well as for regional planning of landscapes, cadastral assessment of land use etc
This paper presents a semi-automatic method based on k-means unsupervised classification to analyze geomorphometric features as landform elements in Armenia
Slope, plan and profile curvatures according to Evans (1972) and Shary (1995, Shary et al 2002) schemas and the flow path length according to Horton method (Horton 1945, Mitasova, Hofierka 1993, Zavoianu et al 2013) were used as a Digital Elevation Models (DEM)-based geomorphometric layers for semi-automated landform classification of the territory of Armenia

Summary

Introduction

Information about landforms is crucial, for instance, for studies of landscape evaluation and, assessment of potential erosion, natural hazards and risk prediction, as well as for regional planning of landscapes, cadastral assessment of land use etc. The classical ways to implement (combine) relief units on landscape assessment is to delineate them during ground truthing and/or by using aerial photographs Such an approach is relatively time consuming and the results are largely dependent on the researcher’s subjective assessment and, in many cases they may not have sufficient transparency and reproducibility (Dragut, Blaschke 2006). Terrain analysis in environmental assessments of landscape was quite rare until the last decade, while terrain topography is a key indicator in a wide range of environmental processes (Bates et al 1998, Butler 2001) Such a landscape approach is an important step in a variety of regional studies (Dragut, Blaschke 2006). Landscape pattern could have a crucial influence on the spread of disturbance (Turner 1990, Forman 1995, Butler 2001), water quality (Hunsaker et al 1992, Arakelyan, Piloyan 2011), environmental quality and biodiversity (Gordon et al 1994), as well as substantiate the mechanisms of material flow (Dalrymple et al 1968)

Objectives

Methods

Results

Conclusion