Physically-based segmentation of the Western Carpathians (Central Europe)

Abstract

Results of a physically-based methodology to delineate morphometrical-morphostructural subdivision of the Western Carpathian region (Central Europe) from DEMs and their derivatives are presented. Previous suitability evaluation of an object-oriented methodology showed its potential in recognition of morphostructural features. In this study we moved towards a more complex object-oriented approach – fusion of segmentation and classification on several hierarchical levels. In addition, physicallybased geomorphometric variables were used as input data, resulting in enhancement of subsequent morphotectonic interpretations. Decrease of local variance of the delineated objects in comparison with simple segmentations without these upgrades confirms the efficiency of our approach.

Sensitivity of multiresolution segmentation to spatial extent

“…Local optimization methods led to up to 1.5% increase in overall accuracy as well as improvements in the geometric accuracy of the image objects. Comparable findings were also reported by Bandura et al (2018), where partitioning of the complex full scene into more homogeneous domains led to delineation of geomorphological features, which vary in shape and homogeneity properties. The authors locally adjusted SP for each domain, which, complemented with subsequent removal of distinct geomorphological features, positively affected the overall level of homogeneity in the MRS.…”

Physically-based land surface segmentation: Theoretical background and outline of interpretations

“…First, morphostructural segmentation of the Western Carpathians ( Fig. 3) uses data and techniques described in [15]. From the variables used for morphostructural segmentation we derive the Index of Steady State (ISS).…”

Physically based morphostructural land surface segmentation: Case of the Alps and Western Carpathians

“…This work is also aimed at delineation of objects at regional scale primarily formed by endogenous geomorphic processes, emphasizing their physical-geomorphic definition and interpretation. The OBIA approach was selected as it proved successful in our task (Bandura, Minár, & Drăguţ, 2018).…”

“…The presented research picks up on the work of Bandura, Minár, Drăguţ, and Harciníková (2017), focused mainly on the segmentation aspect using regular geomorphometric variables and simple segmentations, and Minár et al (2018), where three physically based geomorphometric variables enabling construction of the index of steady state (ISS) were introduced: endogenous geomorphic work (EnW), exogenous geomorphic work (ExW), and relief brake force (RBF). This physically based variant of morphostructural segmentation using OBIA and physically based geomorphometric variables was first tested in the territory of the Western Carpathians (Slovakia), where a specific object-based approach was developed (Bandura et al, 2018). A simple object-based approach refers to one simple segmentation, and usually, but not necessarily, is followed by classification.…”