Application of Object Based Image Analysis for Glacial Cirques Detection. Case Study: The Ţarcu Mountains (Southern Carpathians)

Abstract

Geomorphologic mapping is an important fact in many research studies and the traditional methods are time consuming and expensive. This paper aims to develop a semi-automated rule-based method for the detection of glacial cirques for a test area located in the Ţarcu Mountains (Southern Carpathians) in an objectoriented approach. In this study we have established the morphometric characteristics of the glacial cirques developed in a particularly geomorphologic context at the edge of planation surfaces, using a 10 m horizontal resolution DEM (Digital Elevation Model). The parameters extracted from DEM (i.e. curvature) were further used in segmentation and classification process. Also, other factors were introduced in the rule set, as the context regarding neighboring objects like planation surfaces to the target class. The most important factor in segmentation was the curvature and to choose an appropriate scale factor we have used the available ESP (Estimation of Scale Parameter) tool. The results achieved were very close to the field reality, except for some areas where there are large negative landforms such as gullies and torrents, which were identified as objects belonging to glacial cirques class and also some roches moutonnees with high positive curvature values, objects that could be filtered manually by the user based on previous field knowledge and ancillary data such as orthophotoplans and the geomorphologic map of glacial relief in the Ţarcu Mountains. For further research, we intend to identify the characteristic thresholds for morphometric parameters that can be integrated in a set of rules in order to detect and classify other type of landforms in the alpine domain of the Ţarcu Mountains.