A methodology was developed previously by the authors for the segmentation of the Global Digital Elevation Model (GTOPO30) to three terrain classes (mountains, basins and piedmont slopes) and it was applied to the Great Basin Section (south-west USA). In the present research effort, mountain objects were identified through a connected component-labelling algorithm applied on the mountain terrain class. Taking into account the physical and perceptual attributes of the Great Basin mountain features, 12 morphometric attributes were defined for the mountain objects and were used as descriptors in their parametric representation. Finally, classification of mountain objects through the implementation of a K -means clustering algorithm resulted in four clusters of mountain objects that appeared to be spatially arranged to distinct geographic regions. The results were compared with existing maps and they were found to be in accordance with existing physiographic descriptions. It is concluded that the derived parametric representation of mountain objects carried sufficient physiographic information and it can be used for mountain classification. The conclusions point out the physiographic information content of GTOPO30 and its value and applications to regional geology and space geomorphology.

A reference digital elevation model (RDEM) produced by the Hellenic Military Geographical Service (HMGS) (Athens, Greece) is used to assess the vertical accuracy of the ASTER (advanced spaceborne thermal emission and reflection radiometer) global digital elevation model (GDEM) for a small volcanic island, Nisiros, in south-east Greece. The overall vertical accuracy of the GDEM is within the root mean square error (RMSE) specified by the ASTER DEM Working Group. Despite this, ASTER GDEM is a product with variable vertical accuracy depending on the number of scenes (stack number (SN)) per grid point used for the computation of elevation. The exposure (SN) of a physiographic region to ASTER sensors seems to be elevation- and slope-dependent in this particular study area. The difference between the GDEM and RDEM may partly be explained by the fact that GDEM is a digital surface model and thus the variable landcover heights introduce additional error in accuracy estimates.