Abstract
Circular structures on the ground are one of the characteristic features recognized on satellite images and aerial photographs. Many of the circular structures are comprised of arcuate topographic ridges, which represent the erosional remnants of caldera structures. Distinguishing circular structures is difficult and inconsistent on satellite images and aerial photographs because it is based on photogeological interpretation through human eyes.
To overcome this inconvenience, we have used digital elevation models to detect circular structures, focusing especially on the topographic expression of volcanic calderas that have been extensively modified by erosion. The detection of topographic calderas was conducted by using a simple algorithm that we developed on the basis of topographic characteristics such as the topographic rim of the caldera, which shows a circular pattern, and the direction of maximum slope of the rim, which consists of steep cliffs sloping toward the center of the caldera. By using this procedure, the lines drawn from the cliff of the caldera along the direction of the maximum slope will concentrate in the center of the caldera. Based on this result, score maps are constructed by measuring the density of lines along the above mentioned maximum slope directions.
By applying our algorithm to downward and upward directions along the maximum slope, we were able to automatically detect the central areas of low ground and areas of circular topographic form, including topographic calderas. This could be achieved without having to use photogeological interpretation and statistic analysis obtaining from satellite images and aerial photographs, even if the outline of a topographic circular feature was broken into separate segments by surface erosion.
This algorithm presents the disadvantage that it also identifies as high scoring areas, those areas with topographic domes and flat planes formed by Quaternary lavas and pyroclastic flows which are characteristic topographic features of Quaternary volcanoes.
