The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
Publications Copernicus
Articles | Volume XL-5/W3
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-5/W3, 1–6, 2013
Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci., XL-5/W3, 1–6, 2013
07 Jan 2014
07 Jan 2014


C. Mulsow, R. Koschitzki, and H.-G. Maas C. Mulsow et al.
  • Institute of Photogrammetry and Remote Sensing, Technische Universität Dresden, 01062 Dresden, Germany

Keywords: Photogrammetric network, image sequence processing, disaster monitoring

Abstract. The growing number of glacial margin lakes that have developed due to glacier retreat, have caused an increase of dangerous Glacial Lake Outburst Floods (GLOFs) in several regions over the last decade. A GLOF can occur when the water from the lake finds a path underneath the bottom of the glacier and the lake is draining rapidly. This causes normally a flood wave downstream the glacier. Typically such an event takes about 24 hours. GLOF scenarios may be a significant hazard to life, property, nature and infrastructure in the affected areas.

Together with our partner institute CECS (Valdivia, Chile), a project was initiated on a pilot study for an early warning system for GLOF events in the Northern Patagonian Icefield. A GLOF is normally characterized by a progressive water level drop. By observing the water level of the lake, an imminent GLOF-event can be identified. Common gauging systems are not suitable for the measurement task, as they may be affected by ice fall or landslides in the lake basin. Therefore, in our pilot study the water level is observed by processing images of a terrestrial camera system.

The paper presents the basic principle of a single-camera based GLOF early warning system. Challenges and approaches to solve them are discussed. First results from processed image sequences are presented to show the feasibility of the concept. Water level changes can be determined at decimetre precision.

In the first stage of the project, the waterline was measured manually in the images. A promising approach for reliable automation of this task is the use of a camera, which is sensitive for near infrared. The difference in the reflection of water, ice, and rock in this wavelength is more better than in RGB. This will be discussed in the outlook in deep.