skip to primary navigation skip to content

Department of Geography

 

 

Semi-automated mapping of glacial landforms

High-resolution remotely sensed data are being used increasingly in glaciology to map aspects of the landscape and quantify current environmental change. Despite advances in data collection and analytical software, mapping of glacial geomorphology is still achieved largely through manual techniques. Manual mapping is a relatively slow, inefficient and potentially subjective process, particularly given the extensive spatial coverage of remotely sensed datasets. Thus, glacial geomorphic mapping would benefit from semi-automated techniques able to map large areas quickly and efficiently.

This project has developed a suit of novel, semi-automated techniques (using object based image analysis and cluster analysis) for mapping glacial geomorphology using a wide range of high-resolution data from differing settings. We have used airborne LiDAR gathered as part of a NERC ARSF grant to SPRI, and from Icelandic colleagues, to map the forefields of two Icelandic glaciers, and we have also used marine swath-bathymetry data to map the seafloor of a fjord in Svalbard. Comparison with existing manually classified maps shows a high degree of correspondence, and means that our algorithms should allow large areas to be classified much more quickly than using traditional techniques.

Publications

Papers stemming from this project so far:

  • Robb, C., Willis, I., Arnold, N. and Gudmundsson, S., 2015. A semi-automated method for mapping glacial geomorphology tested at Breiðamerkurjökull, Iceland. Remote Sensing of Environment, v. 163, p.80-90. doi:10.1016/j.rse.2015.03.007.

Figure

Forefield of Breiðamerkurjökull, an outlet from Vatnajökull, Iceland. The map to the left is an annotated hillshade of airborne lidar data used in the study. The black arrow shows the location and direction of the photograph shown on the right. The blue square shows a small area of the forefield where our semi-automatic mapping methodology was developed and tested. From Robb et al, 2015.

Figure

Training the semi-automated algorithms on a subset of lidar data from Breiðamerkurjökull. (A) Manually produced analogue map from Evans and Twigg (2002); (B) Lidar hillshade; (C) Manually digitised reference map from (B) guided by (A); (D) Map produced using semi-automated methods. From Robb et al, 2015.