Automated Mapping of Rock Glaciers Using Deep Learning and OBIA
Rock glaciers can contain significant amounts of ice and are found typically in arid and semi-arid catchments. These are areas with limited water resources already, with climate change in some cases causing shrinking glaciers and altered precipitation patterns. Rock glaciers can therefore be considered a potentially important future water resource and should be inventoried and monitored.
The methods typically used to delineate glacial landforms however do not translate well to rock glaciers due to their spectral similarity to the surrounding terrain. We will present a method that was developed within eCognition that combines deep learning (or Convolutional Neural Networks, CNNs) with Object Based Image Analysis (OBIA) to identify rock glaciers from 10 metre resolution Sentinel 2 imagery.
Date: Wednesday September 17
Time: 9:00 am CDT
Presenters: Keith Peterson (Trimble eCognition) & Dr. Benjamin Robson (University of Bergen)
To attend this webinar, please register below.