“Similar to earthquakes, landslides are next to impossible to predict,” explained Daniel Hölbling, a research scientist at the University of Salzburg, in a recent article entitled “Shifting Approaches to Landslides” published in The American Surveyor.
Hölbling has been researching landslide-applications since 2009, using a range of remote sensing data in combination with Trimble’s eCognition software to develop automated approaches for landslide detection, mapping, inventory and monitoring – tasks that have been dominated by manual image analysis methods.
Hölbling goes on to state that “OBIA, with remote sensing data, is the most powerful tool for detecting and analyzing landslides…You can integrate spectral, spatial, morphological, textual, and contextual properties in one interlinked framework. All that diverse data enables the software to mimic how the human brain identifies and categorizes objects, making it far superior to traditional pixel-based approaches which can’t do that.”
The article presents several landslide projects that Hölbling and his team have investigated throughout the world, including New Zealand, Iceland and Taiwan, all landslide prone areas that will benefit greatly from his research.
In New Zealand, local governments worked hard to create a detailed landslide inventory. But this work is currently based on traditional, manual analysis – “visually interpreting each aerial or satellite image and manually delineating and mapping each landslide identified”. Needless to say, this work is “tedious, slow and subjective”.
In 2016, Hölbling partnered with researchers in New Zealand to test his semi-automated eCognition approach to identify landslide-prone hotspots in a 1000 hectare study area. The NZ-based researcher manually digitized visible landslides on each orthophoto in GIS – the work took them 2 weeks. For comparison purposes, Hölbling worked in parallel – it only took a single day to prepare Trimble’s eCognition software for integrating the datasets and classifying the landslides. “In a few hours, eCognition classified all visible landslides across all five time stamps.”
Hölbling’s work will continue within the MORPH project until 2019, so a complete assessment of the methodology is currently not possible, but he is optimistic given the current results.