Landslides are a natural hazard posing a threat to human life, property, infrastructure, and natural environment. Every year, slope instabilities have a significant impact on societies and economies. Consequently, landslide documentation is used for risk assessments, policy making and enforcing of construction regulations. Landslide monitoring is used to ensure safety of infrastructure operation. Rapid mapping of landslides and associated damages is done for response actions, e.g. of civil protection organizations. As ground surveys are very costly and time-consuming, satellite remote sensing is increasingly used to assess damage resulting from landslides.
Landslides lead to local terrain changes after a downslope movement of material under the effect of gravity. They vary by type of movement (e.g. falling, toppling, gliding and flowing), by size (from small rocks to entire mountain slopes) and velocity (from a couple of millimetres per year up to free-fall speed). Landslides can be triggered both by natural causes (like earthquakes or heavy rainfall events) and human causes, e.g. mining activities that lead to slope failures. Landslides can initiate other natural hazards, e.g. when a landslide blocks a river a lake can be formed which poses a risk for an outburst flood.
Landslides are diverse in appearance, and therefore are challenging to detect. EO-based assessment methods aim for detecting changes to the land surface and surface displacements.
EO satellites and airborne remote sensing use optical sensors for detecting landslides in post-event images and land cover changes caused by landslides, primarily indicated by the removal of vegetation and the exposure of bare soil, by comparing pre-event and post-event images. Typical resolutions of optical EO data for mapping rapid landslides are between 0.4 m and 30 m, depending on the size of landslides caused by the triggering event. Optical data from unmanned aerial vehicles are used in cases where single landslides or concise regions have to be covered. Additionally, synthetic aperture radar (SAR) sensors allow the detection of subtle changes in ground deformation caused by landslides. Therefore, time-series of radar images are used. Further, airborne laser scanning enables the generation of digital elevation models (DEMs) that allow identification of landslide surface structures and, in case of repeated coverage, detection of elevation changes. DEM generation for analysing landslides is also possible with photogrammetry on stereographic optical data and radargrammetry on SAR images.
The diversity of appearances of landslides leads to challenges for (semi-)automatic image processing and makes visual interpretation of EO data by a landslide expert a commonly used method for landslide mapping. However, visual interpretation is subjective and experts’ results can be very diverse. Additionally, it is a slow and time-consuming process. Semi-automated classification based on optical and DEM data using object-based image analysis (OBIA) can achieve detailed interpretations of landslides while reducing the analysis time. Interferometic SAR (InSAR) techniques, such as persistant scatterer interferometry (PSI) or Small Baseline Subset (SBAS), are primarily used to identify and monitor slow-moving landslides and for quantifying movement rates. Integrated analysis of optical, DEM and SAR data allow to fully exploit the potential of EO data from different sensors for landslide mapping and assessment.