Visual interpretation of stereoscopic NDVI satellite images to map rainfall-induced landslides

Federica Fiorucci, Francesca Ardizzone, Alessandro Cesare Mondini, Alessia Viero, Fausto Guzzetti, 2019, Visual interpretation of stereoscopic NDVI satellite images to map rainfall-induced landslides, Landslides (Berl., Print) (2019).,
URL: http://www.cnr.it/prodotto/i/393004

Landslide inventory maps are commonly prepared through the visual interpretation of stereoscopic aerial photographs and field checks. Stereoscopic satellite images can also be interpreted visually to recognize and map landslides. When interpreting stereoscopic imagery, shadows can conceal the photographic elements typical of landslides, hampering the recognition and mapping of the landslides. To mitigate the problem, we propose a method that exploits normalized difference vegetation index (NDVI) images and digital stereoscopy for the 3D visual recognition and mapping of landslides in shadowed areas. We tested the method in the 25 km2 Pogliaschina catchment, northern Italy, where intense rainfall caused abundant landslides on 25 October 2011. Using a PLANAR® StereoMirror(TM) digital stereoscope, we prepared an event landslide inventory map (E-LIM) through the visual interpretation of a pair of NDVI images obtained from a WorldView-2 stereoscopic multispectral bundle. We compared the event inventory with two independent E-LIMs for the same area and landslide event. The 3D vision of the NDVI stereoscopic image pair maximized the use of the radiometric (color and tone) and the terrain (elevation, slope, relief, and convexity) information captured by the stereoscopic multispectral images, allowing for the recognition of more landslides and more landslide areas than the other E-LIMs in the shadowed areas. Our results confirm that use of NDVI images facilitates the visual recognition and mapping of landslides in terrain affected by shadows. We expect that the proposed method can help trained interpreters to map landslides more accurately in areas affected by shadows.

Dati tratti da https://intranet.cnr.it/people/