Expert Landslide and Shaded Area Delineation System
Landslides are destructive geological processes that cause enormous damage to human settlements and other valuable resources. The high-spatial-resolution imagery with broad bandwidths is usually employed to prepare the landslide inventory. Various approaches were proposed and attempted in the past to classify the landslide areas directly from the remote sensing imagery in a fully automatic fashion. However, the cultivated lands, roads, houses, riverbeds, and even clouds, are usually misclassified as landslides from these high-spatial- but low-spectral-resolution imagery.
Export Landslide and Shaded Area Delineation System (ELSADS) serves as a fast and reliable approach that enables users to (1) employs two spectral indexes NDVI and Gn to determine the thresholds of non-vegetation and dark areas on-the-fly, (2) overlays the boundaries of non-vegetation and dark areas on the locally-enhanced image and DTM to get the best viewing direction to inspect each area closely, (3) enables the switch between base image, current image, earlier image, geology map, base map and land use map, to facilitate the interpretation. The user can quickly eliminate those non-landslide areas from the non-vegetation areas, and confirm the rest areas are the landslide areas. Instead of delineating the boundary manually, ELSADS now enables the users to prepare a inventory of landslide and shaded areas in a quick and accurate fashion.
A blob (binary large object) is an area of touching pixels with the same logical state. All pixels in an image that belong to a blob are labeled to one, while the other pixels are labeled to zero. Blob analysis enables us to detect and measure blobs, as well as producing information about any 2D shape in an image. In some cases, the landslide areas would connect to riverbeds, roads, or other man-made land covers. ELSADS provides another function to cut one piece of landslide or dark area based on the curve delineated on the image by users. This cutting curve can be plotted and edited on the image through the graphic interface of ELSADS.
Shaded Area Interpretation
There is no signal in shaded areas which makes it unable to interpret landslides therefore shaded areas of satellite imagery should be deleted. By applying principal component analysis, we find that the intensity of panchromatic represents signal of shaded areas. Base on this method, we establish the principal of shaded area interpretation and increase the accuracy of landslide interpretation.
Dynamically integrated with other geospatial information
Since the occurrence of landslide and shaded areas are closely related to slopes, the most useful information for preparing landslide inventory is DTM. ELSADS enables the users to overlay the standard-false-color image and the boundaries of both the non-vegetated and dark areas onto the corresponding DTM on-the-fly. Furthermore, the users can freely rotate and rescale the DTM in a 3D fashion. This provides the most intuitive way to justify the landslide and shaded areas according to their adjacent topography changes. ELSADS also enables us to switch between the current image and the other geospatial information, such as earlier image, geology map, base map or land use map. Switching among various images and maps provides a convenient and unambiguous way to determine the landslide and shaded areas in a quick and accurate fashion.
Video Clip of Using ELSADS