Deriving digital elevation models using high-resolution satellite imagery in high relief volcanic areas

Over high relief active volcanic areas, customary methods of deriving elevations may be prohibitive. To overcome this, two pairs of stereo Pleiades imagery collected in July 2015 over Mt. Etna volcanic area (Italy) were considered. The Pleiades stereo pairs (1&2 and 2&3) were processed using the NASA Ames Stereo Pipeline (ASP), a suite of open source automated geodesy and stereo-photogrammetry tools intended to process stereo imagery with or without accurate camera pose information. The workflow involves image pre-processing, feature matching and 3D reconstruction. The image processing step prepares the data for feature matching by updating the sensor ephemeris / attitude information for two or more input images. The feature matching procedure prepares the data for 3D point cloud generation. Each pair of stereo satellite imagery was run with two algorithms, one using a local search window (LSW), and the other using a more global matching (MGM) methodology, a modification of the semi-global matching (SGM) algorithm. The first algorithm estimates the apparent motion of the scene points, or disparity, by comparing local windows around each pixel in the stereo pair. This method cannot cope with ambiguities in the image texture such as smoothness or repetitive patterns. The second algorithm compensates for ambiguities by using the smoothness of the disparity map and reducing the two-dimensional smoothness constraint to the average of the one-dimensional line optimization problem. The last step generates a 3D point cloud through triangulation by combining spacecraft ephemeris / attitude information, sensor model and the refined and filtered disparity map of left and right pixel matches.

The planimetric root mean square error (RMSE) of all derived 2-meter spatial resolution DEMs ranges from 1 to 2 pixels. The altimetry validation was computed for both ellipsoid and orthometric elevations. The ellipsoidal elevation values have been converted into orthometric heights using the CGR tool and the geoid ITALGEO 95.

The elevations are positively biased 4.5 to 6.5 meters, with a RMSE between 4.6 to 6.8 meters. The best results for both planimetry and altimetry are obtained for pair 2&3. We expect to decrease the vertical RMSE to 1-meter by adding visually identifiable ground control points.