Optimizing photogrammetric DEMs for glacier volume change assessment using laser-scanning derived ground-control points

Abstract

Photogrammetric processing of archival stereo imagery offers the opportunity to reconstruct glacier volume changes for regions where no such data exist, and to better constrain the contribution to sea-level rise from small glaciers and ice caps. The ability to derive digital elevation model (DEM) measurements of glacier volume from photogrammetry relies on good-quality, well-distributed ground reference data, which may be difficult to acquire. This study shows that ground-control points (GCPs) can be identified and extracted from point-cloud airborne lidar data and used to control photogrammetric glacier models. The technique is applied to midtre Lovénbreen, a small valley glacier in northwest Svalbard. We show that the amount of ground control measured and the elevation accuracy of GCP coordinates (based on known and theoretical error considerations) has a significant effect on photogrammetric model statistics, DEM accuracy and the subsequent geodetic measurement of glacier volume change. Models controlled with fewer than 20 lidar control points or GCPs from sub-optimal areas within the swath footprint overestimated volume change by 14-53% over a 2 year period. DEMs derived from models utilizing 20-25 or more GCPs, however, gave volume change estimates within ∼4% of those from repeat lidar data (-0.51 ma-1 between 2003 and 2005). Our results have important implications for the measurement of glacier volume change from archival stereo-imagery sources.