第九届青年地学论坛

The adoption of Real-Time/Post-Processing Kinematic (RTK/PPK) technologies has facilitated the increasing application of Unmanned Aerial Vehicle (UAV) photogrammetry in glaciological research. To ensure accuracy and minimize the costs in glacial environments, achieving PPK-based optimal quality control with a minimal number of Ground Control Points (GCPs) is crucial. In this study, 8571 images were captured at Sermeq Avannarleq glacier in Greenland from August 4^th, 2021, to 6^th, covering approximately 85 km² with the furthest distance 13.22 km away from the coastline. PPK-geotagged images were processed in Agisoft Metashape Professional utilizing 6 configurations of GCPs and Check Points (CPs) distributions, where GCPs distribution gradually extending longitudinally, to assess the accuracy of the Block Bundle Adjustment (BBA) model. Cross-validation with ICESat-2 and ArcticDEM further confirmed the accuracy of the UAV photogrammetric-DSM in each configuration. The results indicated that RMSE in the X, Y, Z directions of the BBA_PPK model was 0.038m, 0.031m, and 0.146m, respectively. Only applying 4 GCPs along the coastline with certain longitudinal distribution improved the RMSE in the X, Y, Z directions to 0.037m, 0.031m, and 0.081m, and these values stabilize even with increasing GCPs. Cross-validation with ICESat-2 and ArcticDEM also demonstrated that the vertical accuracy of UAV-DSM shows marginal enhancement when GCPs number exceeds 4, but all showing a significant vertical improvement compared to DSM_PPK. No elevation ramp error appears in the inland DSM if GCPs are only distributed at the terminus. Therefore, combining PPK with few GCPs which distributes in both direction of the measured region at the coastline offers a viable solution for studying glaciers at Greenland with cm-level accuracy.
 

UAV; PPK; accuracy assessment; ground control points; GrIS