An incomplete airborne lidar survey of Langjökull, Iceland's second largest ice cap (~900 km2) and the surrounding area was undertaken in August 2007. Elevation data were interpolated between the lidar swaths using the technique of photoclinometry (PC), which relates Sun-parallel slope angles to image brightness. A Landsat Enhanced Thematic Mapper Plus (ETM+) image from March 2002 was used for this purpose. Different bands and band combinations were assessed and Band 4 (760-900 nm) was found to be the most appropriate. Parameters in the slope-brightness equation were derived empirically by comparing the image brightness with lidar elevation data in a 4 km × 4 km region in the centre of the ice cap. This relationship was then used to calculate the slopes, and, by integration between tie points of known lidar elevation, the elevations of the 30 m pixels that were not surveyed by lidar. The root-mean-square (RMS) precision (repeatability) of lidar elevations was 0.18 m and the accuracy was estimated to be 0.25 m. The 68.3% quantile of absolute difference relative to lidar (analogous to root-mean-square error (RMSE)) of all interpolated areas where PC assumptions are met was 5.44 m (4.66 m and 8.73 m for on- and off-ice areas, respectively). Where one or more PC assumptions were not met (e.g. self-shading, sensor saturation), the 68.3% quantile of absolute difference relative to lidar was 27.89 m (18.52 m on the ice cap and 32.91 m off-ice). These accuracies were applicable to 63%, 31%, and 6% of the ice cap and 59%, 28%, and 13% of the final digital elevation model (DEM), respectively. The area-weighted average 68.3% quantiles were 2.89 m for the ice cap and 6.75 m for the entire DEM. The PC technique applied to satellite imagery is a useful and appropriate method for interpolating a lidar survey of an ice cap.
|Number of pages||21|
|Journal||International Journal of Remote Sensing|
|Publication status||Published - Feb 2013|
Bibliographical noteFunding Information:
This work was supported by the UK NERC ARSF – Project IPY07-01 and a studentship from Trinity College. Fieldwork costs were supported by the University of Cambridge, the Scandinavian Studies Fund, the B.B. Roberts Fund, Trinity College, St John’s College, and St Catharine’s College. We thank Cameron Rye and Narelle Baker for help in the field, Bill Mockeridge for initial processing of the GPS data, and Gabriel Amable for processing the raw lidar data. Special thanks are due to Helgi Björnsson and Sverrir Guðmundsson from the Institute of Earth Sciences, University of Iceland, for practical and financial help with fieldwork logistics and for provision of data. We also thank two anonymous reviewers for their suggestions which greatly improved the quality of this paper.