Accurate coastal DEM generation by merging ASTER GDEM and ICESat/GLAS data over Mertz Glacier, Antarctica

Mertz Glacier (MG) calved in February 2010 and a 70-years' calving cycle of MG was reported recently because of the shallow Mertz Bank. To better investigate the calving process, a high accurate surface Digital Elevation Model (DEM) with accurate date marker over MG is urgently required for numerical modeling, especially over the ice tongue. However it is quite challenging to generate an accurate DEM over high relief and steep slope regions. To solve this problem, a new method for accurate coastal DEM production by merging ASTER GDEM and ICESat/GLAS data is designed, which can effectively discriminate accurate elevation data from ASTER GDEM. Then, a DEM corresponding to the end of 2002 over MG is generated, which has an accuracy of 0.99 ± 17.50 m. This DEM is several times better in accuracy than the original ASTER GDEM (accuracy: − 8.17 ± 54.31 m) and two new characteristics of ASTER GDEM have been found through analysis. First, ASTER GDEM has elevation bias (varying from − 23 m to 28 m), spatially correlated to ASTER ground tracks, which was most probably caused by uncertainty of attitude measurements of ASTER. Second, ASTER GDEM grids with stacking number ≥ 4 can be effectively adjusted by ground control points so as to improve elevation accuracy. Additionally, this DEM has the best accuracy by comparing with other DEMs (RAMP DEM: 47.71 ± 91.61 m, ICESat/GLAS DEM: − 1.01 ± 30.33, Bamber DEM: − 1.07 ± 33.04 m, Bedmap-2 DEM: 9.19 ± 48.34 m, and Cryosat-2 DEM: − 5.42 ± 32.02 m). The high performance of our DEM in accuracy indicates that our method is effective and has a potential to be widely used to improve existing ASTER GDEM along Antarctic coast.

Details

Publication status:
Published
Author(s):
Authors: Wang, Xianwei, Holland, David M., Gudmundsson, G. Hilmar

On this site: Hilmar Gudmundsson
Date:
1 March, 2018
Journal/Source:
Remote Sensing of Environment / 206
Page(s):
218-230
Digital Object Identifier (DOI):
https://doi.org/10.1016/j.rse.2017.12.041