Sensitivity of glacier volume change estimation to DEM void interpolation

Abstract

Glacier mass balance has been estimated on individual glacier and regional scales using repeat digital elevation models (DEMs). DEMs often have gaps in coverage ("voids"), the properties of which depend on the nature of the sensor used and the surface being measured. The way that these voids are accounted for has a direct impact on the estimate of geodetic glacier mass balance, though a systematic comparison of different proposed methods has been heretofore lacking. In this study, we determine the impact and sensitivity of void interpolation methods on estimates of volume change. Using two spatially complete, high-resolution DEMs over southeast Alaska, USA, we artificially generate voids in one of the DEMs using correlation values derived from photogrammetric processing of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) scenes. We then compare 11 different void interpolation methods on a glacier-by-glacier and regional basis. We find that a few methods introduce biases of up to 20 % in the regional results, while other methods give results very close ( < 1 % difference) to the true, non-voided volume change estimates. By comparing results from a few of the best-performing methods, an estimate of the uncertainty introduced by interpolating voids can be obtained. Finally, by increasing the number of voids, we show that with these best-performing methods, reliable estimates of glacier-wide volume change can be obtained, even with sparse DEM coverage.Published by Copernicus Publications on behalf of the European Geosciences Union.

Elevation Changes of West-Central Greenland Glaciers From 1985 to 2012 From Remote Sensing

“…The major challenge arises from the ignorance about the true elevation changes across the data gaps. Therefore, McNabb et al (2019) artificially produced data voids in DEMs over southeast Alaska to identify which method reproduces the reality most accurate. However, the bestperforming method in that study (linear interpolation of voids in the dDEM) cannot just be transferred to any other region with data voids in the DEMs.…”

“…Three conditions majorly influence the identification of the appropriate void-filling method: (a) the aim of the study: calculation of region-wide vs. glacierwide elevation changes, (b) the characteristics of the glaciers in the study area (e.g., surging glaciers), and (c) the distribution of the data voids which is not expected to be the same for each region, or even each DEM. Given the circumstances for this study of glaciers in west-central Greenland [(a) aiming for glacier-wide elevation changes, (b) the presence of surging glaciers, and (c) voids mainly in accumulation area], we chose five methods recommended by McNabb et al (2019) to identify the influence on the estimates for our study region. Additionally, this multi-method approach allowed an uncertainty assessment of the void-filling method (see Methods).…”

“…The interpolation of voids in DEMs and dDEMs is a common procedure; however, the precise interpolation method is not standard. McNabb et al (2019) assessed the impact and sensitivity of different void-filling methods on estimates of glacier volume change. Following the recommendations by McNabb et al (2017), we chose to fill voids in the dDEM rather than in the original DEMs and, hence, selected five approaches to fill the voids in our dDEM.…”

“…For methods 2-3, the inclusion of office pixels with (approximately) zero values would most likely depress the mean/median and so we excluded these data from our interpolation. Methods 2-5 are so-called "hypsometric methods" building upon the assumption "that there is a relationship between elevation change and elevation" (e.g., McNabb et al, 2019). For this, we divided the dDEM in 100 m elevation bins (based on the TanDEM-X) and for each bin, we calculated the mean/median elevation change.…”

“…For this study, we used the local mean (per elevation bin) method as our best guess to fill the voids in the dDEM of our study region. This choice is recommended by our results, by the findings in McNabb et al (2019), by the distribution of voids in the data, and by the characteristics of the glaciers in this region as it better accounts for the characteristics of individual glaciers when compared to the other methods (cf. discussion).…”