Assessment of Penetration Depth Bias in TanDEM-X Surface Elevation on Devon Island Ice Cap, Canadian Arctic

Abstract

Remotely sensed surface elevation datasets are key to study mass balance and dynamic behaviours of glaciers, ice caps, and ice sheets. Digital elevation models (DEMs) generated from the TanDEM-X mission from single-pass synthetic aperture radar interferometry (InSAR) at X-band provide an excellent potential resource for this purpose. However, they typically suffer from a negative bias in calculated surface elevations on glaciated terrain due to penetration of radar waves in the snowpack and firn layers. Penetration depth depends on snowpack properties and acquisition parameters and can be estimated from backscatter intensity, interferometric coherence and height of ambiguity of input scenes. To validate modelled penetration depth, we take advantage of the overlap of multiple TanDEM-X scenes with surface elevation data derived from ground, airborne and spaceborne campaigns over Devon Island Ice Cap (75.25° N, 82.5° W) in the Canadian Arctic. Comparison of raw TanDEM-X DEMs from Fall and Winter 2010 to 2014 and 2018 with kinematic surface differential GPS transects, ATM/IceBridge NASA laser altimetry surveys and ArcticDEM Strip DEMs, reveal significant elevation bias of approximatively -3 m wherever firn layers are present below the snowpack. We present the interannual and spatial variability of penetration depths in relation to the near-surface conditions and radar cquisition parameters over Devon Ice Cap. This validation case study aims at improving models of penetration depth for future applications of TanDEM-X DEMs over glaciated terrain.