C41D-0723
Characterising glacier-wide ice-surface roughness using a combined Structure-from-Motion and Terrestrial Laser Scanning approach

Thursday, 17 December 2015
Poster Hall (Moscone South)
Duncan J Quincey1, Mark W Smith1, Jonathan L Carrivick1, David M Rippin2 and Robert G. Bingham3, (1)University of Leeds, School of Geography, Leeds, United Kingdom, (2)University of York, Environment Department, York, YO10, United Kingdom, (3)University of Edinburgh, Edinburgh, EH9, United Kingdom
Abstract:
Ice-surface roughness is a poorly parameterised component of most surface energy balance models, often being represented by a single constant value, or effective roughness length. Yet it plays an important role in determining sensible heat flux, which may contribute ~30% to the overall surface energy balance at high-latitudes. Measurements of ice-surface roughness have thus far been dominated by microtopographic and one-dimensional profile-based approaches, which generate rather crude assessments of roughness length. The availability of fine-resolution and multi-temporal topographic datasets offers an opportunity to develop more robust methods. Using a combined Structure-from-Motion (SfM) and Terrestrial Laser Scanning (TLS) approach, we demonstrate that glacier-wide surface roughness can be characterised over multiple time periods, meaning sensible heat transfer can be much better parameterised in energy balance modelling.

Here we present fine-scale topographic data from 30 plots, each measuring 2 m x 2 m, acquired on Kårsaglaciären, Northern Sweden over a period of four days. Glacier surface types included smooth/superimposed ice, runnels, cryoconite, sun cusps, blocky crystalline ice, supraglacial channels, dirty ice, light/medium/dense scree-covered ice, shallow/deep crevasses and snow. We test multiple methods for extracting z0 and demonstrate a novel 3D approach that uses the dense point cloud rather than a decimated (2D) product. Z0 is shown to vary widely over different surface types, with values ranging over more than an order of magnitude. The evolution of surface roughness is also shown to vary with surface type, with some plots becoming rougher and others becoming smoother through time. Finally, we use a glacier-wide DEM generated from TLS to correlate the detrended standard deviation of elevations in each plot with the calculated z0 values and thus produce a glacier-wide characterisation of ice-surface roughness.