Automated detection of ice cliffs within supraglacial debris cover

Sam Herreid,Francesca Pellicciotti

doi:10.5194/tc-12-1811-2018

Abstract

Abstract. Ice cliffs within a supraglacial debris cover have been identified as a source for high ablation relative to the surrounding debris-covered area. Due to their small relative size and steep orientation, ice cliffs are difficult to detect using nadir-looking space borne sensors. The method presented here uses surface slopes calculated from digital elevation model (DEM) data to map ice cliff geometry and produce an ice cliff probability map. Surface slope thresholds, which can be sensitive to geographic location and/or data quality, are selected automatically. The method also attempts to include area at the (often narrowing) ends of ice cliffs which could otherwise be neglected due to signal saturation in surface slope data. The method was calibrated in the eastern Alaska Range, Alaska, USA, against a control ice cliff dataset derived from high-resolution visible and thermal data. Using the same input parameter set that performed best in Alaska, the method was tested against ice cliffs manually mapped in the Khumbu Himal, Nepal. Our results suggest the method can accommodate different glaciological settings and different DEM data sources without a data intensive (high-resolution, multi-data source) recalibration.

Highlights

Ice cliffs are steep, bare-ice surface features that can develop within a debris-covered portion of a glacier
Melt and surface energy fluxes at specific ice cliffs have been studied in detail (Sakai et al, 1998; Han et al, 2010; Sakai et al, 2002; Reid and Brock, 2014; Buri et al, 2016a) and digital elevation model (DEM) differencing has shown the spatial trends of enhanced glacier melt relative to surrounding debris cover and ice cliff evolution at the scale of several cliffs or a single glacier tongue (Thompson et al, 2016; Brun et al, 2016)
The DEMs used in this study had a spatial resolution of 5 m; for future applications of this method with different input DEM sources, we have derived a relation between coarsening DEM resolution and method performance which can help guide anticipated outcomes

Summary

Introduction

Bare-ice surface features that can develop within a debris-covered portion of a glacier. The direct atmosphere–ice interface can result in significantly higher ablation rates relative to the surrounding debris-covered area and are areas of interest when solving for glacier melt in heavily debris-covered regions (Buri et al, 2016b; Thompson et al, 2016; Brun et al, 2016). Melt and surface energy fluxes at specific ice cliffs have been studied in detail (Sakai et al, 1998; Han et al, 2010; Sakai et al, 2002; Reid and Brock, 2014; Buri et al, 2016a) and digital elevation model (DEM) differencing has shown the spatial trends of enhanced glacier melt relative to surrounding debris cover and ice cliff evolution at the scale of several cliffs or a single glacier tongue (Thompson et al, 2016; Brun et al, 2016). A wide range of ice cliff abundance within a debris-covered area is possible, from no ice cliffs to an abundance capable of possibly negating, or even reversing, the net melt reducing effect of the surrounding debris cover (Kääb et al, 2012; Basnett et al, 2013; Gardelle et al, 2013)

Objectives

Methods

Results

Conclusion