PhD
Dynamical change at tidewater glaciers examined using time-lapse photogrammetry
In a nutshell
Goal: To understand processes linked to dynamical change at tidewater glaciers.
Three main aims:
- Examine subglacial hydrology and its influence on glacier dynamics at Kronebreen, a fast-flowing, tidewater glacier in Svalbard 2. Investigate controls on terminus conditions and calving processes at Tunabreen, a surge-type, tidewater glacier in Svalbard
- Develop a suite of photogrammetry tools for obtaining measurements from oblique time-lapse imagery
Techniques used: Monoscopic time-lapse photogrammetry, hot water borehole drilling, bathymetry surveying, satellite feature-tracking, passive seismic monitoring, melt/runoff modelling
This thesis is freely available to download from the Edinburgh Research Archive
Associated with the Calving Rates And Impact On Sea Level (CRIOS) research group at UNIS
Tidewater glaciers in Svalbard
Svalbard is an archipelago located in the Arctic Ocean (76°N, 16°E), approximately halfway between mainland Norway and the North Pole. Of its total area, 60% of Svalbard is covered by glaciers, making it one of the largest glaciated areas in the Arctic. These glaciers consist of large continuous ice masses that are divided by mountain ridges and nunataks.
[caption id=”attachment_7528” align=”aligncenter” width=”669”]
The two tidewater glaciers in Svalbard that I focused on for my PhD, Kronebreen (B) and Tunabreen (C).[/caption]
Svalbard glaciers have largely been retreating from their last maximum extent during the Little Ice Age. This retreat has been reasonably steady, so they have currently contributed little to current sea level rise. However, they are very sensitive to climate change because of the influence of the North Atlantic Ocean current system bringing warmer water to the Arctic. Predictions indicate that Arctic glaciers will have a larger contribution to sea level rise in the near-future (Blaszcyk et al., 2009), especially those that terminate in fjord water (i.e. tidewater glaciers) due to the direct influence of the ocean (Luckman et al., 2015).
We decided to look at two dynamically different tidewater glaciers in Svalbard in order to monitor their changes over consecutive melt seasons - Kronebreen and Tunabreen.
One of the most persistent fast-flowing glaciers in Svalbard is Kronebreen, which is a tidewater glacier that lies on the west coast (78.8°N, 12.7°E). The glacier has been known to flow up to 4 metres per day, equating to an average annual velocity between 300-800 metres per year (Eiken and Sund, 2012).
Kronebreen is currently in a period of rapid retreat, having receded approximately 1 km between 2011 and 2016. Strong links have been observed between frontal ablation and fjord water temperatures in Kongsfjorden, which suggests that submarine melting is a key factor in ice loss. However, little is understood regarding the short-term changes in terminus conditions and their influence on ice loss at Kronebreen.
[caption id=”attachment_4946” align=”aligncenter” width=”4912”]
Kronebreen (centre) viewed from the west. Kronebreen shares its southern (right) margin with Kongsvegen, a slow-moving surge-type glacier that has been fairly inactive for the past couple of years. The glacier adjacent to Kronebreen, separated by the mountain Collethøgda (left), is called Kongsbreen. Kongsbreen has been retreating from the fjord onto land since approximately 2014 (September 2016)[/caption]
Tunabreen is a surge-type tidewater glacier situated south of Kronebreen, at the head of a large fjord system called Tempelfjorden (78.29°N, 17.25°E). Surging glaciers are identified to undergo multi-year periodic oscillations, switching between an extended phase of normal motion (i.e. the 'quiescent phase') and a relatively short-lived phase of comparatively fast motion (i.e. the 'surge phase').
Tunabreen is one of few surge-type glaciers on Svalbard that has been observed to undergo multiple repeat surface cycles - 1870, 1930, 1971, 2002-05, and most recently in 2016-18 (Flink et al., 2015). At the time of our study (2015-16), Tunabreen was in a quiescent phase and ice motion was largely confined to longitudinal stretching near to the terminus, with little detectable motion in the upper section of the glacier tongue. It has been suggested that oceanic processes are a prevailing control on ice loss at Tunabreen, however, less is understood about controls on short-term changes in terminus stability.
[caption id=”attachment_699” align=”aligncenter” width=”4899”]
Tunabreen terminating into Tempelfjorden. Von Postbreen is a slow-flowing glacier to the east of Tunabreen (top of picture), which is now land-terminating (August 2015)[/caption]
Acquiring data from the field
[caption id=”attachment_1596” align=”alignright” width=”312”]
Camera sites 8a and 8b at Kronebreen, Svalbard (May 2015)[/caption]
High-detail monitoring of glacier termini is challenging. We decided to employ time-lapse photogrammetry as our primary technique in this study given that it can provide high-resolution data acquisition (e.g. 1 image every 3 seconds, over 24 hours) as well as appropriate acquisition rates for longer-term monitoring where needed (e.g. 1 image every hour over the course of a melt season). Therefore we can acquire different temporal frequencies depending on which aspects of the glacier system we want to examine.
We deployed a network of time-lapse cameras at Kronebreen (and in the surrounding area) between 2014 and 2017 to monitor various aspects - ice flow, terminus retreat, supraglacial lake drainage, meltwater plumes, and local fjord circulation. In addition, a borehole was drilled on the glacier tongue in 2013 as part of the CRIOS research project, which provided additional insight into subglacial dynamics.
Four time-lapse cameras were installed at the margin of Tunabreen, monitoring terminus dynamics between 2015 and 2016, including calving activity, ice flow, terminus retreat and meltwater plumes. We also produced a highly detailed sequence from one of our cameras entrained on the calving front, capturing images every three seconds. From this, we could identify individual calving events and calving styles. Additional researchers accompanied us on this fieldwork to conduct bathymetry surveying of the sea bed and side-profiling of the submarine terminus, and to conduct laser scanning of the subaerial part of the terminus.
[caption id=”attachment_698” align=”alignnone” width=”4912”]
The field team carrying time-lapse cameras up Brucebreen to Ultunafjella, which overlooks Tunabreen (August 2015)[/caption]
Main findings
#1 Spatial variation in Kronebreen's ice flow is primarily controlled by meltwater routing at the glacier bed
[caption id=”attachment_7445” align=”alignleft” width=”343”]
Map of Kronebreen, showing the borehole installation (starred) and the time-lapse camera set-up (orange points), along with the supraglacial lake groups which drained over the course of the 2014 melt season (C1, C2 and C3)[/caption]
From our time-lapse images over the 2014 melt season, along with borehole data analysis, melt/runoff modelling and hydropotential modelling, we found that spatial variations in ice flow at Kronebreen were primarily controlled by the location of subglacial meltwater channels.
Efficiency in subglacial water evacuation varied between the north and south regions of the glacier tongue, with the north channel configuration draining a large proportion of the glacier catchment through persistent channels, as indicated by hydropotential modelling. Channel configurations beneath the south region of the terminus were vastly different, with rapid hydrological changes evident and cyclic 'pulsing' suggested from the observed meltwater plume activity.
These differences in subglacial hydrology are reflected in ice flow, with faster velocities experienced in the south region of the glacier, facilitated by enhanced basal lubrication and sliding. Two speed-up events were observed at the beginning of the 2014 melt season, the second being of significant importance given that it occurred at the end of the melt season and enabled fast flow through the winter season. It is suggested that this event was caused by an abnormal high rainfall event which overwhelmed an inefficient hydrological regime entering its winter phase. This phenomena highlights that the timing of rainfall events at tidewater glaciers is fundamental to their impact on ice flow.
[caption id=”attachment_7405” align=”alignnone” width=”1776”]
Sequential velocity maps (left) and velocity change maps (right) of Kronebreen showing the first of two speed-up events experienced during the 2014 melt season.[/caption]
#2 Terminus stability is inherently linked to both atmospheric and oceanic variability at Tunabreen. In particular, calving activity is primarily facilitated by melt-undercutting
Terminus conditions at Tunabreen were examined on two differing temporal scales:
- Over a one month period in peak melt season using time-lapse images acquired every 10 minutes
- Over a 28-hour period in August 2015 using time-lapse images acquired every 3 seconds
Over the one-month observation period, the terminus retreated 73.3 metres, with an average retreat rate of 1.83 metres per day. The frontal ablation rate fluctuated between 0 and 8.85 metres per day, and 1820 calving events were recorded of which 115 events were simultaneously detected from passive seismic signatures recorded in Longyearbyen. Overall, strong links were found between terminus position changes and both sea surface temperature and air temperature, suggesting that atmospheric forcing plays a larger role in terminus stability than previously considered.
[caption id=”attachment_1027” align=”alignleft” width=”389”]
Calving events at Tunabreen over a 28-hour period in August 2015, captured using high-resolution time-lapse photography (one photo every three seconds). Calving events are categorised as subaerial (i.e. ice falling from the front above the waterline), subaqueous (i.e. ice breaking off from the front beneath the waterline), both (i.e. large calving events which contain both subaerial and subaqueous originating ice) and unknown (caused by concealment or poor visibility)[/caption]
Calving activity at Tunabreen consists of frequent events, with 358 calving events detected from the 28-hour, high-frequency time-lapse sequence (i.e. 12.8 events per hour). The majority of these calving events (97%) occurred above the waterline despite the fact that 60-70% of the terminus is subaqueous (i.e. below the waterline). This suggests that ice loss below the waterline is dominated by submarine melting, rather than the break off of large projecting 'ice feet'. In addition, calving events are twice as frequent in the vicinity of the meltwater plumes, with visible undercutting (approximately 5 metres) revealed from the bathymetry side profiles. Overall, this suggests that enhanced submarine melting causes localised terminuinstability at Tunabreen.
#3 PyTrx is a viable Python-alternative toolbox for extracting measurements from oblique imagery of glacier environments
Time-lapse photogrammetry is a growing method in glaciology for providing measurements from oblique sequential imagery, namely glacier velocity. When we began processing our time-lapse images, we found that there were few publicly available toolboxes for what we wanted and the range of their applications was relatively small. For this reason we decided to develop PyTrx, a Python-alternative toolbox, to process our own data and also aid the progression of glacial photogrammetry with a wider range of toolboxes.
[caption id=”attachment_7545” align=”alignright” width=”353”]
An example of PyTrx’s capabilities in deriving surface velocites at Kronebreen, Svalbard. Velocities are calculated from the image using a sparse feature-tracking approach, with unique corner features identified using Shi-Tomasi corner detection and subsequently tracked using Optical Flow approximation. In this example, 50 000 points have been successfully tracked between an image pair from Kronebreen, producing a dense collection of velocity points.[/caption]
PyTrx is an object-oriented toolbox, consisting of six scripts that can be used to obtain velocity, area and line measurements from a series of oblique images. These six scripts are:
- CamEnv: Handles the associated data with the camera environment, namely the Ground Control Points (GCPs), information about the camera distortion, and the camera location and pose
- DEM: Handles data related to the scene, or Digital Elevation Model (DEM)
- FileHandler: Contains functions for reading in data from files (such as image data and calibration information) and exporting output data
- Images: Handles the image sequence and the data associate with each individual image
- Measure: Handles the functionality for calculating homography, velocities, surface areas and distances from oblique imagery
- Utilities: Contains the functions for plotting and interpolating data
PyTrx has been used to process the data presented previously, and is freely available on GitHub with several example applications also. These examples include deriving surface velocities and meltwater plume footprints from time-lapse images of Kronebreen, and terminus profiles and calving event locations from time-lapse images of Tunabreen.
Associated papers and outputs
How et al. (2017) The Cryosphere - Examining the subglacial hydrology of Kronebreen and its influence on glacier dynamics
How et al. (2019) Annals of Glaciology - Observations of calving styles at Tunabreen and the role of submarine melting in calving dynamics
How et al. (2020) Frontiers in Earth Science - Presenting the PyTrx toolbox and its capabilities with oblique imagery of glacial environments
PyTrx - PyTrx toolbox code repository, hosted on GitHub