Stress mapping reveals secrets of coupled supraglacial lake drainages by hydro-fracture

A new study, led by the University of Oxford and Woods Hole Oceanographic Institution, has mapped stress evolution during supraglacial lake drainages, revealing a physical explanation for what causes neighbouring lakes to drain within hours of each other via hydro-fracture. The findings have been published today in the Journal of Geophysical Research.

Greenland, home to the second largest ice sheet in the world, is particularly vulnerable to present-day climate change. Of particular concern to scientists is the rate at which the ice sheet is shrinking, a process which is partially driven by the flow of ice into the surrounding oceans. Surface melting of the ice sheet plays an important role in this: if meltwater reaches the base of the ice where it meets underlying bedrock, this water lubricates the ice and drastically speeds up its movement towards lower elevations.

Throughout the summer season, the ice surface melts and this water collects in topographic lows on the surface of the ice sheet, forming huge bodies of water referred to as “supraglacial lakes”. There are thousands of these lakes across the surface of the Greenland Ice Sheet – expanses of crystal blue water in an otherwise icy landscape. Around 70–90% of supraglacial lakes either refreeze at the end of the summer, or drain via channels over the ice-sheet surface, kind of like lakes on land. The rarer (and somewhat cinematic) phenomenon is where kilometre-scale fractures form across the lake, and the water in the lake drives these fractures down through the ice sheet, draining the lake rapidly all the way to the bedrock. Known as “hydro-fracture”, this process can empty lakes more than three kilometres across in a few hours, like pulling the plug from a sink.

Since the first observations of hydro-fracturing lakes, scientists have been puzzled by the processes which trigger these rifts in the ice. A new study, led by Oxford Earth Sciences’ Associate Professor Laura Stevens, has made use of Global Positioning System (GPS) data from sensors circling a cluster of supraglacial lakes, in addition to satellite imagery, to examine this process more closely.

Building on earlier work led by co-author Dr. Sarah Das, the team mapped the 3D position of the ice surface at 30-second resolution across two melt seasons and two sets of lake drainages. Further geophysical modelling made it possible to examine how ice-sheet stress in the local area changes with time when the lakes drain. Tensile or extensional stress is the forces of tension acting on the ice, and much like ripping open a packet of crisps, tension is a key ingredient in the formation of fractures. For this study, Stevens and their international team of researchers were able to map how stress changed through time, and consider whether the stress changes during fracturing and drainage at one lake could also trigger a similar process in neighbouring lakes.

The lakes in question are located in central west Greenland, to the South East of Ilulissat, the third largest settlement in the country. The lakes are just south of Sermeq Kujalleq, one of the fastest glaciers in the world, and a key ice stream linking the ice sheet with the sea. During fieldwork campaigns between 2011 and 2012, the research team installed 16 GPS instruments in the local area—a fieldwork process involving helicopter transport and specialist GPS equipment—which allowed the scientists to securely attach the instruments to the ice surface in preparation for the approximately 2 metres of melting which occurs here every summer.

Study region in western Greenland (panel a) with lakes investigated boxed in red. (panels c–f) Series of European Space Agency Sentinel-2 images shows the central (L1A) and southern (L1B) lakes draining within the same 24-hour window in late-June 2020 (panels c,d). The lake to the northwest (L1D) drains over the surface into a moulin (panel e). The northern lake (L1C) freezes over in mid-September 2020, at the end of the melt season (panel f).

Initially the goal of this field campaign was to examine the mechanisms driving hydro-fracture for a single, central lake, which Stevens took on during their PhD. It was during the Covid-19 pandemic that Stevens was inspired to re-visit the dataset after looking at images (from the Sentinel-2 satellite) and noticing two lakes in the area draining within the same 24-hour time window during summer 2020 (Satellite Imagery Figure). This observation resulted in a collaboration which allowed for Stevens’ PhD and postdoc mentors and collaborators to join forces.

Piecing together a history of these lakes’ drainage events going back to 2000, the team realized that two lakes in the area drain within the time window of repeat satellite images (2–3 days on average) in exactly half of years when both lakes formed. Stevens said, “I was curious as to whether these coincident lake drainages were just a trick of a temporally aliased satellite-imagery record; or, alternatively, if there was a physical explanation for these drainage events.”

So, Stevens and colleagues went back to the 2011 and 2012 GPS observations, and extended their numerical methods from earlier work to be able to estimate the ice-sheet surface stress state for three lakes within the area as the drainage of the central lake took place. The research team found that by far the biggest influence on the ice stress state during drainages was the formation and migration of kilometre-scale blisters of water beneath the ice sheet. Ice deformation sourced from these blisters opening at the bed can explain more than 95% of stress variations at the ice-sheet surface during the hydro-fracture events. The blisters force the overlaying ice upwards, which causes the ice surface to crack, much like bending a caramel-filled chocolate bar that has been stuck in the freezer.

Not only this, but it was observed that large tensional stresses in one lake can affect neighbouring lakes. Within the study area, two adjacent lakes drained within hours of each other due to this effect, but a third lake was unaffected and did not drain. The authors hypothesize that this was because the ice beneath this stable lake was slightly compressed, rather than stretched, by the ice deformation resulting from the drainage of the other two lakes. “That we observe this lake to the north being placed weakly in compression during the drainage of the other two is a really neat observation,” Stevens commented. “It reveals that one lake drainage doesn’t always promote a neighbouring lake to drain. There’s something more complicated at play in the dynamics between these lakes.”

These findings have prompted even greater questions regarding the contribution of supraglacial lake drainage to the depletion of the Greenland Ice Sheet. A significant – but currently unanswered – question is the maximum elevation at which we should expect hydro-fracturing to occur today, and in future decades. Due to rising air temperatures above the ice sheet, we are observing an increasing number of lakes across Greenland forming at higher altitudes. Scientists want to know if these high-altitude lakes are also a potential source of water delivery to the bed beneath the ice sheet; or rather, if there is a spatial restriction or upper limit on where hydro-fracture can occur. This study offers a first step towards a better understanding of how this process impacts the rate of ice sheet movement, by observing and explaining the physical mechanisms by which these lakes can promote or inhibit neighbouring lakes to drain.

The study “Elastic stress coupling between supraglacial lakes” is available to read in full in the Journal of Geophysical Research: Earth Surface at https://doi.org/10.1029/2023JF007481.