How does a glacier melt underwater?

This post was first published on the ice2ice blog though I have extended and expanded it with some extra details.

 Ice2Ice is a large project funded by the European Research Council in a Synergy grant. It is one of my current main sources of funding and is a pretty exciting piece of work, looking at abrupt climate changes in the past (when the temperature around Greenland suddenly increased by around 10-15C in about a decade) known as Dansgaard-Oeschger cycles.  Go and check out the webpage and also this very nice overview on science nordic including cool film showing ice core processing

Glaciers in Greenland lose mass by melt and runoff, by calving and by submarine melt that happens at the front of outlet glaciers that terminate in the ocean. Submarine melt occurs because the ocean water is (relatively) warmer than the ice, but it goes much faster where there is turbulent water mixing the layers by the glacier. Probably the most important source of turbulence are plumes of water that emerge at the base of the glacier where it terminates in the fjord. Penny How, a scientist at the University of St Andrews recently wrote a very nice blog post giving an overview. Including this nice GIF showing a very clear plume coming out the front of Tunabreen, a glacier in Svalbard:

Note the dark coloured water (indicating high sediment concentration) coming out the front of Tunabreen in Svalbard – Image by Penny How 

The water is generated by melting mostly at the surface though also at the bed of the glacier. Meltwater flows like rivers through systems of englacial channels to finally arrive at the bed where it makes its way, eventually, to the end of the glacier.

Unfortunately these channels are pretty hard to map, and there are lakes and areas at the bed where water can be stored. The plumes themselves are rather hazardous to observe as they are often inaccessible and in front of actively calving sections of the glacier. There have been a few studies, but often these are snapshots in time and it is difficult to assess how important these processes are to the overall mass budget of the ice sheet.

Therefore we have to turn to models to work out how important plume processes are for submarine melt. In our recent paper with Slater et al (2017), we contributed data from the HIRHAM5 RCM to look at runoff within a catchment in Greenland. The case study was based at Kangiata Nunata Sermia glacier, in the Godthåbsfjord area of south western Greenland. It’s a relatively accessible glacier showing many of the common processes for Greenland outlet glaciers and has a fair bit of data available. The Langen et al (2015) paper showed that HIRHAM5 performs pretty well in terms of modelled runoff in this region as I detailed in this post.

The modelled runoff was used in two different models of subglacial plumes, including one implemented in MITgcm, in order to determine what configuration of subglacial hydrology and plume distribution along the ice front was most likely.  The models were compared with a time lapse photos of the ice front showing plume activity at the surface.

Illustrations of plume state classification. a) Plume state = -1, ice tongue present. b) Plume state = 0, no ice tongue and surface expression of a plume. c) Plume state = 1, plume visible adjacent to glacier terminue but is contained within a few hundred metres of the trerminus. d) Plume state = 2, plume visible and flows down-fjord at surface for a number of kilometres. 

For a large proportion of the summer, the modelled catchment runoff greatly exceeds the discharge required to create a plume that would reach the fjord surface, yet there are extended periods when there is no plume visible from the time lapse pictures. This can only be explained by the runoff emerging into the fjord in a spatially distributed fashion. In the paper we therefore argue that subglacial drainage near the glacier terminus is often spatially distributed, formed either from numerous point sources of subglacial discharge, or a single but very wide subglacial channel or possibly a complex combination of the two.

There are two implications from this work. Firstly, a more spatially distributed submarine plume gives a higher total melt than a single concentrated plume but this melt rate is still unable to explain the mass loss at the terminus when considering the ice velocity at the terminus, suggesting that calving is still the most important mass flux term at this glacier. Secondly, the modelling study found that the distributed hydrology, suggested by the results leads to a more direct ice flow response to high surface melt rates and this response most likely scales with catchment size.

Probably the most important result to come out of this study is that longer time series of observations of plumes, in combination with the modelled runoff lead to a dramatically different understanding of key processes within the fjords when compared to those suggested by simple snapshot observations in earlier studies.

a) Air temperature from KNS1 and NUKL PROMICE stations b) Modelled runoff. HIRHAM5 (orange) delays runoff using a parameterisation based on surface slope. PDD model (green) assumes instantaneous runoff. PDD delay (pink) uses a transit velocity of 0.05 m s-1 from point of production to the terminus. PDD rapid (purple) uses a transit velocity of 1 m s-1. the green curve has been smoothed using a 3d moving window, the pink and pruple curves using a 6 h moving window. large discrepancies between HIRHAM5 and the PDD model arise due to rainfall events (e.g. days 177 and 181). c) KNS1 daily ice velocity. d) Plume state as described in the picture above. 

So, does this matter – well, probably. These kind of studies sprang up in the wake of a paper published by Eric Rignot and friends. They were almost the first to really look seriously and consistently at the amount of ocean-driven melting going on at these fronts in Greenland and they found summer melt rates at a number of glaciers that would indicate almost as much mass loss as from calving rates. Out of this early work, the NASA funded OMG (Oceans Melting Greenland) project emerged, which is currently contributing a huge amount of data that will be very useful – including the bathymetry of many fjords in Greenland which are still in general rather poorly mapped.

This focus is welcome, the ice sheet-ocean interaction is incredibly poorly observed and we are very much reliant on imperfect models. However,  our study shows that meltwater driven plumes and melt rates at the front of this glacier is rather a small source of glacier retreat when compared to calving rates or surface melt and runoff. Calving is important because a large amount of ice can be lost very quickly as these bergs show. These calved off the front of the same glacier as in this study but in 2009 when I happened to be in Greenland. the glacier, Kangiata Nunata Sermia was previously called the Godthåbs glacier, after the earlier Danish name for the town of Nuuk. It is also often known by the abbreviation KNS in the scientific community.



If this process is similar elsewhere in Greenland it suggests that submarine melt may be less important to rates of ice mass loss and consequent sea level rise. This is not to say it is not important elsewhere in Greenland, and indeed under climate change scenarios with enhanced melt and much warmer ocean waters making it way into the fjords, this may well change and could potentially become much more important.

The mass of icebergs that are calved from the front of Kangiata Nunata Sermia