The full citation is here: Trusel, L. D., Frey, K. E., Das, S. B., Karnauskas, K. B., Munneke, P. K., Van Meijgaard, E., & Van Den Broeke, M. R. (2015). Divergent trajectories of Antarctic surface melt under two twenty-first-century climate scenarios. Nature Geoscience.
The paper suggests that most Antarctic ice sheets are at risk of collapse by the end of the century, even assuming relatively aggressive emissions reductions consistent with RCP4.5 (the IPCC scenario which results in approximately 2°C of warming, the threshold that is considered ‘safe’, although above the 1.5°C target that has been agreed in the Paris Agreement, and still well below the trajectory we are currently on – RCP8.5, or around 4°C of warming). This includes those that are typically considered relatively stable because the accumulation of snow and ice roughly matches the amount of loss due to melt and runoff, such as the Shackleton and other East Antarctic ice sheets.
The total amount of water contained in Antarctic ice sheets would contribute more than 58 m of sea level rise globally. That’s a catastrophic amount of water – not only would it inundate coastal and island areas which are disproportionately located in developing countries, and which are relatively densely populated, but it would also increase the intensity of storm surges and flooding in low-lying areas across the world, causing further damage to homes, livelihoods and communities. Again, the brunt of the impacts will be borne by those who are least able to adapt or mitigate against the effects of climate change because they live in developing countries with fewer resources or political clout on the international stage to go up against major players like the US and EU. If we think the migrant crisis is bad now, we have a lot to learn.
Trusel et al. present a lot of information-dense figures, but Figure 3 is the one that stands out for me. The study uses a method called ensemble modelling, whereby you run numerous different models with varying strengths and weaknesses and representations of the Earth system, and combine the output to get a ‘best guess’ estimate that minimises the bias introduced by any one of the ensemble ‘members’. The ensemble projects a steady increase in the amount of melting across all Antarctic ice shelves as the century goes on, though the amount of melt under a higher emissions scenario is much more considerable. Y’know; RCP8.5, the scenario we are currently on track to achieve.
My own research focuses on the Antarctic Peninsula, which is the most rapidly warming region on the planet, having warmed by around 3°C since the ‘50s. Specifically, I’m looking at Larsen C, which is the largest ice sheet remaining on the peninsula, its neighbours Larsen A and B having succumbed to melting and collapsed into the sea in 1995 and 2003, respectively.
Trusel et al. use the conditions observed at the time of their collapse as a threshold against which to compare the stability of other ice shelves. This may or may not be a reasonable benchmark because we cannot know for sure whether the conditions that led to the collapse of Larsen A and B are going to be the same for other ice shelves across the continent. Factors such as glacial structure, flow velocity, atmospheric circulation, ocean circulation, latitude, structure of warming and underlying geology all have an impact on the stability of ice shelves, and consequently on how quickly they are likely to collapse, if at all. This threshold is assumed to be representative, but considering this caveat, we can continue.
We know from observations that depletion of the amount of air in the top layers of snow and ice (‘firn’) is a precursor of ice sheet collapse. That’s because this top firn layer is very porous and therefore acts as a buffer, absorbing excess meltwater in warm years, and allowing it to refreeze, rather than run off and be lost into the ocean. However, if this layer becomes saturated, this buffer is lost, and ponds of water begin to form on the surface and fill crevasses that extend under pressure, eventually causing sections of the ice sheet to calve off into icebergs.
Now, as we’re talking about ice sheets breaking up, it seems pertinent to mention this: during the last two summers in Antarctica, a huge crack has begun to extend across Larsen C, sparking fears that its stability may be in question. Further research suggests that the ice sheet as a whole is relatively stable, but that the volume of ice that this crack will cause to break off will represent just over 10% of the total surface area of the ice sheet. What this indicates is that Larsen C is reaching the same kinds of thresholds that were observed before Larsen A and B collapsed, so small changes like a few warm summers in succession could have considerable consequences.
The results of Trusel et al. indicate that Larsen C, where meltwater is currently produced at a rate of around 275 mm water equivalent (w.e.) per year, is on track to hit the threshold of 725 mm w.e. per year by the end of the 21st century under a relatively modest emissions scenario, RCP4.5. What is more shocking though, is by how much the threshold is projected to be overshot by if we keep emitting fossil fuels like we currently are. By 2100, under the RCP8.5 scenario, the meltwater production rates soar to around 2400 mm w.e. per year: that’s nearly nine times the current rates on the Larsen C, and more than three times the threshold level of melt considered a precondition of collapse.
Under the RCP8.5 scenario the outcomes are pretty terrifying: virtually all ice sheets will be lost from the Antarctic Peninsula. While the loss of floating ice shelves in itself will not contribute to sea level rise (they already displace their own weight in water – just like Archimedes in the bath, Eureka!), they in effect act as a plug, preventing all the ice that flows down from the mountains from spilling into the sea. Once these floating ice shelves are gone, there’s nothing to stop all that ice flowing rapidly downhill and into the ocean, as was observed after the loss of other ice sheets like Larsen A and B.
The loss of the Larsen C alone would contribute an estimate 27 cm of sea level rise. Add this to the other ice sheets that would potentially collapse under the level of warming that seems likely right now and we look set to lose a lot of land worldwide, not to mention the increased intensity and damage caused by natural events like flooding and storms.
The major limitation of this study is the use of a model called RACMO. This is a hydrostatic model, which means that it assumes that the atmosphere is in hydrostatic equilibrium (the assumption that flow is at constant speed, being balanced by gravity and the pressure gradient force). While this is broadly true in the atmosphere, because forces acting in opposite directions balance out, this is only the case at larger scales. At finer scales, say, below 8 km, this assumption breaks down. This might sound like model semantics, but this is important.
Over Larsen C in particular, the foehn warming mechanism has been shown to be critical. This is a process whereby air is pushed over the barrier presented by the transantarctic mountains along the peninsula and is warmed and dried as it moves. This is because the air is forced upwards, meaning all the moisture in the air mass is forced to condense out, leading to orographic (driven by the shape and height of the mountain) precipitation on the windward (the first side the air encounters) side of the mountain. Other mechanisms are also in action, such as mechanical mixing of warmer air from higher aloft due to this forced uplift, which are well documented by Elvidge & Renfrew (2016) if you are interested.
The result of these mechanisms is to warm and dry the air on the lee side (the other side) of the mountain, generating warm, dry and speedy foehn winds. In fact, these foehn winds can cause temperature rises of up to 30°C, representing quite a temperature contrast and contributing to melt over the ice shelf. You can see how this might be quite an important way that melt can be induced over ice shelves.
Now, foehn winds happen on a small scale, which means that they are poorly resolved by models at a coarser resolution, for instance 5.5 km and up. To accurately simulate them, resolution at the kilometre scale is required. For instance, Andy Elvidge’s work on foehn winds utilised the MetUM at 1.5 km resolution. I hope to run the same model at even finer scale, say 1 km or even 500 m scale, to investigate the processes at work.
Hydrostatic models’ fundamental assumptions place a theoretical constraint on the resolution at which they can represent climate systems. They cannot really operate at a finer resolution than the one used by Trusel et al. (2015). This means that the effects of foehn winds are not included in the projections they make, which have been shown in several papers to be critical in determining the extent, intensity and distribution of melt over Larsen C in particular. Because they use Larsen C to benchmark melt over other ice shelves, this may affect their projections and limit their understanding of the processes at work.
All in all, this is a beautifully written paper. However, the logic is a little too elegant and simple, and perhaps irons out some of the complexities at work. Further work at finer resolution is evidently required to improve upon research in this area.