The two pools of the Churchill Marine Observatory, where we grew sea ice and flooded it. Photo provided by U. Manitoba to Global News. Click to enlarge.
I recently received a fellowship award from the Scientific Committee on Antarctic Research (SCAR). Unlike its sister fellowship from IASC (which I’ve also written about, and more on that soon), the SCAR fellowship awards money for a scientific project that you propose, but the activity must happen in a different country to where the applicant works.
A succesful SCAR fellowship proposal must address an Antarctic scientific question, and gets bonus points if the host organisation isn’t a traditional powerhouse of Antarctic research. I proposed to study the oxygen-isotope signature of sea ice flooding at the Churchill Marine Observatory in Canada.
The Snow-Ice Challenge
The Antarctic sea ice pack sees a lot more snowfall than its northern cousin: probably between twice and four-times as much. That happens while Antarctic sea ice itself sits at a typically lower latitude than of the Arctic, so grows thermodynamically more slowly. Long story short: the ice is weighed down more by accumulated snow, and is less capable of holding it up anyway. This often leads to the ice surface being pushed below the waterline, and the snow on top flooding with seawater. After a cold-snap, that flooded layer freezes and forms snow-ice.
Some flooded snow on Antarctic sea ice that I encountered with Vishnu Nandan in 2023. The sea ice surface was pushed 1 cm below the surface, and capillary action had soaked the snowpack 8 cm above the waterline. Photo from Mallett et al., 2024.
It’s worth trying to partition sea-ice formation between the “traditional” mechanism of downward growth at the floe’s base, and snow-ice formation at the floe’s top. That’s because sea ice models are one of the worst performing components of earth system models, and are particularly bad in Antarctica. So we’d really like to improve them, and one big step would be knowing how much snow-ice formation we’re supposed to be simulating in the first place. Step one to doing that is being able to identify snow-ice in the field!
Snow-ice identification in the field is tricky, because after a while snow-ice can take on a very similar salinity, density and grain structure to “normal” (experts read: frazil) sea ice. So when we show up to an Antarctic sea ice floe in the wild, how can we figure out what’s normal sea ice and what’s snow-ice? Isotopes. The oxygen atoms in water typically have two different weights (isotopes). So when water molecules evaporate from the ocean, the molecules with the lighter isotope of oxygen evaporate first, leaving heavier ones behind. Because the water vapour in clouds isn’t the same mix of isotopes as in the ocean, we say it’s isotopically light: you can measure the ratio of light isotopes to heavy isotopes with a mass-spectrometer. When it condenses and falls as snow, that snow is also isotopically light compared to the sea ice.
When isotopically light snow turned into snow-ice by flooding, the snow-ice also appears isotopically light by comparison to the “normal” sea ice. We can therefore identify snow-ice in the lab with a mass-spectrometer. Unfortunately, the flooding process is normally reconstructed post-hoc. That is to say, a scientist will encounter a sea ice floe, suspect it has experienced flooding, extract an ice core, and then analyse it later in a laboratory setting to determine the snow-ice content. It’s rare to sample a floe before and after flooding. This leaves a lot of uncertainties about the snow-ice identification process. I wanted to investigate this by fully characterising the snow and ice before and after flooding and snow-ice formation.
The Churchill Experiment
Clement Soriot cutting out a mini-ice floe
I joined a pre-planned radar experiment at the Churchill Marine Observatory (CMO), organised by Julienne Stroeve and John Yackel (of U Manitoba and U Calgary, the two owners of the facility). It costs several thousand dollars per day to rent the CMO, so I was very lucky to be able to piggyback on their project! The CMO’s main facility is essentially a pair of outdoor swimming pools, each eight metres on a side and five metres deep. There’s a movable roof above, and a movable gantry that allows access to the interior of the pool. Churchill’s a pretty cold place in winter and spring (-30 °C at times when we were there), so the pools freeze over with sea ice rapidly unless you artificially heat the water.
To make sure our snow patches were vertically homegenous to start with, we manufactured two types of snow: large-grained (depth-hoar) and small-grained (wind-slab). By manufactured, I mean we found snow outside, mixed it, and sieved it into 10cm thick patches on the sea ice. I did a full physical characterisation of the snow, ice and seawater at this point. We then cut around our mini ice floes with an ice saw until they began to float freely.
Photograph showing capillary action from 5mm of flooding near the base of our manufactured, small-grained snowpack.
We then gently weighed down our artificial ice floes to simulate flooding, starting with the one covered with large-grained depth-hoar. There was an immediately striking effect: even a small amount of flooding led to upward capillary action in the snow, which weighed it down further; this essentially represented a small tipping point in the hydrostatic system, and was a dynamic that had never occurred to me. Something to look into I think.
The mini ice floe froze up within the larger floe over time, and the snow-ice froze up too. Once it had formed, I took vertical samples, but had to cut through with a saw and a corer to do so because I couldn’t use our snow density cutter in the ice. I split each sample into two, measuring the salinity of one part of the split, and bottling the other part for isotopic analysis. I then took the bottles back to the University of Calgary’s isotopic science laboratory, where they ran some fancy machinery to tell me the exact ratio of heavy to light oxygen isotopes in the samples. I just received the isotope data back, and it looks solid!
The Future + Gratitude
I’m planning to write up the method and results of our study, probably as a technical note in Ocean Sciences. The reason that I’m not (at this stage) thinking about a “full” journal article is because we really didn’t know how this study would go, and as such it wasn’t designed to test a hypothesis. Plan A was just having the flooding event work, having the isotopic analysis work too, and producing a baseline dataset. But still, I think the community will benefit from having a citable reference for our method, and I hope that by publishing this “proof of concept”, we can inspire others to tackle deeper methodological questions.
The on-site team. Left to right: me, Anton Komarov, John Yackel, Clement Soriot and Kiledar Tomar.
I’m grateful to John Yackel for supporting my SCAR fellowship application, and for handling a lot of the administrative work involved in using CMO, as well as contributing funding to make my fellowship application competitive. He also processed a lot of my samples for me when I was busy elsewhere! I’d also like to thank Julienne Stroeve (my old PhD supervisor and postdoc boss) for driving the campaign planning forward (particularly at a time when it was in doubt!), lending lots of field gear, and also bankrolling large parts campaign. In a series of logistics meetings before the campaign with John and Julienne, they were both very encouraging and supportive about trying to simulate the flooding event, in a way that many scientists would perhaps not be. I’d like to thank Clement Soriot, Anton Komarov, and Kiledar Tomar who worked super hard to make this happen too. Anton showed me how to manufacture the snow, and Clement and Kiledar both went early to CMO to make sure the shipping worked out. They were all then critical in the snow manufacturing itself (which is very physically taxing), the snow-pit measurements, the processing of samples in the lab, and orchestrating the cutting-and-flooding operation. It really was a team effort from start to finish, and was pleasingly international in terms of our nationalities: thanks again from me (GBR) to John (CAN) Julienne (USA), Clement (FRA), Anton (RUS) and Kiledar (IND).
I should close by thanking Cyril Fredlund, CMO’s resident technician. Cyril did everything in his power to make sure we had a succesful ten days at the facility, and tolerated an enourmous amount of mess and short-notice requests. He’s a consumate northerner, so he was also able to offer us really useful practical advice on snow and ice, alongside logistical help and local connections. Cyril will be retiring soon, and it will be a huge loss for the facility and sea ice research in Churchill when he does.