MOSAiC: A Year-Long City on Sea Ice

The Arctic Ocean is complicated, important and undergoing rapid climate change. As the sea ice thins and retreats, sunlight and Arctic winds are warming and stirring the ocean in an unprecedented way. This has serious implications for regional ecosystems, northern hemisphere weather and global climate.

Today I leave for the shrinking ice aboard Russian icebreaker Akademik Federov to help build a “city on ice”. The formal name for the city is the Multidisciplinary Observatory for the Study of Arctic Climate - MOSAiC for short. Its “downtown” will be the German research vessel RV Polarstern.

German icebreaker RV Polarstern will form the centerpiece of the drifting climate observatory.

Stacked full of food, fuel and scientists, Polarstern will be deliberately frozen into the sea ice for a full year in order to monitor the environment.

Surrounding the ship will be the largest array of technical instruments ever assembled on the sea ice in a historic, multimillion pound, multinational research endeavour.

MOSAiC has taken most of a decade to plan and prepare for. It represents the largest expedition to the central Arctic in history (by some margin!)

I’ll help set up and deploy these instruments and also participate in perhaps the greatest graduate summer school ever conceived: twenty PhD students will live, work and learn alongside science journalists and communicators, thrashing out the details of the Arctic climate system with the help of the MOSAiC scientists. But freezing a ship into the ice for a year isn’t easy or cheap. It requires an enormous amount of planning and money: 600 scientists from 19 countries will contribute to the 390 day expedition at a cost of more than €140m. So why bother?

We need to do this because the Arctic is the epicenter of climate change. Due to a phenomenon known as Arctic amplification, the region has seen a rise in temperature unparalleled anywhere else on the globe.

This shocking rise is forecast to continue as the world warms over the next century. Given that the Arctic drives much of Northern Hemisphere weather and sets global temperatures through its high reflectivity, this change in climate is extremely concerning. To compound our concern, scientific uncertainty abounds in the Arctic. While all mainstream climate models predict an unparallelled warming in the Arctic, inadequate representation of sea ice and cloud physics leads to large variation between the models with respect to size of the change. 

Observed change of temperature 1970 – 2017 [Degrees C]. The Arctic region has experienced unparalleled warming in recent years.

Observed change of temperature 1970 – 2017 [Degrees C]. The Arctic region has experienced unparalleled warming in recent years.

The data taken during MOSAiC’s operation will change the face of Arctic climate science and hopefully radically reduce these uncertainties. My research into the properties of snow on sea ice will be propelled forward as the expedition generates insight into how the internal properties of the snow-pack evolve. The complex issue of Arctic clouds and boundary-layer processes will be tackled head-on with measurements never before taken in the central arctic. All these measurements will ultimately be fed back into climate models and improve our understanding and predictions of future climate change.

A MOSAiC researcher practices with a tethered weather observation balloon in Svalbard

Polar scientists now frequently refer to the “New Arctic” - one characterised by diminished sea ice which is starkly younger and thinner than before. Understanding the New Arctic may require a New Science, one that’s multidisciplinary and multinational; this what we hope MOSAiC will be.

For more information, you can visit www.mosaic-expedition.org

To track the expedition as it’s built and as it drifts, visit follow.mosaic-expedition.org/

To read even more, visit en.wikipedia.org/wiki/MOSAiC_Expedition (I wrote the original article!)

America’s next top modelers: two weeks of climate science at NCAR, Colorado.

At the Eastern foot of the Rocky mountains in Wyoming sits the town of Cheyenne, home to about sixty thousand people. Founded shortly after the civil war, its stereotype is one of hard-working middle-America. But if you take Highway 210 to the outskirts of town, you’ll find an inhabitant that redefines the concept of hard work: the fortieth most powerful supercomputer in the world.

Cheyenne was ranked as the 40th most powerful supercomputer in the world by  TOP500 in June 2019 .

Cheyenne was ranked as the 40th most powerful supercomputer in the world by TOP500 in June 2019.

The eponymous “Cheyenne” is maintained by the US National Center for Atmospheric Research (NCAR), and is almost exclusively used for running computer models of the climate system. Cheyenne is mostly controlled remotely via the internet by scientists in Boulder, Colorado, and often from the iconic Mesa Lab which overlooks the city from a picturesque mountain plateau. I was invited to the Mesa Lab for two weeks to learn about NCAR’s in-house climate model and run it on Cheyenne to study the polar climate system.

The Mesa Lab sits on a small plateau at the foot of the Rocky Mountains looking over Boulder

The Mesa Lab sits on a small plateau at the foot of the Rocky Mountains looking over Boulder

Modern study of the climate depends on modeling the different systems that define it (such as sea ice, forests and ocean biology). Climate scientists now talk about “earth system” models rather than just ‘climate models’ to reflect this complexity. My first week in Boulder would be dedicated to exploring one of these earth system models: CESM (the Community Earth System Model). CESM simulates seven systems and their interactions in around two million lines of code.

The case of full coupling between atmosphere, ocean, sea ice, land ice, waves and the land surface is extremely computationally expensive and rarely run by specialist groups such as mine at the Center for Polar Observation and Modelling. Instead, fully coupled runs are carried out centrally on supercomputers like Cheyenne and the results are uploaded onto online archives for analysis.

 
Schematic of the earth systems coupled together in CICE v2

Schematic of the earth systems coupled together in CICE v2

 

Rather than run CESM in “fully coupled” mode, most scientists run simplified “cases” of the model to study specific systems. For instance, you may not need to compute the distribution of ocean waves in a study of how volcanic eruptions affect ozone production in the upper atmosphere. In the first week I became familiar with running these simplified cases and examining the output. The mornings were generally taken up by lectures from expert modelers, and afternoons were spent running CESM in different configurations and with different settings.

Keith Oleson helps tutorial students run CESM. Photo:  @NCAR_CGD

Keith Oleson helps tutorial students run CESM. Photo: @NCAR_CGD

Week two was dedicated to modeling the polar climate system, and the attendees were reduced from around 80 to 20. We began week two by forming teams and making predictions for this year’s minimum area of Arctic sea ice. The week then ran with a similar format to the first. Exercises now included modifying the source code and running the model at different complexity levels to deduce the behavior of individual systems.

In the evenings we made the most of the Mesa Lab’s dramatic geography, hiking the mountains above us and walking back into town for late dinners. The generous per diem from the workshop’s organisers allowed us to eat out together most nights, discussing science and a lot else. During the middle weekend, some of us rented a car for more ambitious hikes and experienced Boulder’s wildly changeable weather first hand. One social highlight was being kindly invited to Marika Holland’s house for an evening of empanadas, conversation and corn-hole (British readers should google it).

An opportunistic evening hike up the Bear Creek trail. Photo: Adi Nalam

An opportunistic evening hike up the Bear Creek trail. Photo: Adi Nalam

Over the fortnight I went from a total climate modeling novice to a confident user of CESM. Looking at the notes I took, I can now happily modify the source code, design my own model runs and process the output. More broadly, I have a much better understanding of nuanced but important concepts like the role of internal variability and model hierarchies. What’s more, I made lots of great contacts with sea ice scientists and polar modelers and have plans to do some cool stuff with them in the near future!

As well as the course being free, my flights, accommodation and per diem were generously funded by NCAR and its Polar Climate Working Group, which are in turn sponsored by the US National Science Foundation.

NCAR 2019 Polar Modeling Workshop participants and organizers. Photo:  David Bailey / Todd Amodeo

NCAR 2019 Polar Modeling Workshop participants and organizers. Photo: David Bailey/ Todd Amodeo

Lasers in space - NASA's ICESat-2 satellite counts photons from orbit

This week I attended a five day hack-week organised by the University of Washington Polar Science Center and the eScience institute, and taught by members of the NASA ICEsat-2 science team. The hack-week was made up of tutorials on data access and manipulation as well as collaborative projects on polar science.

group_photo.jpg

Light detection and ranging (Lidar) is nothing new in environmental science - it's used to map topography and vegetation in 3D and measure distances with extreme precision, helping scientists monitor systems from volcanic hazards to beach erosion.

But despite the heritage its the technology, NASA’s new ICEsat-2 satellite is a major leap forward, firing laser pulses at earth from low earth orbit and counting individual photons in as they return. By measuring their time of flight, it calculates elevations of the reflecting surfaces with centimetre accuracy. Even better, it does this with six laser beams, scanning a swath of terrain 9km wide. Each beam has a footprint of 17m diameter, offering unprecedented spatial resolution with a measurement every 70cm.

The six beams are divided into three pairs, with paired lasers 90m apart. This allows for measurements of gradient in the and 'across-track' directions, critical for steep and complex surfaces like glaciers.

poster.jpg

During the workshop we were quickly encouraged to pitch our ideas for projects, with attendees proposing ocean wave detection, mapping of ice sheet grounding lines and calculation of floe size distribution. I proposed a project to investigate the automatic blowing snow detection algorithm and was very pleased to have five people join my project! I wanted to compare the data to weather data from climate models, but we quickly branched out to mapping the distribution of blowing snow too.

We made a convincing climatology and our comparisons to reanalysis were encouraging, but we limited our scope to land-ice for practical reasons. We're now planning to extend our work to the ICEsat-2 sea ice product, where a climatology of blowing snow has never been made (to our knowledge).

Eric Keenan’s (@EricKeenanCU) plot of blowing snow optical depth over Antarctica. The optical thickness is higher over East Antarctica (figure right) in agreement with published work.

Eric Keenan’s (@EricKeenanCU) plot of blowing snow optical depth over Antarctica. The optical thickness is higher over East Antarctica (figure right) in agreement with published work.

A significant part of the week focussed on software tools like cloud computing in Jupyter Lab and Git, a collaboration and version control tool. It was great to be pushed to use these tools, as I wouldn't have done so otherwise. Git in particular offers our blowing snow team a chance to continue developing our product even now the hack-week is over.

As well as the chance for ongoing collaboration on blowing snow, ICEsat-2 has a lot to bring to my PhD project. I'm currently working on radar altimetry of the sea ice surface and encumbent assumptions about the spatial patterns of snow cover. ICEsat-2 offers the chance to validate radar altimetry, and also to shed light on model-generated snow distributions. During the hackweek I also had a couple of other ideas for novel uses of the data, which I'm going to keep under my hat for now! Perhaps just as valuably, I made some great connections with other PhD students with expertise in connected areas.

It clearly took a lot of time and effort to make this happen; thanks go in particular to the University of Washington eScience institute and Polar Science Center, and Anthony Arendt who brought it all together. I'm looking forward to all the icey science to follow!

Five days of Antarctic sea ice chat in Bern

Last week I joined twelve leading sea ice scientists at the International Space Science Institute (ISSI) in Bern for a meeting to discuss satellite measurements of Antarctic sea ice. ISSI offers generous support to young scientists like me to attend these meetings. But why is Antarctic sea ice so important but so difficult to measure by comparison to its Northern sibling?

Like that of the Arctic, Antarctic sea ice plays a key role in our climate. However, it’s chronically understudied, partly because nobody lives on Antarctica (apart from scientists). It also is less of a barrier to international shipping, a feature which motivates a significant chunk of Arctic sea ice research. Because of these reasons and others (discussed later), it’s relatively poorly understood.

While Arctic sea ice extent has declined sharply with global warming, Antarctic sea ice extent has slightly increased. Plot credit:  Zeke Hausfather, Yale Climate Connections

While Arctic sea ice extent has declined sharply with global warming, Antarctic sea ice extent has slightly increased. Plot credit: Zeke Hausfather, Yale Climate Connections

Unlike the Arctic Ocean, which is a sea almost entirely surrounded by continents, the Southern Ocean surrounds its own continental island. Because of this shape and the fact that the island sits over the pole, winds and ocean currents can swirl unimpeded round the bottom of the earth. This swirling of air and water isolates frozen continent from many of earth’s atmospheric and oceanic processes, keeping it cool and partly shielding it from polar amplification. This shielding has even helped sea ice increase its spatial coverage with recent global warming. Even with this shielding, Southern sea ice has its own challenges to growth. Rather than sitting in a protected ocean, southern sea ice is free to float northwards into warmer water and melt, meaning that southern sea ice rarely survives a summer and is in general young and thin compared to northern sea ice.

Antarctic sea ice surrounds an island, whereas Arctic sea ice is largely enclosed by continents. Figure shows northern hemisphere summer sea ice extent, so Arctic sea ice coverage is anomalously low and Antarctic coverage is anomalously high. Figure credit:  NASA Global Climate Change Blog

Antarctic sea ice surrounds an island, whereas Arctic sea ice is largely enclosed by continents. Figure shows northern hemisphere summer sea ice extent, so Arctic sea ice coverage is anomalously low and Antarctic coverage is anomalously high. Figure credit: NASA Global Climate Change Blog

While we have a good idea how much area is covered by Antarctic sea ice (on average just less than Arctic sea ice at ~10 million square kilometres), data on its thickness is patchy at best. Because of this, we really don’t know how much sea ice actually exists around Antarctica. In the Arctic we can use radar from satellites to measure how much the ice protrudes out of the water, and from this work out how much ice is submerged. In contrast, in the Antarctic, the snow is often so thick that it weighs down the surface of the ice enough to totally submerge it. When this happens we can’t measure ice thickness directly. Adding to the difficulty, Arctic snow generally permits radar penetration and allows us to measure the ice surface. Antarctic snow is much more prone to reflecting satellite radar and so blocks direct measurment of the sea ice.

In fact, snow is regarded among scientists as the major barrier to a thorough understanding of Antarctic sea ice thickness. Its depth, density and microscopic properties affect both our view of the ice from satellites and the processes that form and melt of the ice itself. Scientists are yet to develop a “snow map” for the Antarctic sea ice, a source of consternation for those wanting to study the ice below.

Despite these barriers to research, a full understanding Antarctic sea ice processes is an attractive prize. Our ability to model and predict the climate depends on our estimation of the fresh-water locked up in sea ice each winter. Extremely salty water is rejected from the ice as it freezes and drives global ocean circulation. Finally, the thickness of the ice (and the overlying snow) regulates the light that reaches the surface ocean, fuelling the Southern Ocean food web. Scientists continue to untangle these interrelated processes, and it was great to be a part of it in Bern last week – thanks again to ISSI and to Petra Heil and Rachel Tilling for inviting me.