CR
Cryospheric Sciences

Image of the Week

Image of the Week – The shape of (frozen sea) water

 

Figure 1: Annual evolution of the sea ice area with two different floe shape parameters of 0.44 (red) and 0.88 (blue). The model is spun-up between 2000 – 2006 and then evaluated for a further ten years between 2007 – 2016 and the mean values over this period displayed by the thick lines. Thin lines show the results for individual years. [Credit: Adam Bateson]

Polar sea ice exists as isolated units of ice that we describe as floes. These floes do not have a constant shape (see here for instance); they can vary from almost circular to being jagged and rectangular. However, sea ice models currently assume that all floes have the same shape. Much focus has been paid to the size of floes recently, but do we also need to reconsider how floe shape is treated in models?


Why might floe shape matter?

In recent years, sea ice models have started to examine more and more how individual floes influence the overall evolution of sea ice.

A particular focus has been the size of floes (see here and here) and the parameterisation of processes which influence floe size (see here for example). However less attention has been given to the shape of the floe. The shape of the floe is important for several reasons:

  • Lateral melt rate: the lateral melt rate describes how quickly a floe melts from its sides. Two floes with the same area but different shape can have a different perimeter; the lateral melt rate  is proportional to the floe perimeter.
  • Wave propagation: a straight floe edge will impact propagating waves differently to a curved or jagged floe edge. The distance waves travel under the sea ice and hence the extent of sea ice that waves can fragment will be dependent on these wave-floe edge interactions.
  • Floe mechanics: an elongated floe (i.e. much longer in one direction than another) will be more likely to break from incoming waves if its longer edge is aligned with the direction the waves are travelling.

How do models currently treat floe shape?

One approach used within sea ice models to define floe shape is the use is the use of a parameter, α. The smaller the floe shape parameter, the longer the floe perimeter (and hence, the higher the lateral melt rate). A standard value used for the parameter is 0.66 (Steele, 1992). Figure 2 shows how this floe shape parameter varies for some common shapes.

Figure 2: The floe shape parameters for some common shapes are given for comparison to the standard value of 0.66. [Credit: Adam Bateson]

The standard value of the floe shape parameter, 0.66, was obtained from taking the mean floe shape parameter measured over all floes greater than 1 km from a singular study area of 110 km x 95 km at one snapshot in time. Despite the limited data set used to estimate this shape parameter, it is being used for all sea ice throughout the year for all floe sizes. However, this would only be a concern to the accuracy of modelling if it turns out that sea ice evolution in models is sensitive to the floe shape parameter.

 

Model sensitivity to floe shape

To investigate the model sensitivity to the floe shape parameter two simulations have been run: one uses a floe shape parameter of 0.88 and the other uses 0.44, chosen to represent likely extremes. The two simulations are run from 2000 – 2016, with 2000 – 2006 used as a spin-up period. Figure 1 displays the mean total ice area throughout the year and results of individual years for each simulation. Figure 3 is an equivalent plot to show the annual evolution of total ice volume for each simulation.

The results show that the perturbation from reducing the floe shape parameter is smaller than the variation between years within the same simulation.  However, the model does show a permanent reduction in volume throughout the year and a 10 – 20 % reduction in the September sea ice minimum. The impact of the floe shape is hence small but significant, particularly for predicting the annual minimum sea ice extent and volume.

Figure 3: Annual evolution of the sea ice volume with two different floe shape parameters of 0.44 (red) and 0.88 (blue). The model is spun-up between 2000 – 2006 and then evaluated for a further ten years between 2007 – 2016 and the mean values over this period displayed by the thick lines. Thin lines show the results for individual years.

More recent studies on floe shape

In 2015, Gherardi and Lagomarsino analysed the floe shape behaviour from four separate samples of satellite imagery from both the Arctic and Antarctic. The study found different distributions of floe shapes in different locations, however there was no correlation between floe shape and size. This property would allow models to treat floe shape and size as independent properties. More recently, in 2018, Herman et al. analysed the results of laboratory experiments of ice breaking by waves. It was found that wave break-up influenced the shape of the floes, tending to produce straight edges and sharp angles.  These features are associated with a smaller floe parameter i.e. would produce an increased lateral melt rate.

What next?

More observations are needed to identify whether the use of a constant floe shape parameter is justified. The following questions are important:

  • Do further observations support the finding that floe size and shape are uncorrelated?
  • What range of values for the floe shape parameter can be observed in reality?
  • Do we see significant variations in the floe shape parameter between locations?
  • Do these variations occur over a large enough scale that they can be represented within existing model resolutions?

Further reading

Edited by Violaine Coulon and Sophie Berger


Adam Bateson is a PhD student at the University of Reading (United Kingdom), working with Danny Feltham. His project involves investigating the fragmentation and melting of the Arctic seasonal sea-ice cover, specifically improving the representation of relevant processes within sea-ice models. In particular he is looking at lateral melting and wave induced fragmentation of sea-ice as drivers of break up, as well as the role of the ocean mixed layer as either an amplifier or dampener to the impacts of particular processes. Contact: a.w.bateson@pgr.reading.ac.uk or @a_w_bateson on twitter.

Image of the Week – Making waves: assessing supraglacial water storage for debris-covered glaciers

Fig. 1: Deriving the bathymetry and temperature of a large supraglacial pond on Khumbu Glacier, Everest region of Nepal. The sonar-equipped unmanned surface vessel nicknamed ‘BathyBot’ (left), and kayak retrieval of temperature loggers (right) [Credit: Scott Watson].

A creeping flux of ice descends Everest, creating the dynamic environment of Khumbu Glacier. Ice and snow tumble, debris slumps, ice cliffs melt, englacial cavities collapse, ponds form and drain, all responding to a variable energy balance. Indeed, Khumbu Glacier is a debris-covered glacier, meaning it features a layer of sediment, rocks and house-sized boulders that covers the ice beneath. Recent advances in understanding debris-covered glacier hydrology come from combining in situ surveys with remotely sensed satellite data.


Khumbu Glacier

The dramatic beauty of Nepal’s Everest region attracts a mix of trekkers, climbers, and scientists. Flowing down from the slopes of Mount Everest, the debris-covered Khumbu Glacier has drawn scientists from the mid-1900s, and offers temporary residence for research teams and a myriad of climbers. In some locations, Khumbu Glacier has thinned by up to 80 m in the last three decades, leading to moraines overlooking the glacier with impressive topographic relief and providing an instant visualisation of glacier mass loss for trekkers heading to Everest Base Camp.

Melt at the surface of this glacier is moderated by an undulating debris layer, which insulates the ice beneath,   and enhanced locally by dynamic surface features such as supraglacial ponds and ice cliffs thinly veiled by debris. These features contribute disproportionately to melt and lead to the development of hummocky, pitted surface topography. The resulting variable surface topography and melt rates complicate meltwater runoff and flow routing across the glacier. To better understand them, in situ surveying (Fig. 1) is increasingly combined with fine spatial-temporal resolution satellite imagery to reveal the hydrological evolution of debris-covered glaciers, which is closely linked to their mass loss.

Hydrology of Khumbu Glacier

As with debris-free glaciers, water may be routed through supraglacial, englacial, and subglacial pathways, which are conceptually distinct but physically link to one another.

At Khumbu Glacier, surface channels collect and rapidly convey meltwater generated in the upper ablation area (Fig. 2), just below the treacherous Khumbu Icefall, incising at a faster rate than the surface melt. In the middle of the debris-covered area, such streams disappear into the glacier’s interior through cut-and-closure and/or hydrofracture.

Fig.2: The upper ablation area of Khumbu is drained by supraglacial channels which enter the glacier’s interior through hydrofracture and cut-and-closure, while the lower portion is characterised by pitted surface depressions and an increasing density of ponds. Right panel looking east to west shows the hummocky topography and ponding on Khumbu Glacier. [Credit: Evan Miles (left), Ann Rowan (right)].

In areas of low surface gradient , and particularly throughout the hummocky lower reaches of the glacier, supraglacial ponds collect water in surface depressions. These features haveregulate the runoff of debris-covered glaciers by seasonally storing meltwater. The annual melt cycle thus leads to pond expansion and contraction, or their disappearance when the protecting debris layer thaws and relict meltwater conduits become avenues for drainage (Fig 3). The areal fluctuation of ponds can be quantified using  satellite images at different times, but cloud cover during the summer monsoon season limits useable imagery at a time when the ponds are most dynamic. Therefore, field-instrumented ponds provide valuable insights into their active melt season behaviour.

Fig. 3: A small 4.5 m deep pond that drained over the course of a year [Credit: Watson et al., 2017a].

Turbid ponds associated with debris influx from ice cliffs are often ephemeral but some can grow to hold vast quantities of water (Fig. 1). Stored water absorbs and transmits solar energy to melt adjacent ice, which generates additional meltwater and leads to pond expansion. The ponds also thermally undercut ice cliffs, leading to both subaqueous and subaerial  retreat (Fig. 4). Khumbu Glacier has been developing a growing network of ponds in recent years, which means meltwater is increasingly stored on the surface of the glacier before contributing to downstream river discharge. Ponds that coalesce into larger and more persistent lakes behind unstable deposits of sediment can in some cases pose a hazard  to downstream communities. Field and satellite-based techniques are therefore used simultaneously to monitor lake development.

 

Fig. 4. Supraglacial ponds often exist alongside ice cliffs. These ‘hot spots’ of melt can be observed with repeat point cloud differencing [Credit: Watson et al., 2017b]. An interactive view of the drained pond basin (right) is available here.

What lies beneath?

Ephemeral ponds drain into the ‘black box’ glacier interior, where relatively little is known about the internal structure and hydrology. Scientists have occasionally ventured into the subsurfac e realm through networks of englacial conduits that become exposed as the glacier thins (Fig. 5); such conduits often re-emerge at the glacier surface but may also lead to the bed. The conduits carry meltwater through the glacier but can become dormant if blocked by falling debris or creeping ice, or when the meltwater that sustains them finds a route of lesser resistance. Whilst satellite data can be used to infer the presence of conduits, field-based methods are required for hydrological budgeting and quantifying meltwater transit times. For example, dye tracing can detect the subsurface passage of meltwater where strategically placed fluorometers measure the receipt and dilution of the dye upon re-emergence. Such methods are crucial for developing an improved understanding of the links between, for example, flow in the supraglacial channels up-glacier and discharge at the outlet.

Fig. 5: An exposed conduit on Lirung Glacier (left) [Credit: Miles et al., 2017] and researchers inside a conduit on Ngozumpa Glacier (right) [Credit: Benn et al., 2017].

 

Outlook

Multiple teams working across the Himalaya are advancing our understanding of debris-covered glacier hydrology, which is essential to forecast their future and quantify their downstream impact. With the ready availability of increasingly high temporal resolution satellite imagery (e.g. Sentinel-2, Planet Labs), the link between field and spacebourne observations will become increasingly complementary. Developing these links is crucial to upscale observations from specific sites more broadly across the Himalaya.

Further reading

Edited by Violaine Coulon and Sophie Berger


Scott Watson is a Postdoc at the University of Arizona, USA. He studies glaciers in the Everest region and the surface interactions of supraglacial ponds and ice cliffs. He also investigates natural hazards and the implications of glacial lake outburst floods.
Tweets @CScottWatson. Website: www.rockyglaciers.co.uk

 

 

Evan Miles is a Research Fellow at the University of Leeds, UK, where he is a part of the EverDrill project’s hot-water drilling at Khumbu Glacier. His recent work has examined the seasonal hydrology and dynamics of debris-covered glaciers, with a focus on the melt associated with dynamic surface features such as supraglacial ice cliffs and ponds.
Tweets @Miles_of_Ice

EverDrill website: www.EverDrill.org

Image of the Week — Quantifying Antarctica’s ice loss

Fig. 1 Cumulative Antarctic Ice Sheet mass change since 1992. [Credit: Fig 2. from The IMBIE team (2018), reprinted with permission from Nature]

It is this time of the year, where any news outlet is full of tips on how to lose weight rapidly to  become beach-body ready. According to the media avalanche following the publication of the ice sheet mass balance inter-comparison exercise (IMBIE) team’s Nature paper, Antarctica is the biggest loser out there. In this Image of the Week, we explain how the international team managed to weight Antarctica’s ice sheet and what they found.


Estimating the Antarctic ice sheet’s mass change

There are many ways to quantify Antarctica’s mass and mass change and most of them rely on satellites. In fact, the IMBIE team notes that there are more than 150 papers published on the topic. Their paper that we highlight this week is remarkable in that it combines all the methods in order to produce just one, easy to follow, time series of Antarctica’s mass change. But what are these methods? The IMBIE team  used estimates from three types of methods:

  •  altimetry: tracking changes in elevation of the ice sheet, e.g. to detect a thinning;
  •  gravimetry: tracking changes in the gravitational pull caused by a change in mass;
  •  input-output: comparing changes in snow accumulation and solid ice discharge.

To simplify, let’s imagine that you’re trying to keep track of how much weight you’re losing/gaining. Then  altimetry would be like looking at yourself in a mirror, gravimetry would be stepping on a scale, and input-output would be counting all the calories you’re taking in and  burning out. None of these methods will tell you directly whether you have lost belly fat, but combining them will.

The actual details of each methods are rather complex and cover more pages than the core of the paper, so I invite you to read them by yourself (from page 5 onwards). But long story short, all estimates were turned into one unique time series of ice sheet mass balance (purple line on Fig. 1). Furthermore, to understand how each region of Antarctica contributed to the time series, the scientists also produced one time series per main  Antarctic region (Fig. 2): the West Antarctic Ice Sheet (green line), the East Antarctic Ice Sheet (yellow line), and the Antarctic Peninsula (red line) .

Antarctica overview map. [Credit: NASA]

Antarctica is losing ice

The results are clear: the Antarctic ice sheet as a whole is losing mass, and this mass loss is accelerating. Nearly 3000 Giga tonnes since 1992. That is 400 billion elephants in 25 years, or on average 500 elephants per second.

Most of this signal originates from West Antarctica, with a current trend of 159 Gt (22 billion elephants) per year. And most of this West Antarctic signal comes from the Amundsen Sea sector, host notably to the infamous  Pine Island  and Thwaites Glaciers.

The Antarctic ice sheet has lost “400 billion elephants in 25 years”

But how is the ice disappearing? Rather, is the ice really disappearing, or is there simply less ice added to Antarctica than ice naturally removed, i.e. a change in surface mass balance? The IMBIE team studied this as well. And they found that there is no Antarctic ice sheet wide trend in surface mass balance; in other words Antarctica is shrinking because more and more ice is discharged into the ocean, not because it receives less snow from the atmosphere.

Floating ice shelf in the Halley embayment, East Antarctica [Credit: Céline Heuzé]

What is happening in East Antarctica?

Yet another issue with determining Antarctica’s weight loss is Glacial Isostatic Adjustment. In a nutshell, ice is heavy, and its weight pushes the ground down. When the ice disappears, the ground goes back up, but much more slowly than the rate of ice melting . This process has been ongoing in Scandinavia notably since the end of the last ice age 21 000 years ago, but it is also happening in East Antarctica by about 5 to 7 mm per year (more information here). Except that there are very few on site GPS measurements in Antarctica to determine how much land is rising, and the many estimations of this uplifting disagree.

So as summarised by the IMBIE team, we do not know yet what the change in ice thickness is where glacial isostatic adjustment is strong, because we are unsure how strong this adjustment is there. As a result in East Antarctica, we do not know whether there is ice loss or not, because it is unclear what the ground is doing.

What do we do now?

The IMBIE team concludes their paper with a list of required actions to improve the ice loss time series: more in-situ observations using airborne radars and GPS, and uninterrupted satellite observations (which we already insisted on earlier).

What about sea level rise, you may think. Or worse, looking at our image of the week, you see the tiny +6mm trend in 10 years and think that it is not much. No, it is not. But note that the trend is far from linear and has been actually accelerating in the last decades…

 

Reference/Further reading

The IMBIE Team, 2018. Mass balance of the Antarctic Ice Sheet from 1992-2017. Nature 558, 219–222.

Edited by Sophie Berger

Image of the Week — Orange is the new white

Figure 1. True color composite of a Sentinel-2 image showing the dust plume off the coast of Libya on 22-Mar-2018 (see also on the ESA website) [Credit: processed by S. Gascoin]

On 22 March 2018, large amounts of Saharan dust were blown off the Libyan coast to be further deposited in the Mediterranean, turning the usually white snow-capped Mountains of Turkey, Romania and even Caucasus into Martian landscapes.  As many people were struck by this peculiar color of the snow, they started documenting this event on social media using the “#orangesnow hashtag”. Instagram and twitter are fun, but satellite remote sensing is more convenient to use to track the orange snow across mountain ranges. In this new image of the week, we explore dusty snow with the Sentinel-2 satellites…

Марс атакует 🌔 #smurygins_family_trip

A post shared by Alina Smurygina (@sinyaya_ptiza) on


Sentinel-2: a great tool for observing dust deposition

Sentinel-2 is a satellite mission of the Copernicus programme and consists of two twin satellites (Sentinel-2A and 2B). Although the main application of Sentinel-2 is crop monitoring, it is also particularly well suited for characterizing the effect of dust deposition on the snowy mountains because:

  1. Sentinel-2A and 2B satellites provide high-resolution images with a pixel size of 10 m to 20 m (depending on the spectral band), which enables to detect dust on snow at the scale of hillslopes.
  2. Sentinel-2 has a high revisit capacity of 5 days which increases the probability to capture cloud-free  images shortly after the dust deposition.
  3. Sentinel-2 has many spectral bands in the visible and near infrared region of the light spectrum, making easy to separate the effect of dust on snow reflectance — i.e. the proportion of light reflected by snow — from other effects due to snow evolution. The dust particles mostly reduce snow reflectance in the visible, while coarsening of the snow by metamorphism (i.e. the change of microstructure due to transport of vapor at the micrometer scale) tends to reduce snow reflectance in the near infrared (Fig. 2).
  4. Sentinel-2 radiometric observations have high dynamic range and are accurate and well calibrated (in contrast to some trendy miniature satellites), hence they can be used to retrieve accurate surface dust concentration, provided that the influence of the atmosphere and the topography on surface reflectance are removed.

Figure 2: Diffuse reflectance for different types of snowpack. These spectra were computed with 10 nm resolution using the TARTES model (Libois et al, 2013) using the following parameters: snowpack density: 300 kg/m3, thickness: 2 m, fine snow specific surface area (SSA): 40 m2/kg, coarse snow SSA: 20 m2/kg, dust content: 100 μg/g. The optical properties of the dust are those of a sample of fine dust particles from Libya with a diameter of 2.5 μm or less (PM2.5) (Caponi et al, 2017). The Sentinel-2 spectral bands are indicated in grey. [Credit: S. Gascoin]

Dust on snow from Turkey to Spain

The region of Mount Artos in the Armenian Highlands (Turkey) was one of the first mountains to be imaged by Sentinel-2 after the dust event. Actually Sentinel-2 even captured the dust aloft on March 23, before its deposition (Fig. 3)

Figure 3: Time series of three Sentinel-2 images near Mount Artos in Turkey (true color composites of level 1C images, i.e. orthorectified products, without atmospheric correction). [Credit: Contains modified Copernicus Sentinel data, processed by S. Gascoin]

Later in April another storm from the Sahara brought large amounts of dust in southwestern Europe.

Figure 4: Sentinel-2 images of the Sierra Nevada in Spain (true color composites of level 1C images). [Credit: Contains modified Copernicus Sentinel data, processed by S. Gascoin]

This example in the spanish Sierra Nevada nicely illustrates the value of the Sentinel-2 mission since both images were captured only 5 days apart. The high resolution of Sentinel-2 is also important given the topographic variability of this mountain range. This is how it looks in MODIS images, having a 250 m resolution.

Figure 5: MODIS Terra (19) and Aqua (24) images of the Sierra Nevada in Spain. True Color composites of MODIS corrected reflectance. [Credit: NASA, processed by S. Gascoin]

Sentinel-2 satellites enable to track the small-scale variability of the dust concentration in surface snow, even at the scale of the ski runs as shown in Fig. 6.

Figure 6: Comparison of a true color Sentinel-2 image and a photograph of the Pradollano ski resort, Sierra Nevada. [Credit: photograph taken by J. Herrero / Contains modified Copernicus Sentinel data, processed by S. Gascoin]

A current limitation of Sentinel-2, however, is the relative shortness of the observation time series. Sentinel-2A was only launched in 2015 and Sentinel-2B in 2017. With three entire snow seasons, we can just start looking at interannual variability. An example in the Prokletije mountains in Albania is shown in Fig. 7.

Figure 7. Sentinel-2 images of the Prokletije mountains in Albania (true color composites of level 1C images) [Credit: Contains modified Copernicus Sentinel data, processed by S. Gascoin]

These images suggest that the dust event of March 2018 was not exceptional in this region, as 2016 also highlights a similar event. The Sentinel-2 archive will keep growing for many years since the EU Commission seems determined to support the continuity and development of Copernicus programme in the next decades. In the meantime to study the interannual variability the best option is to exploit the long-term records from other satellites like MODIS or Landsat.

Beyond the color of snow, the water resource

Dust on snow is important for water resource management since dust increases the amount of solar energy absorbed by the snowpack, thereby accelerating the melt. A recent study showed that dust controls springtime river flow in the Western USA (Painter et al, 2018).

“It almost doesn’t matter how warm the spring is, it really just matters how dark the snow is.”

said snow hydrologist Jeff Deems in an interview about this study in Science Magazine. Little is known about how this applies to Europe…

Further reading

 Edited by Sophie Berger


Simon Gascoin is a CNRS researcher at Centre d’Etudes Spatiales de la Biosphère (CESBIO), in Toulouse. He obtained a PhD in hydrology from Sorbonne University in Paris and did a postdoc on snow and glacier hydrology at the Centro de Estudios Avanzados en Zonas Áridas (CEAZA) in Chile. His research is now focusing on the application of satellite remote sensing to snow hydrology. He tweets here and blog here.

 

 

Marie Dumont is a researcher, leading the snow processes, observations and modelling research team at the snow study centre (CNRM/CEN, Grenoble, France). Her research focuses on snow evolution mostly in alpine region using numerical modelling and optical remote sensing.

 

 

 

Ghislain Picard is a lecturer working at the Institute of Geosciences and Environment at the University Grenoble Alpes, in the climate and ice-sheets research group. His research focuses on snow evolution in polar regions in the context of climate change. Optical and microwave remote sensing is one of its main tools.

Image of the Week – Antarctica: A decade of dynamic change

Fig. 1 – Annual rate of change in ice sheet height attributable to ice dynamics. Zoomed regions show (a) the Amundsen Sea Embayment and West Marie Byrd Land sectors of West Antarctica, (b) the Bellingshausen Sea Sector including the Fox and Ferrigno Ice Streams and glaciers draining into the George VI ice shelf and (c) the Totten Ice Shelf. The results are overlaid on a hill shade map of ice sheet elevation from Bedmap2 (Fretwell et al. 2013) and the grounding line and ice shelves are shown in grey (Depoorter et al. 2013). [Credit: Stephen Chuter]

  

Whilst we tend to think of the ice flow in Antarctica as a very slow and steady process, the wonders of satellites have shown over the last two decades it is one of the most dynamic places on Earth! This image of the week maps this dynamical change using all the satellite tools at a scientist’s disposal with novel statistical methods to work out why the change has recently been so rapid.


Why do we care about dynamic changes in Antarctica ?!

The West Antarctic Ice Sheet has the potential to contribute an approximate 3.3 m to global sea level rise (Bamber et al. 2009). Therefore, being able to accurately quantify observed ice sheet mass losses and gains is imperative for assessing not only their current contribution to the sea level budget, but also to inform ice sheet models to help better predict future ice sheet behaviour.

An ice sheet can gain or lose mass primarily through two different processes:

  • changes in surface mass balance (variations in snowfall and surface melt driven by atmospheric processes) or
  • ice dynamics, which is where variations in the flow of the ice sheet (such as an increase in its velocity) leads to changes in the amount of solid ice discharged from the continent into the ocean. In Antarctica ice flow dynamics are typically regulated by the ice shelves that surround the ice sheet; which provide a buttressing stress to help hold back the rate of flow.

Understanding the magnitude of each of these two components is key to understanding the external forcing driving the observed ice sheet changes.

This Image of the Week shows the annual rates of ice sheet elevation change which are attributed to changes in ice dynamics between 2003 and 2013 (Fig. 1) (Martín-Español et al. 2016). This is calculated by combining observations from multiple satellites (GRACE, ENVISAT, ICESat and CryoSat-2) with in-situ GPS measurements in  a Bayesian Hierarchical Model. The challenge we face is that the observations we have of ice sheet change (whether that being total height change from altimetry or mass changes from GRACE) vary on their spatial and temporal scales and can only tell us the total mass change signal, not the magnitudes or proportions of the underlying processes driving it. The Bayesian statistical approach used here takes these observations and separates them proportionally into their most likely processes, aided by prior knowledge of the spatial and temporal characteristics for each process we want to resolve. This allows us reducing the reliance on using forward model outputs to resolve for processes we cannot observe. As a result, it is unique from other methods of determining ice sheet mass change, which rely on model outputs which in some cases have hard to quantify uncertainties.  This methodology has been applied to Antarctica and is currently being used to resolve the sea level budget and its constituent components through the ERC GlobalMass project.

What can we learn from Bayesian statistical approach?

This approach firstly allows us to quantitively assess the annual contribution that the Antarctic ice sheet is making to the global sea level budget, which is vital to better understanding the magnitude each Earth system process is playing in sea level change. Additionally, by being able to break down the total change into its component processes, we can better understand what external factors are driving this change. Ice dynamics has been the dominant component of mass loss in recent years over the West Antarctic Ice Sheet and is therefore the process being focussed on in this image.

Amundsen Sea Embayment : a rapidly thinning area

Since 2003 there have been major changes in the dynamic behaviour over the Amundsen Sea Embayment and West Marie Byrd Land region (Fig 1, inset a). This region is undergoing some of the most rapid dynamical changes across Antarctica, with a 5 m/yr ice dynamical thinning near the outlet of the Pope and Smith Glacier. Additionally the Bayesian hierarchical model results show that dynamic thinning has spread inland from the margins of Pine Island Glacier, agreeing with elevation trends measured by satellite altimetry over the last two decades (Konrad et al. 2016).

These changes are driven primarily by the rapid thinning of the floating ice shelves at the ice sheet margin in this region

The importance of ice dynamics  is also illustrated in Fig 2, which shows  surface processes and ice dynamics components of mass changes over the Amundsen Sea Embayment from the bayesian hierarchical model. Fig 2 demonstrates that ice dynamics is the primary driver of mass losses in the region. Ice dynamic mass loss increased dramatically from 2003-2011, potentially stabilising to a new steady state since 2011.

Fig. 2 – Annual mass changes due to ice dynamics (pink line) and SMB (blue line) for the period 2003-2013 from the Bayesian hierarchical model approach. Red dots represent mass change anomaly (changes from the long term mean) due to surface mass balance calculated by the RACMO2.3 model and allow for comparison with our Bayesian framework results. (calculated from observations of ice velocity and ice thickness at the grounding line and allow for comparison with our Bayesian framework results (Mouginot et al, 2014). [Credit: Fig. 9b from Martín-Español et al., 2016].

 

The onset of  dynamic thinning can also be seen in glaciers draining into the Getz Ice Shelf, which is experiencing high localised rates of ice shelf thinning up to 66.5 m per decade (Paolo et al. 2015) . This corroborates with ice speed-up recently seen in the region (Chuter et al. 2017; Gardner et al. 2018). We have limited field observations of ice characteristics in this region and therefore more extensive surveys are required to fully understand causes of this dynamic response.

Bellingshausen Sea Sector :  Not as stable as previously thought…

 The Bellingshausen Sea Sector (Fig 1, inset b) was previously considered relatively a dynamically stable section of the Antarctic coastline, however recent analysis from a forty year record of satellite imagery has shown that the majority of the grounding line in this region has retreated  (Christie et al. 2016). This is reflected in the presence of a dynamic thinning signal in the bayesian hierarchical model results near the Fox and Ferrigno Ice streams and over some glaciers draining into the George VI ice shelf, which have been observed from CryoSat-2 radar altimetry (Wouters et al. 2015). The dynamic changes in this region over the last decade highlight the importance of continually monitoring all regions of the ice sheet with satellite remote sensing in order to understand the what the long term response over multiple decades is to changes in the Earth’s climate and ocean forcing.

Outlook

Multiple  satellite missions have allowed us to measure changes occurring across the ice sheet in unprecedented detail over the last decade. The launch of the GRACE-Follow On mission earlier this week and the expected launch of ICESat-2 in September will ensure this capability continues well into the future. This will provide much needed further observations to allow us to understand ice sheet dynamics over time scales of multiple decades. The bayesian hierarchical approach being demonstrated will be developed further to encompass these new data sets and extend the results into the next decade. In addition to satellite measurements, the launch of the International Thwaites Glacier Collaboration  between NERC and NSF will provide much needed field observations for the Thwaites Glacier region of the Amundsen Sea Embayment, to better understand whether it has entered a state of irreversible instability .

Data
The  Bayesian hierarchical model mass trends shown here (Martín-Español et al. 2016) are available from the UK Polar Data Centre. In addition, the time series has been extended until 2015 and is available on request from Stephen Chuter (s.chuter@bristol.ac.uk). This work is part of the ongoing ERC GlobalMass project, which aims to attribute global sea level rise into its constituent components using a Bayesian Hierarchical Model approach. The GlobalMass project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 69418.

References

Christie, Frazer D. W. et al. 2016. “Four-Decade Record of Pervasive Grounding Line Retreat along the Bellingshausen Margin of West Antarctica.” Geophysical Research Letters 43(11): 5741–49. http://doi.wiley.com/10.1002/2016GL068972.

Chuter, S.J., A. Martín-Español, B. Wouters, and J.L. Bamber. 2017. “Mass Balance Reassessment of Glaciers Draining into the Abbot and Getz Ice Shelves of West Antarctica.” Geophysical Research Letters 44(14).

Gardner, Alex S. et al. 2018. “Increased West Antarctic and Unchanged East Antarctic Ice Discharge over the Last 7 Years.” Cryosphere 12(2): 521–47.

Martín-Español, Alba et al. 2016. “Spatial and Temporal Antarctic Ice Sheet Mass Trends, Glacio-Isostatic Adjustment, and Surface Processes from a Joint Inversion of Satellite Altimeter, Gravity, and GPS Data.” Journal of Geophysical Research: Earth Surface 121(2): 182–200. http://dx.doi.org/10.1002/2015JF003550.

Mouginot, J, E Rignot, and B Scheuchl. 2014. “Sustained Increase in Ice Discharge from the Amundsen Sea Embayment, West Antarctica, from 1973 to 2013.” Geophysical Research Letters 41(5): 1576–84.

Paolo, Fernando S, Helen A Fricker, and Laurie Padman. 2015. “Volume Loss from Antarctic Ice Shelves Is Accelerating.” Science 348(6232): 327–31. http://www.sciencemag.org/content/early/2015/03/31/science.aaa0940.abstract.

Edited by Violaine Coulon and Sophie Berger


Stephen Chuter is a post-doctoral research associate in Polar Remote Sensing and Sea Level at the University of Bristol. He combines multiple satellite and ground observations of ice sheet and glacier change with novel statistical modelling techniques to better determine their contribution to the global sea level budget. He tweets as @StephenChuter and can be found at www.stephenchuter.wordpress.com. Contact email: s.chuter@bristol.ac.uk

Image of the Week — Biscuits in the Permafrost

Fig. 1: A network of low-centred ice-wedge polygons (5 to 20 m in diameter) in Adventdalen, Svalbard [Credit: Ben Giles/Matobo Ltd]

In Svalbard, the snow melts to reveal a mysterious honeycomb network of irregular shapes (fig. 1). These shapes may look as though they have been created by a rogue baker with an unusual set of biscuit cutters, but they are in fact distinctive permafrost landforms known as ice-wedge polygons, and they play an important role in the global climate.


Ice-wedge polygons: Nature’s biscuit-cutter

In winter, cracks form when plummeting air temperatures cause the ground to cool and contract. O’Neill and Christiansen (2018) used miniature accelerometers to detect this cracking, and found that it causes tiny earthquakes, with large magnitude accelerations (from 5 g to at least 100 g (where g = normal gravity)!). Water fills the cracks when snow melts. When the temperature drops, the water refreezes and expands, widening the cracks. Over successive winters, the low tensile strength of the ice compared to the surrounding sediment means that cracking tends to reoccur in the ice. As the cycle of cracking, infilling, and refreezing continues over centuries to millennia, ice wedges develop.

Subsurface ice wedge growth causes small changes in the ground surface microtopography. There are linear depressions, known as troughs, above the ice wedges (fig. 2). Adjacent to the troughs, the soil is pushed up into raised rims. From these raised rims, the elevation drops off into the polygon centre, forming low-centred polygons (fig. 2a).

Shaping Arctic landscapes

Permafrost in the Northern hemisphere is warming due to increasing air temperatures (Romanovsky et al. (2010). As air temperatures rise, the active layer (the ground that thaws each summer and refreezes in winter) deepens.

As permafrost with a high ice content thaws out, the ice melts and the ground subsides. On the other hand, permafrost containing no ice does not experience subsidence. Consequently, permafrost thaw can cause differential subsidence in ice-wedge polygon networks. This re-arranges the surface microtopography: ice wedges melt, the rims collapse into the troughs, and the polygons become flat-centred and then eventually high-centred (fig. 2b and c; Lara et al. (2015)). Wedge ice is ~20 % of the uppermost permafrost volume, and so this degradation could have a big impact on the shape of Arctic landscapes.

Are ice wedge polygons climate amplifiers?

Fig. 2: Schematic diagrams of polygon types and features [Credit: Wainwright et al. (2015)].

The transition from low-centred to high-centred ice-wedge polygons affects water distribution across the polygonal ground. The rims of low-centred polygons tend to block water drainage, whereas the troughs facilitate relatively fast and effective drainage of water from the polygonal networks (Liljedahl et al., 2012). So, during summer, the centres of low-centred polygons are frequently flooded with stagnant water, whereas the central mounds of high centred polygons are well drained (and good to sit on at lunchtime!). The contrast in hydrology influences vegetation, surface energy transfer, and biogeochemistry, in turn influencing carbon cycling and the release of greenhouse gases into the atmosphere.

High-centred polygons can have increased carbon dioxide emissions compared to low-centred polygons, on account of their lower soil moisture, reduced cover of green vascular vegetation and the well-drained soil (Wainwright et al., 2015). On the other hand, once plant growth during peak growing season is accounted for, this can actually cause a net drawdown of carbon dioxide in high-centred polygons (Lara et al., 2015). In contrast, there is general agreement that low-centred polygons are associated with high summer methane flux (Lara et al., 2015; Sachs et al., 2010; Wainwright et al., 2015). This is due to multiple interacting environmental factors. Firstly, low centred polygons have a higher temperature, which increases methane production rates. Secondly, they also have moister soil, which decreases the consumption of methane, owing to the lower oxygen availability. Thirdly, the low-centred polygons often have more vascular plants that help transport the methane away from its production site and up into the atmosphere. Lastly, the low-centred polygons had higher concentrations of aqueous total organic carbon, which provides a good food source for methanogens.

Outlook

As the climate warms, ice wedge polygons will increasingly degrade. The challenge now is to figure out whether the transition from low-centred to high-centred polygons will enhance or mitigate climate warming. This depends on the balance between the uptake and release of methane and carbon dioxide, as well as the rate of transition from high- to low-centred polygons.

Further Reading

Lara, M.J., et al. (2015), Polygonal tundra geomorphological change in response to warming alters future CO2 and CH4 flux on the Barrow Peninsula. Global Change Biology, 21(4), 1634-1651

Liljedahl, A.K., et al. (2016), Pan-Arctic ice-wedge degradation in warming permafrost and its influence on tundra hydrology. Nature Geoscience, 9, 312-316.

Wainwright, H.M., et al. (2015), Identifying multiscale zonation and assessing the relative importance of polygons geomorphology on carbon fluxes in an Arctic tundra ecosystem. Journal of Geophysical Research: Biogeosciences, 707-723.

On permafrost instability: Image of the Week – When the dirty cryosphere destabilizes! | EGU Cryosphere Blog

On polygons in wetlands: Polygon ponds at sunset | Geolog

Edited by Joe Cook and Sophie Berger


Eleanor Jones is a NERC PhD student on the EU-JPI LowPerm project based at the University of Sheffield and the University Centre in Svalbard. She is investigating the biogeochemistry of ice-wedge polygon wetlands in Svalbard. She tweets as @ElouJones. Contact Email: eljones3@sheffield.ac.uk

Image of the week — Making pancakes

A drifting SWIFT buoy surrounded by new pancake floes. [Credit: Maddie Smith]

It’s pitch black and twenty degrees below zero; so cold that the hairs in your nose freeze. The Arctic Ocean in autumn and winter is inhospitable for both humans and most scientific equipment. This means there are very few close-up observations of sea ice made during these times.

Recently, rapidly declining coverage of sea ice in the Arctic Ocean due to warming climate and the impending likelihood of an ‘ice-free Arctic’ have increased research and interest in the polar regions. But despite the warming trends, every autumn and winter the polar oceans still get cold, dark, and icy. If we want to truly understand how sea ice cover is evolving now and into the future, we need to better understand how it is growing as well as how it is melting.


Nilas or thin sheets of sea ice [Credit: Brocken Inaglory (distributed via Wikimedia Commons) ]

Sea ice formation

Sea ice formation during the autumn and winter is complex. Interactions between ocean waves and sea ice cover determine how far waves penetrate into the ice, and how the sea ice forms in the first place. If the ocean is still, sea ice forms as large, thin sheets called ‘nilas’. If there are waves on the ocean surface, sea ice forms as ‘pancake’ floes – small circular pieces of ice. As the Arctic transitions to a seasonally ice-free state, there are larger and larger areas of open water (fetch) over which ocean surface waves can travel and gain intensity. Over time, with the continued action of waves in the ice, pancake ice floes develop raised edges —  as seen in our image of the week — from repeatedly bumping into each other. Pancake ice is becoming more common in the Arctic, and it is already very common in the Antarctic, where almost all of the sea ice grows and melts every year.

Nilas vs pancakes

Nilas and pancake sea ice are different at the crystal level (see previous post), and regions of pancake ice and nilas of the same age may have different average ice thickness and ice concentration. As a result, the interaction of the ocean and atmosphere in these two ice types may be very different. Gaps of open water between pancake ice floes allow heat fluxes to be exchanged between the ocean and atmosphere – which can have very different temperatures during winter. Nilas and pancakes also interact with waves differently – nilas might simply flex with a low-intensity wave field, or break into pieces if disturbed by large waves, while pancakes bob around in waves, causing a viscous damping of the wave field. The two ice types have very different floe sizes (see previous posts here and here). Nilas is by definition is a large, uniform sheet of ice; pancake floes are initially very small and grow laterally as more frazil crystals in the ocean adhere to their sides, and multiple floes weld together into sheets of cemented pancakes.

How to make observations?

Sea ice models have only recently begun to be able to separate different sizes of sea ice. This allows more accurate inclusion of growth and melt processes that occur with the different sea ice types. However, observations of how sea ice floe size changes during freeze-up are required to inform these new models, and these observations have never been made before. Pancake sea ice floes are often around only 10 cm in diameter initially, which is far too small to observe by satellite. This means that observations of pancake growth need to be made close-up, but the dynamic ocean conditions in which pancakes are created makes it difficult to deploy instruments in-situ. So how can we observe pancake sea ice in this challenging environment?

In a recent paper (Roach et al, 2018), we used drifting wave buoys, called SWIFTs, to capture the growth of sea ice floes in the Arctic Ocean. SWIFTs are unique platforms (see image of the week) which drift in step with sea ice floes, recording air temperature, water temperature, ocean wave data and – crucially for sea ice – images of the surrounding ice. Analysis of the series of images captured has provided the first-ever measurements of pancake freezing processes in the field, giving unique insight into how pancake floes evolve over time as a result of wave and freezing conditions. This dataset has been compared with theoretical predictions to help inform the next generation of sea ice models. The new models will allow researchers to investigate whether describing physical processes that occur on the scale of centimetres is important for prediction of the polar climate system.

Edited by Sophie Berger


Lettie Roach is a PhD student at Victoria University of Wellington and the National Institute for Water and Atmospheric Research in New Zealand. Her project is on the representation of sea ice in large-scale models, including model development, model-observation comparisons and observation of small-scale sea ice processes.  

 

 

 

Maddie Smith is a PhD student at the Applied Physics Lab at the University of Washington in Seattle, United States. She uses observations to improve understanding of air-sea interactions in polar, ice-covered oceans.

Image of the Week — Seasonal and regional considerations for Arctic sea ice changes

Monthly trends in sea ice extent for the Northern Hemisphere’s regional seas, 1979–2016. [Credit: adapted from Onarheim et al (2018), Fig. 7]

The Arctic sea ice is disappearing. There is no debate anymore. The problem is, we have so far been unable to model this disappearance correctly. And without correct simulations, we cannot project when the Arctic will become ice free. In this blog post, we explain why we want to know this in the first place, and present a fresh early-online release paper by Ingrid Onarheim and colleagues in Bergen, Norway, which highlights (one of) the reason(s) why our modelling attempts have failed so far… 


Why do we want to know when the Arctic will become ice free anyway? 

As we already mentioned on this blog, whether you see the disappearance of the Arctic sea ice as an opportunity or a catastrophe honestly depends on your scientific and economic interests.  

It is an opportunity because the Arctic Ocean will finally be accessible to, for example: 

  • tourism; 
  • fisheries; 
  • fast and safe transport of goods between Europe and Asia; 
  • scientific exploration. 

All those activities would no longer need to rely on heavy ice breakers, hence becoming more economically viable. In fact, the Arctic industry has already started: in summer 2016, the 1700-passenger Crystal Serenity became the first large cruise ship to safely navigate the North-West passage, from Alaska to New York. Then in summer 2017, the Christophe de Margerie became the first tanker to sail through the North-East passage, carrying liquefied gas from Norway to South Korea without an ice breaker escort, while the Eduard Toll became the first tanker to do so in winter just two months ago. 

On the other hand, the disappearance of the Arctic sea ice could be catastrophic as having more ships in the area increases the risk of an accident. But not only. The loss of Arctic sea ice has societal and ecological impacts, causing coastal erosion, disappearance of a traditional way of life, and threatening the whole Arctic food chain that we do not fully understand yet. Not to mention all of the risks on the other components of the climate system. (See our list of further readings at the end of this post for excellent reviews on this topic). 

Either way, we need to plan for the disappearance of the sea ice, and hence need to know when it will disappear. 

Arctic sea ice decrease varies with region and season 

In a nutshell, the new paper published by Onarheim and colleagues says that talking about “the Arctic sea ice extent” is an over simplification. They instead separated the Arctic into its 13 distinct basins, and calculated the trends in sea ice extent for each basin and each month of the year. They found a totally different behaviour between the peripheral seas (in blue on this image of the week) and the Arctic proper, i.e. north of Fram and Bering Straits (in red). As is shown by all the little boxes on the image, the peripheral seas have experienced their largest long term sea ice loss in winter, whereas those in the Arctic proper have been losing their ice in summer only. In practice, what is happening to the Arctic proper is that the melt season starts earlier (note how the distribution is not symmetric, with largest values on the top half of the image).  

Talking about Arctic sea ice extent is an over simplification

Moreover, Onarheim and colleagues performed a simple linear extrapolation of the observed trends shown on this image, and found that the Arctic proper may become ice-free in summer from the 2020s. As they point out, some seas of the Arctic proper have in fact already been ice free in recent summers. The trends are less strong in the peripheral seas, and the authors write that they will probably have sea ice in winter until at least the 2050s. 

So, although Arctic navigation should become possible fairly soon, in summer, you may need to choose a different holiday destination for the next 30 winters. 

Melting summer ice. [Credit: Mikhail Varentsov (distributed via imaggeo.egu.eu)]

But why should WE consider the regions separately? 

The same way that you would not plan for the risk of winter flood in New York based on yearly average of the whole US, you should not base your plan for winter navigation from Arkhangelsk to South Korea on the yearly Arctic-wide average of sea-ice behaviour. 

Scientifically, this paper is exciting because different trends at different locations and seasons will also have different consequences on the rest of the climate system. If you have less sea ice in autumn or winter, you will lose more heat from the ocean to the atmosphere, and hence impact both components’ heat and humidity budget. If you have less sea ice in spring, you may trigger an earlier algae bloom. 

As often, this paper highlights that the Earth system behaves in a more complex fashion that it first appears. Just like global warming does not prevent the occurrence of unpleasantly cold days, the disappearance of Arctic sea ice is not as simple as ice cubes melting in your beverage on a sunny day.  

Reference/Further reading

Bhatt, U. S., et al. (2014), Implications of Arctic sea ice decline for the Earth system. Ann. Rev. Environ. Res., 39, 57-89 

Meier, W. N., et al. (2014), Arctic sea ice in transformation: A review of recent observed changes and impacts on biology and human activity. Reviews of Geophysics, 52(3), 185-217. 

Onarheim, I., et al. (2018), Seasonal and regional manifestation of Arctic sea ice loss. Journal of Climate, EOR.  

Post, E., et al. (2013), Ecological consequences of sea-ice decline. Science, 341, 519-524 

Edited by Sophie Berger

Image of the Week – Geothermal heat flux in Antarctica: do we really know anything?

Spatial distributions of geothermal heat flux: (A) Pollard et al. (2005) constant values, (B) Shapiro and Ritzwoller (2004): seismic model, (C) Fox Maule et al. (2005): magnetic measurements, (D) Purucker (2013): magnetic measurements, (E) An et al. (2015): seismic model and (F) Martos et al. (2017): high resolution magnetic measurements. The color scale is truncated at 30 and 80 mW m-2. The black line locates the grounding line. Note, (B)-(F) are in order of publication from oldest to most recent. [Credit: Brice Van Liefferinge, (2018), PhD thesis]

Geothermal heat flux is the major unknown when we evaluate the temperature and the presence/absence of water at the bed of the Antarctic Ice Sheet. This information is crucial for the Beyond Epica Oldest Ice project, which aims to find a continuous ice core spanning 1.5 million years (see this previous post). A lot of work has been done* to determine geothermal heat flux under the entire Antarctic Ice Sheet, and all conclude that additional direct measurements are necessary to refine basal conditions! However direct measurements are difficult to obtain, due to the thick layer of ice that covers the bedrock. Our new image of the week goes over what we currently know about the geothermal heat flux in Antarctica and presents the five data sets that currently exist. But first, let’s see where this heat flux come from?


What determines geothermal heat flux and how can we estimate it?

Heat flux measured at the surface of the Earth has two sources: (i) primordial heat remaining from when the Earth formed and (ii) contemporary-sourced heat coming from radioactive isotopes present in the mantle and the crust. This heat, concentrated in the Earth’s centre, can propagate to the surface through both conduction in the solid earth (inner core and crust) and convection in the liquid-viscous earth (outer core, lower and upper mantles). The net heat flux to reach the surface of the crust and penetrate the overlying ice is what we refer to as the ‘geothermal heat flux’. Wherever the crust is thinner, convection in the mantle can transfer heat more efficiently to the surface. In those locations, the net geothermal heat flux is higher, and vice versa. At mid-ocean ridges and in active volcanic areas, the heat can be delivered almost directly to the surface by advection (i.e. by the movement of magma), therefore leading to a higher net surface geothermal heat flux (think of Iceland, where the shallow crust allows them to take advantage of geothermal heat flux directly).

As a result, we know that the geology determines the magnitude of the geothermal heat flux and the geology is not homogeneous underneath the Antarctic Ice Sheet:  West Antarctica and East Antarctica are significantly distinct in their crustal rock formation processes and ages.

Nowadays, five independent global geothermal heat flux data sets exist: Shapiro and Ritzwoller, (2004); Fox Maule et al., (2005); Purucker, (2013); An et al., (2015); Martos et al., (2017) (see image of the week). All geothermal heat flux data sets compiled and currently used have been inferred from the properties of the crust and the upper mantle, as geology dictates the magnitude of geothermal heat flux spatially. Let’s see together how the estimation of geothermal heat flux has evolved over the years….

Using constant values (Panel A)

The simplest method, which consists in using a constant value of geothermal heat flux over the entire continent, was common at first and is still sometimes used (e.g. sensitivity tests and model intercomparison projects) as it facilitates model inter-comparisons. Pollard et al. (2005), in panel A, used bands of constant geothermal heat flux values (70, 60, 55 and 41 mW m-2), with geothermal heat flux decreasing from West Antarctica to East Antarctica, consistent with the known geology.

2004, a seismic model (Panel B)

Shapiro and Ritzwoller (2004) are the first to propose a geothermal heat flux distribution map based on seismic methods, and not strictly on rock composition. They extrapolate the geothermal heat flux from a global seismic model of the crust and the upper mantle which is an analysis of seismicity all over the world. Regions of the globe are grouped by their similarity in seismic structure. Assuming that a certain magnitude of seismicity represents a certain geothermal heat flux value, they assign geothermal heat flux value to regions where geothermal heat flux cannot be directly measured by using geothermal heat flux data from regions of similar seismicity. The geothermal heat flux spatial distribution obtained, with values up to 80 mW m-2 in West Antarctica and 48 mW m-2 in East Antarctica, agrees with that of Pollard et al. (2005). However, errors associated with this method are quite large, reaching 50% of the geothermal heat flux value.

 

2005, magnetic measurements (Panel C)

A year later, Fox Maule et al. (2005) derive a geothermal heat flux map based on satellite magnetic measurements and a thermal model. The objective is to determine the depth to the Curie temperature, the temperature at which a material loses its permanent magnetic properties. They set the Curie temperature to 580 °C, while the temperature at the ice-bedrock interface is set at 0 °C. Satellite magnetic measurements allow the calculation of the depth of each of these boundaries. The geothermal heat flux is then obtained using a thermal model of the crust between the depth of the two boundary temperatures. This method also has a large associated error, 60% of the geothermal heat flux value for the East Antarctic interior.

2013, reanalysis of magnetic measurements (Panel D)

In 2013, Purucker updates the Fox Maule et al. (2005) geothermal heat flux map with new magnetic data. The spatial geothermal heat flux pattern obtained still retains the characteristic pattern of low values in West Antarctica and high values in East Antarctica, but predicts lower absolute values for East Antarctica and around the West Antarctic coast.

2015, new seismic model (Panel E)

More recently, An et al. (2015) derive a new geothermal heat flux distribution based on seismic velocities. The method is similar to that used by Shapiro and Ritzwoller (2004). They analyse the Earth’s mantle properties using a new 3D crustal shear velocity model to calculate crustal temperatures and the surface geothermal heat flux. However, their spatial distribution of geothermal heat flux differs quite a bit from the other data sets, particularly in East Antarctica where geothermal heat flux values differ by 10 mW m-2 from those of Shapiro and Ritzwoller (2004). An et al. (2015) find very low geothermal heat flux values at the domes, which is good news for the search of Oldest Ice, but rather high overall values for East Antarctica compared to the other data sets. They explain that the model is invalid for geothermal heat flux values exceeding 90 mW m-2. But such high values should only impact young crust areas, mainly West Antarctica and therefore the variability observed in East Antarctica cannot be explained.

2017, high resolution magnetic measurements (Panel F)

In 2017, Martos et al. provide a high resolution geothermal heat flux map based on the spectral analysis of airborne magnetic data. They use a compilation of all existing airborne magnetic data to determine the depth to the Curie temperature and infer the geothermal heat flux using a thermal model. Their continent-wide spatial distribution of geothermal heat flux obtained agrees with previous studies, but they show higher overall magnitudes of geothermal heat flux including East Antarctica. They report an error of 10 mW m-2 which is interestingly smaller than for the other data sets. However, their data set does not take into account point measurements of geothermal heat flux. The same year, Goodge (2017) calculates an average geothermal heat flux value of 48 mW m-2 for East Antarctica with a standard deviation of 13.6 mW m from the analysis of clasts in the region between Dome A and the Ross Sea. A geothermal heat flux value of 48 mW m-2 is consistent with the mean value of the data sets described above.

All in all

To sum up, although all geothermal heat flux data sets agree on continent scales (with higher values under the West Antarctic ice sheet and lower values under East Antarctica), there is a lot of variability in the predicted geothermal heat flux from one data set to the next on smaller scales. A lot of work remains to be done …

* (e.g. Shapiro and Ritzwoller, 2004; Fox Maule et al., 2005; Purucker, 2013; An et al., 2015; Fisher et al., 2015; Parrenin et al., 2017; Seroussi et al., 2017; Martos et al., 2017; Goodge, 2017)

References

Van Liefferinge, B., Pattyn, F., Cavitte, M. G. P., Karlsson, N. B., Young, D. A., Sutter, J., and Eisen, O.: Promising Oldest Ice sites in East Antarctica based on thermodynamical modelling, The Cryosphere Discuss., https://doi.org/10.5194/tc-2017-276, in review, 2018.

Van Liefferinge, B. Thermal state uncertainty assessment of glaciers and ice sheets: Detecting promising Oldest Ice sites in Antarctica, PhD thesis, Université libre de Bruxelles, Brussels, 2018.

Edited by Sophie Berger


Brice Van Liefferinge  has just earned his PhD at the Laboratoire de Glaciology, Universite Libre de Bruxelles, Belgium. His research focuses on the basal conditions of the ice sheets. He tweets as @bvlieffe.

Image of the Week – The world in a grain of cryoconite

Fig 1: A single grain of cryoconite (top left) is home to a microscopic city of microbes, revealed here by chlorophyll fluorescence microscopy – a technique that causes photosynthesising microbes to emit light (top right) and portable DNA sequencing (bottom panel) [credit: Arwyn Edwards]

Microbes growing on glaciers are recognized for their importance in accelerating glacier melting by darkening their surface and for maintaining biogeochemical cycles in Earth’s largest freshwater ecosystem. However, the microbial biodiversity of glaciers remains mysterious. Today, new DNA sequencing techniques are helping to reveal glaciers as icy hotspots of biodiversity.


To see a world in a grain of…cryoconite

Earth’s glaciers and ice sheets are among its most impressive features, yet this majesty conceals their microscopic riches. We must turn to the microscope and the DNA sequencer to reveal the natural history of glaciers. Rather than a grain of sand, this world lies hidden in a grain of cryoconite. Cryoconite ecosystems are microbe-mineral aggregates which darken the surface of glaciers world-wide which – along with algae – enhance absorption of solar energy and promote glacier melting through so-called bioalbedo feedbacks. Microscopy studies from the late 19th and early 20th century reveal that a diverse range of algae, cyanobacteria, heterotrophic bacteria, protists, fungi and even tardigrades live within cryoconite, but it is only in the last decade that we have started to resolve the genetic diversity of life within cryoconite.

From glaciers to genomes…and back again

Considering glaciers are Earth’s largest freshwater ecosystems, we know very little about the genetic diversity of their inhabitants. Of all known glaciers, fewer than 0.05% have any form of DNA datasets associated with them. Such DNA datasets are commonplace for other environments, as demonstrated by the Earth Microbiome Project. From the limited studies performed, it appears the microbial ecosystems of glaciers are no less diverse than temperate environments: even dark, cold and isolated subglacial lakes harbour thousands of bacterial species. As climatic warming increasingly threatens glaciers, unpicking the interactions between microbes and melt is vital, as is establishing the extent to which glacier biodiversity is threatened. Sequencing microbial genomes from glacial ecosystems is therefore urgent.

Fig 2: Preserving microbial samples from cryoconite for return to the author’s home lab is a conventional approach to studying genetic diversity on glaciers, but the portability of MinION DNA sequencing brings the lab to the field [Credit: Arwyn Edwards].

A DNA sequencer in your rucksack

Such genetic studies have required the collection of samples from glaciers and their return to state-of-the-art laboratories equipped with high throughput DNA sequencers. However, new, portable DNA sequencers are being trialled on cryoconite to permit sequencing of DNA in field labs. Using a pocket-sized DNA sequencer called a MinION connected to the USB port of a laptop, it is possible to extract, sequence and analyse microbial genomes while still in the field. While Nanopore DNA sequencing using MinION devices are increasingly applied to medical emergencies such as Ebola or antibiotic resistance, their highly portable nature means that glacier scientists will be able to collect and analyse microbial genomes while in the field, making the genetic diversity of glaciers accessible.

Who’s who in cryoconite?

Using MinION for DNA sequencing in a field lab at the UK Arctic Station in Ny Ålesund, it was possible to generate rapid profiles of microbial diversity in cryoconite. So who lives in Arctic cryoconite? The most abundant bacterial group identified is a close match to Phormidesmis priestleyi, a filamentous cyanobacterium responsible for engineering the growth of cryoconite grains on Arctic glaciers. In Figure 1 above, Phormidesmis is visible as the bright red, chlorophyll-rich filaments binding together the cryoconite grain. Other cyanobacteria are present, including a species matching sequences from Phormidium autumnale found in an Antarctic lake. However, MinION sequencing is useful in revealing less charismatic microbes. Also abundant within the community are members of the Polaromonas genus. Found in both cold and highly polluted environments worldwide, Polaromonas bacteria are highly flexible in their lifestyles, able to adapt to using highly poisonous compounds as food sources, or even anoxygenic phototrophy (photosynthesis without using water or producing oxygen) on alpine glaciers. Cryoconite sequences matching DNA from Methylibium found in Tibetan permafrost also hint at the need for flexible metabolism to survive on glacier surfaces. Finally, Ferruginibacter sequences best matching DNA data from iron-rich dust aggregates forming on snow in the Japanese mountains suggest that cold-tolerant iron cycling may be occurring within cryoconite.

In a grain of cryoconite, we see relatives of cyanobacteria from Arctic glaciers, but also Antarctic lakes, metabolically flexible bacteria found in cold and contaminated environments, and even bacteria living by respiring iron on snow in the Japanese mountains. We see a world.

Edited by Joe Cook and Sophie Berger


Dr Arwyn Edwards is a Senior Lecturer in Biology at Aberystwyth University and the present Royal Geographic Society’s Walters Kundert Arctic Fellow. His research on genomic diversity in glacial environments is supported by the Leverhulme Trust.