CR
Cryospheric Sciences

numerical modelling

Image of the Week – A high-resolution picture of Greenland’s surface mass balance

Image of the Week – A high-resolution picture of Greenland’s surface mass balance

The Greenland ice sheet – the world’s second largest ice mass – stores about one tenth of the Earth’s freshwater. If totally melted, this would rise global sea level by 7.4 m, affecting low-lying regions worldwide. Since the 1990s, the warmer atmosphere and ocean have increased the melt at the surface of the Greenland ice sheet, accelerating the ice loss through increased runoff of meltwater and iceberg discharge in the ocean.


Simulating the climate with a regional model

To understand the causes of the recent ice loss acceleration in Greenland, we use the Regional Atmospheric Climate Model RACMO2.3 (Noël et al. 2015) that simulates the evolution of the surface mass balance, that is the difference between mass gain from snowfall and mass loss from sublimation, drifting snow erosion and meltwater runoff. Using this data set, we identify three different regions on the ice sheet (Fig. 1):

  • the inland accumulation zone (blue) where Greenland gains mass at the surface as snowfall exceeds sublimation and runoff,

  • the ablation zone (red) at the ice sheet margins which loses mass as meltwater runoff exceeds snowfall.

  • the equilibrium line (white) that separates these two areas.

From 11 km to 1 km : downscaling RACMO2.3

To cover large areas while overcoming time-consuming computations, RACMO2.3 is run at a relatively coarse horizontal resolution of 11 km for the period 1958-2015. At this resolution, the model does not resolve small glaciated bodies (Fig. 2a), such as narrow marginal glaciers (few km wide) and small peripheral ice caps (ice masses detached from the big ice sheet). Yet, these areas contribute significantly to ongoing sea-level rise. To solve this, we developed a downscaling algorithm (Noël et al., 2016) that reprojects the original RACMO2.3 output on a 1 km ice mask and topography derived from the Greenland Ice Mapping Project (GIMP) digital elevation model (Howat et al., 2014). The downscaled product accurately reproduces the large mass loss rates in narrow ablation zones, marginal outlet glaciers, and peripheral ice caps (Fig. 2b).

Fig. 2: Surface mass balance (SMB) of central east Greenland a) modelled by RACMO2.3 at 11 km, b) downscaled to 1 km (1958-2015). The 1 km product (b) resolves the large mass loss rates over marginal outlet glaciers [Credit: Brice Noël].

 

The high-resolution data set has been successfully evaluated using in situ measurements and independent satellite records derived from ICESat/CryoSat-2 (Noël et al., 2016, 2017). Recently, the downscaling method has also been applied to the Canadian Arctic Archipelago, for which a similar product is now also available on request.

Endangered peripheral ice caps

Using the new 1 km data set (Fig. 1), we identified 1997 as a tipping point for the mass balance of Greenland’s peripheral ice caps (Noël et al., 2017). Before 1997, ablation (red) and accumulation zones (blue) were in approximate balance, and the ice caps remained stable (Fig. 3a). After 1997, the accumulation zone retreated to the highest sectors of the ice caps and the mass loss accelerated (Fig. 3b). This mass loss acceleration was already reported by ICESat/CryoSat-2 satellite measurements, but no clear explanation was provided. The 1 km surface mass balance provides a valuable tool to identify the processes that triggered this recent mass loss acceleration.

Fig. 3: Surface mass balance of Hans Tausen ice cap and surrounding small ice bodies in northern Greenland before (a) and after the tipping point in 1997 (b). Since 1997, the accumulation zone (blue) has shrunk and the ablation zone (red) has grown further inland, tripling the pre-1997 mass loss [Credit: Brice Noël].

 

Greenland ice caps are located in relatively dry regions where summer melt (ME) nominally exceeds winter snowfall (PR). To sustain the ice caps, refreezing of meltwater (RF) in the snow is therefore a key process. The snow acts as a “sponge” that buffers a large amount of meltwater which refreezes in winter. The remaining meltwater runs off to the ocean (RU) and contributes to mass loss (Fig. 4a).

Before 1997, the snow in the interior of these ice caps could compensate for additional melt by refreezing more meltwater. In 1997, following decades of increased melt, the snow became saturated with refrozen meltwater, so that any additional summer melt was forced to run off to the ocean (Fig. 4b), tripling the mass loss.

Fig. 4: Surface processes on an ice cap: the ice cap gains mass from precipitation (PR), in the form of rain and snow. a) In healthy conditions (e.g. before 1997), meltwater (ME) is partially refrozen (RF) inside the snow layer and the remainder runs off (RU) to the ocean. The mass of the ice cap is constant when the amount of precipitation equals the amount of meltwater that runs off. b) When the firn layer is saturated with refrozen meltwater, additional meltwater can no longer be refrozen, causing all meltwater to run off to the ocean. In this case, the ice cap loses mass, because the amount of precipitation is smaller than the amount of meltwater that runs off [Credit: Brice Noël].

  In 1997, following decades of increased melt, the snow became saturated with refrozen meltwater, so that any additional summer melt was forced to run off to the ocean, tripling the mass loss.

We call this a “tipping point” as it would take decades to regrow a new, healthy snow layer over these ice caps that could buffer enough summer meltwater again. In a warmer climate, rainfall will increase at the expense of snowfall, further hampering the formation of a new snow cover. In the absence of refreezing, these ice caps will undergo irreversible mass loss.

What about the Greenland ice sheet?

For now, the big Greenland ice sheet is still safe as snow in the extensive inland accumulation zone still buffers most of the summer melt (Fig. 1). At the current rate of mass loss (~300 Gt per year), it would still take 10,000 years to melt the ice sheet completely (van den Broeke et al., 2016). However, the tipping point reached for the peripheral ice caps must be regarded as an alarm-signal for the Greenland ice sheet in the near future, if temperatures continue to increase.

Data availability

The daily, 1 km Surface Mass Balance product (1958-2015) is available on request without conditions for the Greenland ice sheet, the peripheral ice caps and the Canadian Arctic Archipelago.

Further reading

Edited by Sophie Berger


Brice Noël is a PhD Student at IMAU (Institute for Marine and Atmospheric Research at Utrecht University), Netherlands. He simulates the climate of the Arctic region, including the ice masses of Greenland, Svalbard, Iceland and the Canadian Arctic, using the regional climate model RACMO2. His main focus is to identify snow/ice processes affecting the surface mass balance of these ice-covered regions. He tweets as: @BricepyNoel Contact Email: b.p.y.noel@uu.nl

Image of the Week – Supraglacial debris variations in space and time!

Image of the Week – Supraglacial debris variations in space and time!

There is still a huge amount we don’t know about how glaciers respond to climate change. One of the most challenging areas is determining the response of debris-covered glaciers. Previously, we have reported on a number of fieldwork expeditions to debris-covered glaciers but with this Image of The Week we want to show you another way to investigate these complex glaciers – numerical modelling!


Debris-covered glaciers

Debris-covered glaciers occur globally, with a great many being found in the Himalaya-Karakoram mountain range. For example, in the Everest Region of Nepal 33% of glacier area is debris covered (Thakuri et al., 2014). The response of debris-covered glaciers to future climate change in such regions has huge implications for water resources, with one fifth of the world’s population relying on water from the Himalayan region for their survival (Immerzeel et al., 2010).

Debris-covered glaciers respond to climate change differently to debris-free glaciers as the supraglacial debris layer acts as a barrier between the atmosphere and glacier (Reznichenko et al., 2010). The supraglacial debris layer has several key influences on the glacier dynamics:

  • Glacier ablation (loss of mass from the ice surface) is enhanced or inhibited depending on debris layer thickness and properties – see our previous post.
  • Supraglacial debris causes glaciers to reduce in volume through surface lowering rather than terminus retreat (typical of debris free mountain glaciers).

Understanding the influence of a supraglacial debris layer on mass loss or gain is, therefore, key in determining the future of these glaciers. The properties of supraglacial debris layers can vary in time and space both in debris layer thickness and distribution, as well as properties of the rocks which make up the debris (e.g. albedo, surface roughness, porosity, size and moisture content). It is these characteristics of the debris-cover which control the heat transfer through the debris and therefore the amount of thermal energy that reaches the underlying ice causing melting (Nicholson and Benn, 2006). In order to better predict the future of debris-covered glaciers we needs to be able to numerically model their behaviour. This means we need a better understanding of the variations in debris cover and how this affects the ice dynamics.

How does a supraglacial debris layer vary in time and space?

Our Image of the Week (Fig. 1) shows a schematic of how debris distribution can vary spatially across a glacier surface and also this can change through time. The main inputs of debris are:

  • Upper regions: snow and ice avalanches in the upper reaches of the glacier.
  • Mid and Lower regions: rock avalanches and rock falls (Mihalcea et al., 2006).

These irregular mass movement events vary in frequency and magnitude, and therefore affect debris distribution across the glacier surface but also through time. The irregularity of them makes it really hard to predict and simulate! Luckily, debris transport is a little more predictable.

Figure 2: An ice cliff emerging out of the supraglacial debris layer on Khumbu Glacier, Nepal, with Nuptse in the background. [Credit: M. Gibson]

Debris is initially transported along medial moraines (glacially transported debris)  in the upper and mid-sections of the glacier, this is known as entrained debris. The various sources of entrained debris combine to form a continuous debris cover in the lower reaches of the glacier (Fig. 1). As a supraglacial debris layer is forming, such as for Baltoro glacier (Fig. 1), the boundary between the continuous debris layer and entrained debris sections progresses further upglacier over time.

Eventually transported debris will reach the terminus of the glacier and be deposited (Fig. 1), mainly due to a decrease in surface velocity of the glacier towards the terminus. However, once debris is deposited it doesn’t just sit there; debris is constantly being shifted around as ablation (surface melting) occurs. As ablation occurs the debris surface ablates unevenly, as the thickness of the debris layer is spatially variable. Uneven ablation, otherwise known as differential surface lowering, causes the glacier surface to be made up of topographic highs and lows, the latter of which sometimes become filled with water, forming supraglacial ponds (Fig. 1) . Another product of debris shifting is that ice cliffs, such as the one seen in Fig. 2, are exposed. These features are initially formed when englacial channels collapse  or debris layers slide (Kirkbride, 1993). All this movement and shifting means that not only do glacier models have to consider variation in debris layers across the glacier and through time, but also the presence of ice cliffs and supraglacial ponds. They are important as they have a very different surface energy balance to debris-covered ice. To complicate things further the frequency and area of ice cliffs and supraglacial ponds also vary through time! You see the complexity of the problem…

Modelling spatially and temporally varying debris layers

Numerical modelling is key to understanding how supraglacial debris layers affect glacier mass balance. However, current numerical modelling often either omits the presence of a supraglacial debris layer entirely, or a debris layer that is static in time and/or space (e.g. Collier et al., 2013; Rowan et al., 2015; Shea et al., 2014). However, as outlined earlier, these supraglacial debris layers are not static in time or space. Understanding the extent to which spatiotemporal variations in supraglacial debris distribution occur could aid identification of when glaciers became debris-covered, glaciers that will become debris-covered glaciers in the future, and the timescales over which supraglacial debris layers vary. The latter is particularly relevant to numerical modelling as it would result in total glacier ablation being calculated more precisely throughout the modelling time period. Understanding the interaction between glacier dynamics and debris distribution is therefore key to reconstructing debris-covered glacier systems as accurately as possible.

Edited by Emma Smith


Morgan Gibson is a PhD student at Aberystwyth University, UK, and is researching the role of supraglacial debris in ablation of Himalaya-Karakoram debris-covered glaciers. Morgan’s work focuses on: the extent to which supraglacial debris properties vary spatially; how glacier dynamics control supraglacial debris distribution; and the importance of spatial and temporal variations in debris properties on ablation of Himalaya-Karakoram debris-covered glaciers. Morgan tweets at @morgan_gibson, contact email address: mog2@aber.ac.uk.