Whither Greenland Goeth
by Christina Hulbe
Greenland's outlet glaciers are restless. They speed up, they slow down, they thin, and they thicken, on time scales of years to decades. Over the full arc of the modern observational era (starting in the mid 1800's), the pattern is one of gradual retreat. This is the same pattern we observe for glaciers across the globe. Within that broad context though, from the 1950's forward, most Greenland outlet glaciers maintained relatively stable front positions and experienced relatively small changes in volume. Sometime in the late 1990's, that situation changed. Outlet glaciers around much of the continent are speeding up, discharging more ice into the ocean than in decades past, and surface melting has increased at lower elevations. Together, these two changes are causing the ice sheet to lose a total volume of somewhere between 124 and 224 cubic kilometers/year (two estimates made using different methods), equivalent to between 0.34 and 0.57 millimeters/year of sea level rise.
What's causing the change now underway in Greenland? Enhanced melting is a result of regional warming. The cause for the glacier speed-ups may also be related to that warming but the connection is not direct and not completely understood. Here, I'll discuss some basic ideas regarding ice sheet mass balance and some possible causes for the recent, dramatic change in Greenland outlet glacier flow.
the Greenland Ice Sheet
Greenland has been extensively glaciated for about the last 3.3 million years and has experienced regional glaciation for the last 30 million years or so. Over that time the ice sheet has waxed and waned, in step with global glacial-to-interglacial cycles. The island's 2.17 million square kilometer area, a little more than a fifth the size of the continental US, is at present about 80% ice covered. Were all of that ice to melt it would contribute between 6 and 7 meters to global sea level.
The Greenland Ice Sheet looks something like a big, oddly shaped pillow. The ice sheet is thick and relatively flat in the interior and becomes relatively steep only near the coasts (here are some maps I made using data from Bamber et al., 2000, JGR). Underneath the ice, the bedrock is relatively low in the interior and relatively high around the coast. Maximum ice thicknesses, in the interior of the sheet, are a bit more than 3,000 meters.
The speed at which glacier ice flows depends on its surface slope, thickness, temperature, and the nature of the bed beneath it. Simply, steeper surface slope, warmer temperature, and greater thickness all yield faster flow. In Greenland, ice in the interior of the ice sheet flows slowly, 10's of meters per year (the ice is thick but relatively cold and surface slopes are small). Ice moves away from the interior toward outlet glaciers that flow through coastal fjords, discharging ice into the sea. Outlet glacier speeds range from 100's to 1000's of meters per year. Two of the fastest outlet glaciers, Kangerdlugssuaq in the east and Jakobshavn in the west, were moving at about 12 kilometers per year at their downstream ends in 2005, discharging 27.8 and 23.6 cubic kilometers of ice per year, respectively, into the ocean. These are the speediest glaciers on the planet.
Ice sheets gain mass by snow accumulation and lose mass by melting (on both the upper and lower surfaces) and by iceberg calving where the ice flows into the ocean (as a calving glacier). The difference between accumulation and ablation (all forms of mass loss) is called the mass balance. When the volumes of ice accumulating and ablating are equal, the system is in balance. When the mass balance is positive, more accumulation than ablation, the glacier grows larger and makes a withdrawal from global sea level. When the mass balance is negative, glacier size decreases and a contribution is made to sea level.
surface accumulation & melt
Global warming is causing an increase in total snowfall over Greenland. This is the case because a warmer atmosphere can hold more moisture than can a colder atmosphere. If that extra snow is not lost to melting, the net effect is mass accumulation. This is just what we see in the high, cold interior of the ice sheet, while at lower elevations melting dominates (pdf of a technical paper).
Surface melting on the Greenland Ice Sheet varies from year to year, both in duration and geographic extent. Melt season (typically April to September) duration has increased by up to a month since the late 1980's. Over most of the observational record, the 2000 meter elevation contour (the heavy yellow line in the first map here) has served as a sort of "rule of thumb" boundary between the accumulation zone (above) and the ablation zone (below). The transition from one to the other can be seen from space in visible wavelengths as a change from a snowy white surface to blue and grey wet snow and bare glacier ice. Bright blue ponds of meltwater are often observed in the ablation zone. The transition from accumulation to ablation zone can also be observed at other wavelengths and such remotely-sensed data are used to establish the chronology of melt extent over the ice sheet.
The year to year variation in melt extent on the Greenland Ice Sheet is significant, though the late 1990's and early 2000's have seen relatively large melt areas. Surface melting has been observed at higher and higher atlitudes over time. A positive trend line may be fit to the melt extent data but the observational period is short (starting in 1979) and decadal-scale variability in this region is large (the Northern Annular Mode of atmospheric variability). That said, regional changes due to global warming have changed both snow accumulation and surface melt in Greenland (pdf of a technical paper).
The other part of the mass balance equation is ice flow. An increase in glacier speed will yield negative mass balance if the increased volume flux results in losses by calving (and secondarily by melting) that exceed the volume of ice accumulated in the upstream catchment area. Glaciers and ice sheets can speed up and slow down on many time scales, for many reasons. We can't cover this topic in any detail here but a few broad concepts will help us to think about what's going on in Greenland.
Some simple online model explorations of glacier flow and change are presented in a website built by some of my colleagues here at Portland State. A climate-driven, spreadsheet-based model also intended for a wide audience was created by a friend at the University of Brussels (this model runs on your computer).
Glacier ice behaves as a viscous fluid, compelled to flow by gravity. Its deformation (flow) is a reflection of the balance of forces acting on the material. The gravitational force is the product of mass and the acceleration due to gravity. The magnitude of the resulting gravitational driving stress within the ice depends on the surface slope and thickness of the ice. The driving stress must be balanced by other stresses (stress is the product of a force and the area over which it is applied). Resistive stresses act at the base of the ice (at the ice/bedrock contact), along the the sides of outlet glaciers (at the ice/valley wall contact), and at the seaward fronts of tidewater glaciers (the ice/ocean water contact). The basal resistance is relatively large where the ice flows over a rigid substrate (such as the bedrock in the interior of Greenland) and relatively small where the ice flows over a soft substrate (such as ocean water under the floating downstream ends, called "ice tongues" of some outlet glaciers).
Liquid water at the ice/bed interface can increase the rate of ice flow by allowing the ice to slide over the rough bedrock surface. That water may come from melting at the ice/bed interface or may be transported there from the surface via crevasses and moulins. If the amount of water reaching the bed increases, glacier flow may speed up.
Changes in resistive stresses at the downstream end of an outlet glacier can also cause it to change speed. In Greenland, the large outlet glaciers are tidewater glaciers, a name given to glaciers that terminate in water. The mechanics of tidewater glaciers are controlled to a large extent by the shape of the sea floor beneath the ice. Shoals in the sea floor (often piles of glacier-transported sediment) act as "pinning points" that help the glacier maintain a stable terminus position. Side drag along valley walls also plays a role in the force balance. In the stable state, ice mass accumulating in the interior of the ice sheet flows down through the outlet glacier and eventually calves off in icebergs at the glacier front but the front position and glacier geometry remain about the same over time.
If a tidewater glacier loses contact with a downstream pinning point, the rate of ice discharge can increase, causing the glacier front to retreat and the glacier to go into a period of negative mass balance. Over time, the glacier will reach a new equilibrium state. Depending on what caused the perturbation, the glacier front may readvance to an equilibrium state similar to the prior one or it may retreat to a new (smaller) stable geometry. Alaskan tidewater glaciers are observed to both advance and retreat.
Outlet glaciers around much of Greenland have increased their discharge dramatically in recent years, from near balance to rapid drawdown. Friends of mine (linked site includes video of what the work is like) at the University of Maine who visited several southeastern Greenland outlet glaciers in 2005 couldn't land their helicopter on a site they had chosen, using satellite imagery, near the downstream end of Kangerdlugssuaq Glacier because the ice was gone. They were surprised (as is everybody I know) at how rapidly the ice is changing.
Greenland mass balance
There is more than one method by which changes in the mass of the ice sheet may be estimated. One approach is to compare the net volume of ice accumulation (new snow accumulation minus losses to melting) in a glacier's catchment region in a given year with the volume of ice flowing out to the sea. A second, recently developed, approach is the measure the changing gravitational attraction of the ice sheet via satellite. As the ice mass increases, so does its gravitational attraction, and visa versa. This measurement is made using NASA’s Gravity Recovery and Climate Experiment (GRACE) satellites.
Eric Rignot (at NASA's JPL) and Pannir Kanagaratnam (U. Kansas) conducted a catchment-by-catchment inventory of changes in Greenland ice thickness and outlet speed from 1996 to 2005. Thirty of the 33 catchment basins were experiencing net mass loss, and the losses were larger in the southern part of the ice sheet than in the north. The two glaciers noted earlier, Kangerdlugssuaq in the east and Jakobshavn in the west, lost more ice than they gained from year to year. Kangerdlugssuaq was out of balance by -5.2 cubic kilometers per year in 2000 and -35.8 cubic kilometers per year in 2005. Jakobshavn was out of balance by -12.5 and -16.0 cubic kilometers per year in 2000 and 2005, respectively. Across the whole ice sheet, the year 2000 net mass change was -138 +/- 31 cubic kilometers and the year 2005 change was -224 +/- 41 cubic kilometers of ice. The 2005 value is equivalent to 0.57 +/- 0.1 millimeters of sea level rise.
The gravity method has been used by two different groups, with slightly different outcomes. The results differ simply because one group took a higher-resolution view of the ice sheet than did the other. The more recent, higher resolution, calculation shows thickening in the northern interior of the ice sheet (as we would expect in a snowier, warming world) and thinning elsewhere. The largest thinning rates in the southeastern sector of the ice sheet. These authors, led by Scott Luthcke and Jay Zwally, both at NASA's Goddard Space Flight Center, calculate a net mass loss of 124 +/- 19 cubic kilometers of ice per year from 2003 to 2005, equivalent to 0.34 +/- 0.05 millimeters per year of sea level rise.
why are the glaciers speeding up?
If you were to poll the world's glaciologists about the cause of the observed outlet glacier speed up, you would get at least three answers: enhanced surface meltwater transfer to the glacier bed; changes in resistive stresses at the downstream ends of the glaciers due to climate-related breakup of floating ice tongues; or a combination of both effects. It will take more measurements and numerical modeling studies to sort this out (and believe me, folks are on the job).
Mass loss due to enhanced melting will continue as long as the warming continues. How long the glacier speed ups due to changes in ice flow will last is less clear. The changes in outlet glacier discharge are so recent that we have more questions than answers but all of the possible scenarios are linked to global warming. The changes are so pervasive and so uniform in sign that they are not likely to be stochastic variations in individual glaciers but beyond that, there are important unanswered questions regarding glacier mechanics.
The two Greenland mass balance estimates reported above agree in sign but not in magnitude. This may in part be due to differences in measurement technique but may also reflect differences in the time intervals over which the observations were made. A very recent, high temporal resolution investigation of Kangerdlugssuaq and Helheim glaciers, both draining the southeastern part of the ice sheet, revealed significant year-to-year variations in ice speed. This makes sense, because as an outlet glacier speeds up, its thickness and surface slope also change. Either thinning or slope reduction could cause the rate of discharge to decrease (thinning near a grounded-to-floating transition in a glacier with a floating ice tongue and appropriate bed geometry could be more complicated).
It is also unclear how far inland the recent drawdowns might propagate. For example, the recent slowdowns of Kangerdlugssuaq and Helheim might indicate a trend toward stabilization of those two glaciers. If so, the the glacier mechanics perturbation will have made its contribution to sea level but future contributions would require new perturbations. Alternatively, the slowing may simply reflect thinning of the downstream reach of the glaciers, while the perturbation continues to propagate upstream. And of course, trends on other glaciers may be different. To make progress here, we need to use numerical models to test the ideas about what's driving the recent change against observational data. Once that is accomplished, we can begin to think about the future.
While appropriate models exist for testing hypotheses regarding the flow of one particular glacier or another, the current generation of ice sheet models used in future climate studies cannot reproduce the changes now underway in Greenland outlet glaciers (though they can reproduce changes in snow fall and surface melting). This is because the climate-model ice sheets lack the detailed ice physics necessary for the task (a nice technical review paper can be found here, as a pdf). Back in January I was summoned (really, summoned is the right word) to a meeting at NOAA's Geophysical Fluid Dynamics Laboratory in Princeton for a discussion of this issue among folks from the climate modeling and ice sheet modeling communities. Everybody understands that this is a big problem and our two communities are working toward a resolution.
Together, enhanced surface melting and speed-ups in outlet glacier flow are causing the Greenland Ice Sheet to lose a total volume of somewhere between 124 and 224 cubic kilometers of ice per year, equivalent to between 0.34 and 0.57 millimeters/year of sea level rise. This is a significant part of the current 3 millimeter/year rate. There is at present no way to predict how the glacier mechanics term in the mass balance equation will change in the future but it is clear that the surface melting term will continue to drive mass loss from the Greenland Ice Sheet. All in all, I don't think we know enough to warrant the Greenland alarm bells being rung in some corners but that doesn't mean that the bell ringers are not right. There is too much yet to learn for me to have confidence one way or the other.