Prism logo

Polar Radar for Ice Sheet Measurements

curved ice line
jagged ice
Home>Prism Update>Summer Seminar 2000- Dr. Jezek


So just to give you a brief introduction to RADARSAT-1. RADARSAT-1 is a C-band synthetic aperture radar. And you heard Chris discuss the attributes of synthetic aperture radars.

This is a 5.3 Ghz system or about a 5 or 6 cm wavelength transmitted signal. It's HH pole, that means it transmits a horizontally polarized wave, and it receives a horizontally polarized wave.

This is the instrument itself. These are solar panels, this is the bus, and the radar electronics and this is the antenna which seems to me is 12 or 13 m long, I don't remember exactly how long it is, but it's quite a long antenna to minimize the aperture in the cross-track direction.

Operation of RADARSAT

The radar operates in several different modes. The standard beams operate from about 20 degrees incidence out to about 49 degrees incidence angle.There is an extended low beam that operates from about 10 degrees incidence angle and some extended high beams that operate out at least to 50 degrees. Maybe a little bit more than that, I don't recall off the top of my head.

For the Radarsat projects we used the standard beams and the extended high and the extended low beams. There's also two other modes. ScanSAR modes which provide about a 500 km swathe.The resolution is about a hundred meters. The ScanSAR also provides narrower swathes,but at 50 meter resolution. So the instrument has a whole variety of different observing modes which make it a particularly powerful sensor because it can sweep out a fairly large swathe. It can capture information over a fairly large swathe, not imaging the entirety of the swathe, obviously. But it can also capture information at very fine resolution. The standard beam 3-lookmulti-product is 25 meters. The fine beam dataare a single-look product, or can be single-look products, with resolutions about as fine as 5 X 8 meters.

Research Questions

We're interested in using RADARSAT to look at the details of how ice moves through the polar ice sheets to the oceans. There are several fundamental questions that we want to ask about ice sheets. Are they thickening? Are they thinning? What forces within the ice sheet are modulating that thinning or thickening rate? What external forces in terms of climate are modulating that thinning or thickening rate? Can we make predictions as to whether or not the ice sheets will change in the future given different climate scenarios? Can we predict what that change in ice sheet behavior will be and its consequences in terms of global sea level?

When Is RADARSAT Appropriate?

RADARSAT obviously cannot answer all those questions. To understand whether or not the ice sheet is thickening or thinning, you're probably better off using an altimeter system. Either one of the ERS or TOPEX or GeoSat type altimeters or the upcoming ICESAT and CRYOSAT altimeters. But they too can't answer all the questions. They are primarily measuring the geometry of the ice sheet and changes in the geometry of the ice sheet.

We saw a little bit about how the properties and the forces acting on the ice sheet affect geometry. But the other parts of the equation, the dynamics of the ice sheet, are really well captured by RADARSAT. This is for two reasons. One is that RADARSAT takes an image of the ice sheet allows you to see features that are associated with dynamics, such as crevasses that form along the shear margins of ice streams. In addition, since it's a coherent imaging system you can operate it in interferometric mode,allowing you to capture information that tells you something both about the topography of the ice sheet (not in great detail), but also about the motion of the ice sheet which is extremely important.

And as we saw earlier today, the motion of the ice sheet tells you something about the strain rate of the ice sheet, the derivative of the velocity field. And that derivative of the velocity field is directly related to the stresses that act on the ice sheet.

So in a sense, the RADARSAT program is very complimentary to these altimetric missions that are coming up because it provides that additional bit of information about ice sheet dynamics that ultimately enables you to develop predictions about ice sheet behavior and its response to climate.

Project Phase 1997

The project has been divided up into two phases. The first phase occurred in 1997 and the second phase occurred in 2000.

This illustrates the acquisition phase organization that we managed in order to complete the acquisitions in 2000. This is somewhat similar to the strategy that we employed in 1997,although in 2000, the project
had its own set of complications.

In 1997, the primarychallenge was to obtain complete imaging of the entirety of the Antarctic continent. This was tricky because RADARSAT is a synthetic aperture radar. It looks off to the side and it normally is operated in such a way that it looks to the side and will image the entirety of the Northern Hemisphere. In that mode, a hole from about 81° S latitude to the pole .occurs in southerly coverage and you're pretty much out of luck. In 1997, the satellite itself was rotated in orbit. So for a period of about 30 days or so, the satellite was capable of imaging to the South Pole, enabling us to capture that complete, high-resolution radar image of Antarctica that appeared on my first slide.

Project Phase 2000

In 2000 we, partly for operational reasons, partly because of the solar terrestrial physics that was occurring at the time (a solar maximum with considerable variability in solar flux) and partly because of the age of the satellite, we really were unable to rotate the satellite.

We also had slightly different science objectives. One of the things that we learned from the 1997 mission was that we could do radar interferometry with the RADARSAT-1. We could do really good interferometry. Because of the constraints on our system and because of our science objectives, we opted instead to design the program to capture as much interferometric information as we could. This meant acquiring three complete 24-day cycles of RADARSAT imagery of the area north of about 81° South latitude and capturing the information in ascending and descending modes.

This was the network of teams that were required to manage the data acquisition phase. Basically the Canadian Space Agency handled satellite operations. And those operations were guided by a mission plan developed by the Jet Propulsion Laboratory. The Alaska SAR Facility submitted that mission plan to the CSA mission planning teamand participated in the initial conflict resolution. We used a considerable
percentage of the satellite resources to accomplish this mission, so there were conflicts with other users that had to be resolved.

There was a re-planning team on station at CSA during the mission to provide any re-planning help that we could offer. Because we were doing an interferometric campaign, we required very stringent controls on the repositioning of the satellite, so we had support from JPL orbit maintenance in both developing our plan and doing baseline analysis during the actual acquisitions.

We also had help from the Deep Space Network Orbital Determination Group to monitor the actual position of the satellite.

Downloading the Data

Once the satellite was commanded to acquire data, data were stored onboard, primarily onboard tape recorders which then dumped data to McMurdo. This actually occurred in real time as long as McMurdo was functioning. It had some technical problems that prohibited it from operating during the entirety of the mission.

Most of the data then were either downlinked to Prince Albert Station in Saskatchewan,a ??? ground receiving station north of Ottawa, and the Alaska SAR facility. And as long as data were going into McMurdo we actually also operated a near-real-time link from McMurdo to White Sands facility via TEDRIS(?).

This interface was controlled by Wallops Flight Facility at Goddard Space Flight Center. And enabled us essentially to take data from McMurdo, transmit it through TEDRIS and ____ SAT to JPL and then do real-time coherence validation during the mission to determine whether or not the mission plan was meeting the specifications that we needed. Once data were received, Alaska SAR facility did a quick-look support, again taking a look at the data making sure it was meeting specifications. We used tools developed at ASF for doing some of that validation.

The science team participated by doing coherence validation by assuring that we were getting the coverage that we wanted and these functions were duplicated in several different locations just for redundancy.

And finally the results of the science team were forwarded on to project management which coordinated the unfolding of the mission with the Canadian Space Agency and NASA headquarters. So it was a fairly complex network of organizations involved in the acquisition phase.

1997 Coverage

This is the coverage that was acquired in each of the two missions with the objectives basically being: 1) to study the ice sheet structure and extent; 2) to measure ice sheet surface velocity and to study ice sheet dynamics (As I mentioned this is kind of the complimentary aspect to what Cryosat and GLAS [Geoscience Laser Altimeter System] will be doing); and 3) is to establish benchmarks for assessing changes in ice sheet extent, dynamics and interactions with the coastal environment.

Of course image products are particularly useful for this latter goal and provide a unique snapshot-type glimpse of what the southern continent looks like at any particular time. The 1997 mission consisted of essentially three parts: a pre-nominal acquisition phase, a nominal campaign and an ENSAR campaign. Basically the nominal campaign lasted for 18 days, the minimal amount of time we felt needed for getting complete coverage of Antarctica. The rotation of the satellite in 1997 occurred nearly flawlessly and as a result we were able to begin our acquisitions earlier than the nominal campaign. Kind of putting something in the bank just in case problems arose later in the mission. In fact no problems did arise
and we completed the nominal campaign pretty much on schedule
allowing us to acquire an additional set of data 24 days after the start of the pre-nominal campaign.

Some of you may know that RADARSAT is in a 24 day exact repeat orbit making these data useful for interferometric studies and, in fact, have been used quite widely now by a number of investigators.

2000 Coverage

The 2000 campaign consisted of acquiring data north of about 81 degrees South latitude, but in fact this represents triple, or maybe even perhaps quadruple, the amount of data that was collected in 1997.

The reason for that is that we collected three cycles of descending data and three cycles of ascending data. Two cycles of data are required for getting interferometric observations. Three cycles of data allow you to remove topography effects from the velocity components of the interferometric data without a priori knowledge about the surface elevation of the continent.

And ascending and descending data provide you to get two looks at the surface velocity vector. Obviously a single sending pass only provides a single component of the vector. These two data sets combined provide a means of extracting both horizontal components of the surface velocity vector.

Processing The Data

As I mentioned the processing phase of the project is now unfolding. ASF has done level 0 processing. They've calibrated the data and they finished level 1 processing to the single-look complex imagery. Vexcel has been developing the RAMS2 system. And OSU has now begun the processing of image data - coherence mosaics and the production of velocity products.

The processing phase itself is, in itself, quite complicated. The data for inteferometric processing need to be handled in such a way that requirements on the Doppler centroid, the change in Doppler across a frame, the segmenting frames across interesting glacial object boundaries all have to be taken into consideration. So the actual processing of the data by ASF is, in fact, carefully controlled by an interaction between ASF and the RAMS2 processing system, with ASF providing initial level 0 data, essentially coverages, with quick-look imagery for a single data-take.

This is ingested into the RAMS2 system as a triplet, each of the three cycles, either ascending or descending. Based on that image, and based on the information associated with the Doppler centroid, a variation of the Doppler centroid along-track, the images are reframed into several different frames.

The processing parameters for each frame then are sent back to ASF, which processes the level 0 data to level 1 products which are then reingested into the RAMS2 system. This, in itself, was quite a complicated task to complete.

RAMS2 System

Once that SLC data (Single Look Complex) is available, it's ingested into the RAMS2 system. RAMS2 consists of the old mosaicing component and that system is shown in gray here and is highlighted in the upper left-hand portion of this image.

It basically consists of planning functions - block processing functions - which essentially refine the Ephemeris and Mosaic and does the initial orthorectification of the images using the digital elevation model. Once individual blocks are produced, there's a grand adjustment among all the different blocks, of which there's something like 50 or so, before the data are output as tiles and are provided to the science community through the National Snow and Ice Data Center.

Those components still exist in the RAMS2 system, but in addition to that we now have a component that deals with the interferometric attributes of the data set, that allow formation of interferograms between any two of the triplets acquired in either ascending or descending mode.

The computation of coherence which as you'll see later is quite an interesting diagnostic tool for looking at glacier dynamic signatures for calculating registration offsets and then taking the registration offsets or essentially speckle re-tracking information, combine that with the interferogram, sticking all that information into an archive for developing the final velocity estimates, which essentially conflates all the different possible ways in which you could compute velocity, of which there are a number because of the redundancy in our data set, before a final velocity is established.

Products From RADARSAT

The products that we've gotten out of the RADARSAT project are several now. This is one of the most dramatic, I think, and this is, in fact, the first high-resolution radar image of all of Antarctica. You can see a number of different things in this image.

You can, first and foremost, quite clearly see the coast of Antarctica. There's excellent contrast in the backscatter, or as Chris was pointing out earlier, the normalized backscatter coefficient that he had defined earlier, and the signal loss between sea ice which are shown by these darker gray colors and the coast of Antarctica which is characterized frequently by areas where melting occurs which then result in very high backscatter. And just about anybody could take a pencil and with a little care could quite accurately draw the coastline of Antarctica. In fact they would capture what the coastline of Antarctica looked like in September of 1997.

The scaling is many thousands of kilometers. Looking at a slightly smaller scale, just thousands of kilometers, you can see these long arcing features snaking through the interior of East Antarctica and, to some extent, also into West Antarctica as well. Slightly different tones. But it turns out that if you overlay the known ice divides, sort of the equivalent of the continental divides in more northerly climes, that these patterns for reasons that are still somewhat mysterious, very definitely represent the ice divides in East Antarctica. And you can clearly see the drainage basin associated with this divide that drains primarily down into Byrd Glacier here.

Tour of Antarctica - Part 1

Rather than talking about some of the details by looking at that map, let me instead take you on a bit of a tour of Antarctica. This tour was done collaboratively with the Scientific Visualization Laboratory at Goddard Space Flight Center and is basically a virtual tour of Antarctica, using the RADARSAT image data, the OSU digital elevation model of Antarctica, and draping the two together and then producing this animation.

The animation starts off over Ross Island. Here's McMurdo Station. Here you can see the sea ice runway. This is Black Island. This is White Island. It pans away from McMurdo toward the Trans-Antarctic Mountains and the McMurdo Dry Valleys, before turning south back across the Ross Ice Shelf.

Here is Byrd Glacier which is the largest outlet glacier in Antarctica. It moves up over the Beardmore Glacier. This is essentially the route that Scott took toward the South Pole.

South Pole Station is coming up here and in the RADARSAT imagery, it is indeed visible. Here's South Pole Station here with one of the runways.

Tour of Antarctica - Part 2

Moving away from South Pole back toward West Antarctica
crossing the Trans-Antarctic Mountains. You can see the Willems Ice Stream, which I mentioned earlier today, one of the large ice streams in West Antarctica. Here's Ice Streams D and E.

Small volcanoes poke up through the shield of the ice sheet here. This is the Crary Mountains, I believe. Here's Mount Takahe, another shield volcano.

We move now east toward Pine Island glacier, and the Antarctic Peninsula. The Peninsula has been the site of a lot of interest because of the retreat of the Larsen Ice Shelf. This is the Southern Larsen Ice Shelf, Larsen B. This is Larsen A, you can see in 1997, it's essentially obliterated. This portion of Larsen is now gone as well. The remainder seems relatively stable.

This is panning back over the Antarctic Peninsula toward the Ronne Ice Shelf. You can see the large rifts that develop and eventually cause the formation of the huge icebergs characteristic of Antarctica. Here's the ice streams draining Coates Land into the Filchner Ice Shelf. These ice streams are quite long and, in fact, were first revealed in their entirety, by the RADARSAT project.

You can see the Recovery Glacier Ice Stream extending hundreds of kilometers into the interior of East Antarctica. This is the Fimbul Ice Shelf. You can see the Fimbul is punctuated by numerous ice rises that retard its flow outward toward the Southern Ocean.

This is Jutulstraumen Glacier. You can see it's a large ice tongue that punches out through the Fimbul ice shelves. This is just moving along the coast of the Fimbul ice shelf. You can see a lot of structure in the surface of the ice sheet, some of which are small outcrops but most of which is texture associated with the motion of the ice sheet over a complex glacier bed.

Tour of Antarctica - Part 3

As we move farther, now toward the east, we now come upon the Amery Ice Shelf and the Lambert Glacier. Lambert is also one of the largest glaciers in the world and its effect on Antarctica is significant as I pointed out earlier, you can see the down-draw of the surface of the ice sheet due to the effect of the Lambert Glacier.

This is a movie across East Antarctica, now toward Lake Vostok. Lake Vostok is buried beneath 4000 meters of ice, yet it is a fresh-water lake about 250 meters deep. Here's the outline of Lake Vostok. Here is the road from the coast to Vostok station where people are about 100 meters above the surface of the lake drilling in anticipation of eventually drilling into the lake.

These are large snow dunes. These are features several kilometers long with a few kilometer wave length. These characterize much of East Antarctica. Here you can see some more of the dunes.

And eventually we are going to come back upon the Victoria Land coast and wind up in McMurdo again. Here you can see it coming up over the mountains. Here again is McMurdo Station, Willy Field and Erebus Ice Tongue. So it provides kind of a graphic and kind of an interesting overview of Antarctica, that it would be almost impossible to obtain any other way.

Block Coverage for MAMM

The primary product of AMM1 was that mosaic. Just to illustrate that MAMM, the Modified Antarctic Mission, is going to be producing somewhat different products, here's the block coverage that we imagined for the Modified Antarctic Mapping Project.

It's slightly different in that the yellow areas will, because of the beam coverage that we selected, consist of 25 m resolution images. But utilizing the fine beam mode in MAMM, not used in AMM1, we've also acquired very high-resolution data essentially 10 m after final processing over the fast glacier areas. So all of these areas will have what we are calling mini-mosaics, at 10 m resolution, and these are really quite beautiful.

Comparison of AAM1 and MAMM Maps

Just to give you a feeling for the quality of the mini-mosaics. This is AMM1 25m multi-look imagery over a brush stroke of crevices in the Getz Ice shelf area. You can certainly see the crevices in this area but you can't see any particularly clear definition. In the 10 m finding product you can see these crevasses quite clearly, and in fact, you can see them splaying across much of the imagery.

There's two reasons for this: one is, (we) might as well have as good detailed information as possible on the data, but it's also interesting to note that MAMM occurred with three cycles - 24 days per cycle. So there's an opportunity for capturing data on day one and day 48 say. With this definition of crevices, if in places where glacier speeds are in excess of about 250 to 300 meters per year it's actually possible to do routine feature re-tracking, not speckle re-tracking but feature re-tracking, over these kinds of areas to get an estimate of what the surface velocity is because we've got this very good, very high resolution data.

Mapping Glacier Flow

In terms of derived science products there's a number of things that you can do with the data once they're cast in the map format. As I mentioned earlier the simplest thing to do is to just take a pencil and make an outline of the map and, in fact, this is the coastline that we derived for Antarctica using the MAMM data. This is coastline derived using 125 m data, we're in the process now of finishing up some automated algorithms to extract the coast line from the 25 m resolution data. We think the automated algorithms will be useful because it's an extremely boring process to go around with a mouse and try and measure the map at that great detail. We think the operator error alone complicates things. We think that the automated algorithms we will have available shortly should do the job quite well.

The other thing you can do is to map flow stripes. Because the ice sheet is deforming, because it is deforming over a complex bed, and because it's being channeled between mountain-passes in many places, the surface of the ice sheet becomes disrupted and results in long flow patterns being superimposed on the ice sheet surface. These are quite easily visible in the RADARSAT imagery and we've now taken that information and compiled many of the flow stripes -not all of them, we haven't done all of the small outlet glaciers here in the Transantarctic's - but we've tried to characterize places where fast glacier flow is occurring around most of the continent. And, we think this is the first map of fast glaciers derived from actual data not a model result for all of Antarctica.

You can see that there are incredibly detailed and complicated networks of fast glacier flow draining east Antarctica here into the Filcher Ice Shelf. This set of ice streams drains portions of east and west Antarctica into the Irani. Here's the west Antarctic Ice Streams, Pine Island Glacier and Thwaites Glacier, Lambert Glacier, a whole network of systems draining through the, I can't remember the name of this mountain range, but there's a mountain range that pokes up along this flank of Queen Maud Land and drains out toward the Fimbul.

One interesting thing, as I suggested this morning, is that you don't really see any evidence for flow stripes here in Wilkes Land. Whether that's simply a consequence of not detecting fast glacier flow for some reason in this area, or that there really isn't fast glacier flow is one of the science questions that we would like to answer when we analyze more of the velocity products from the MAMM project.

Detecting Changes Over Time

Once you have a baseline for measuring changes in the ice sheet, the next natural thing to do is just go ahead and do that to see what changes you can detect. Here is the front of the Ross Ice Shelf. Here is Ross Island. Here is Roosevelt Island in September of 1997. Here is the margin of the Ross Ice Shelf in September, 2000. And in March of 2000, one of the largest icebergs ever observed broke free from the snout of the Ross Ice Shelf. And now you can see the portion of the ice shelf that was removed. You can see that the calving of that iceberg was associated with these large rifts that span across the front of the Ross Ice Shelf. These seem to be nucleated quite far upstream. And you can see that they are kind of a regular process.

The calving of the large icebergs from the Ross Ice Shelf, and from the Filchner-Ronne, has received a lot of attention in the press over the years. But I think, as you can see just by looking at this image and by the talk that I gave earlier today, you can realize that this is basically an episodic process. Snow is dumped on the surface of the ice sheet and for the ice sheet to remain in equilibrium some amount of that snow has to eventually be melted off or calved off into the sea. So we expect for an equilibrium ice sheet to see these kind of, admittedly, dramatic processes occurring, but we expect these processes to be episodic and the natural consequence of an equilibrium ice sheet.

Fimbul Ice Shelf Changes

This an area along the Fimbul Ice Shelf that fronts the coast of Queen Maud Land. We've looked at ice shelf margin advance and retreat in this area as well. In this case we've used 1963 corona imagery to get an early estimate of ice margin positions. We've used 1997 RADARSAT data as well and, in fact, we've thrown in 1970's era Landsat imagery.

As I think you can see along this coast there has not been very much change in the position of ice shelf margins over the roughly 34 year period of observations. Some of the snouts of these large ice tongues that extend out into the ocean have in fact broken off. The Jutulstraumen Glacier, is a good example of this. However I think again you can intuitively sense that these will be the most mechanically unstable portions of the ice sheet and in situations where the tongue gets large and the storm systems begin to impact upon it, it seems only reasonable that these things will break free. In other areas you can actually see some of these ice tongues starting to re-advance.

Comparisons Across the Continent

We think this area is interesting because you'll notice that it is about the same latitude as the southerly extent of the ice shelves that are starting to disintegrate in the Antarctic Peninsula. One of the powerful things about the RADARSAT data, in fact, any of the satellite data, is that it gives you a continental scale view of processes occurring.

So, as we'll see in a moment the events that are occurring in the Antarctic Peninsula are really quite dramatic; they're really quite astonishing. But they're also relatively confined to this area and are not necessarily representative of processes occurring around the rest of the continent.

Retreat of the Larsen Ice Shelf

This is the Larsen Ice Shelf. This is an image taken of Larsen B in April of 2002, just a few months ago. Here is Seal Nunataks. Here is Seal Nunataks in about 1992. This area of ice shelf in 1992 is now this area of ice shelf in 2002. You can see quite a dramatic retreat.

Sustained retreat of Larsen A, completely gone in 1997, and the retreat of Larsen B, resulting in the complete demise of Larsen B in 2002.

I've speculated and now there's been a recent paper suggesting
that, in fact, one factor that might be influencing the retreat of these ice shelves is changes in the ocean environment. And in fact, if you watch this animation, you can see that the ice shelf is not being replaced by multiyear ice characteristic of this part of the Weddell Sea. But rather it's being replaced by thin ice or, probably in the summer, open-water polynea. And I wondered whether or not the retreat of the ice shelf is as much of a consequence of change in ocean circulation and the bringing of heat from the ocean to the bottom of these ice shelves, as much as the retreat of the ice shelf is a consequence of changing atmospheric temperatures, which in fact are known to be warming in the Antarctic peninsula.

Wordie Ice Shelf

Here's another animation done over the same years. This is the Wordie Ice Shelf on the western flank of the Antarctic Peninsula. The Wordie used to fill the entirety of Marguerite Bay, but you can see that it's pretty much wiped out. This is 2000. This is '92. And this is '97. As I say, the Wordie Ice Shelf used to cover this entire bay. By 2000, it has retreated pretty much up into the interior of the peninsula near the grounding line. But the one thing that's interesting to watch is this little glacier here. As the ice shelf has retreated, it looks as though the confining effects of that ice shelf have been removed, and that little glacier, you can see, shoots out about 7 kilometers between '92 and '97.

This, to me, seems not terribly unexpected, but it does reveal the role
of ice shelves in restraining parts of the interior ice sheet. I think this is quite good graphic evidence of the restraining effect of ice shelves, which has been, for some reason, an issue of some controversy over the last number of years.

Shirase Glacier

This is the final movie of this sort. This is kind of an interesting movie because it doesn't show a lot of interesting science, perhaps, but it is kind of neat in that it basically captures the history of remote sensing.

The imagery starts out with 1963 declassified imagery over the
(I can't think of the name) Shirase Glacier in East Antarctica. it goes on to 1988 LANDSAT data, 1997 RADARSAT data, 2000 RADARSAT data. Here's the DSP?? you can see the Shirase extending far out into Lutzow-Holm Bay. In 2000, basically Lutzow-Holm Bay is devoid of fast ice and Shirase Glacier is kind of a stub. 1997 it starts to grow a little bit, then in 2000 starts to retreat. Whether or not the presence of fast ice in 1963 is contributing to the extent of the ice tongue associated with Shirase Glacier is I can't answer, but nevertheless it is kind of an interesting movie to watch.

Land Glacier

Just to show you that things can be awfully constant, in fact, surprisingly constant in Antarctica, this is the Land Glacier, which is located in West Antarctica. And it's dumping onto the Amundsen Sea. What's kind of amazing about this image is that this is the 1997 image of Land Glacier. This is the 2000 image of Land Glacier. This is sea ice out in here. These are small icebergs that have calved off of the front of Land Glacier. These are a little bit larger icebergs. But when you compare this image and this image, people often ask me if I've goofed up and if I've just used the same image. You can see that this block of ice is essentially in the same position. So is this one. This one has actually changed just slightly. You can see the patterns of the sea ice fractures are about the same. But even this ensemble of little icebergs is almost exactly the same. You can see many of the same fragments.

It's just astonishing. I can't really explain it. Why this area, which is known to be buffeted by some considerable storms, can maintain itself as well as it does.

Capturing Surface Velocity Information

One of the important features of the RADARSAT data is the ability to use it in interferometric mode and to capture information about the surface velocity of the ice sheet. This is a surface velocity map derived from AMM1 interferometric data by colleagues at JPL over the West Antarctic ice streams and you can see the streaming nature of these ice streams quite clearly revealed in the velocity data. You can see that there is a complex interplay of trunks and tributaries that feed down through these things until they finally reach the Ross Ice Shelf.

This is the Lambert Glacier and Amery Ice Shelf compiled using data from AMM1 and MAMM. Again you can see the network of tributaries, the Fisher and the Miller glaciers, draining into the Lambert Glacier and eventually winding up in the Amery Ice Shelf.

Lambert Glacier Velocities

And that's particularly nicely captured in this animation that again superimposes AMM1 data onto the OSUDEM and then also includes a graphic representation of the velocity field.

It's panning down onto the Lambert Glacier. These are the interferometrically derived velocities. This is panning up the glacier. This is the Mawson Enscarpment. Here's the Lambert Glacier. Miller-Fisher Tributaries. You can see the increase in speed here where there is presumably some bottom melting occurring, allowing the ice to move a little bit faster before it reaches this part of the ice shelf where presumably freezing is occurring.

East Antarctic Ice Streem Velocities

These are velocities that were obtained over the east Antarctic ice streams and again you can see a very strong contrast in surface velocity between the ice streams, here of Recovery, Slessor and Bailey ice streams and the surrounding areas of nearly stagnant ice.

We've looked at this area in quite some detail and it looks as though these ice streams, unlike their west Antarctic counterparts, are moving through depressions in the sub-glacial topography. They're also associated with these rather peculiar tributary glaciers. This is what we're calling Ramp Glacier. You can see it's about 250 km long but it's everywhere about 50 km wide. It's very difficult also to see, when you inspect this imagery, exactly where the ice that feeds this tributary comes from. There seems to be very little input from the sides. There must be a tremendous amount of ice pouring in from this snout here.

One of the things that we've been able to do with these data is to actually compare crevices and rifts that we see on the surface in 1963 imagery and track their position in the 1997 imagery to get 30 year average velocities. Those average velocities, along this profile line here, are shown in this curve, the green ones are the average velocities. We can compare those to the NSAR velocities, which are essentially instantaneous velocities, obtained in 1997. And you can see that the two curves match up quite well. There's only one place right here near the snout of the glacier, where there may be a statistically significant difference between the average velocity and the instantaneous velocity, but even that's just barely significant.

It's interesting because this area of the Filchner Ice Shelf has undergone significant calving. There was an iceberg that calved off about this much of the ice shelf five or six years ago. And yet there seems to be very little effect on the interior velocity field. Presumably that's because the terminus of the ice shelf hasn't retreated past this pinch-point between Coats Land and Berkner Island. Were that pinch-point to be breached, then presumably there would be a much more dramatic change in velocities and a probably more dramatic retreat of the ice sheet in this area.

Drygalski Ice Tongue

As I mentioned, we're quite excited about getting velocities using the interferometery. But in places where the glacier is moving fast enough, you can just take a look and see the motion of the ice sheet.

This is AMM1 and AMM2 imagery of the Drygalski Ice Tongue in Victoria Land. And by just doing a flicker on the two co-registered images, you can quite clearly see the forward motion of the Drygalski Ice Tongue between 1997 and 2000. And you can also see the rotation of this block of ice as it's being dragged away from this glacier, whatever it is, along with the Drygalski Ice hang. You can see it being rotated out. This is actually the derived velocities using feature retracking techniques on the AMM1 and the MAMM data. But as I mentioned, and even in the case of the Drygalski Ice Tongue, it's moving fast enough we can probably do feature retracking on these crevasses and other serrated features just using MAMM data alone - by comparing Cycle 1 and Cycle 3.

Coherence Maps

Along with getting a interferometric map and a velocity field, a by-product of the coherent analysis is a coherence map. This is the coherence over the east Antarctic ice streams. Regions of high coherence are shown in red, grading into these blue areas of very low coherence. You can see a number of important features in the coherence data. You can quite clearly see the margins of the ice streams where the rapid shearing essentially de-correlates the signal even over the 24 day period. You can also see the grounding line, the place where the interior ice sheet goes afloat and become the ice shelf, as is well evidenced by these bands of de-correlation.

Aviator Glacier Coherence

For the MAMM data we've been taking a very careful look at how we calculate coherence. This is coherence calculated using the complex data not just the amplitude data alone. This is actually the magnitude of the correlation peak for the Aviator Glacier.

This is Aviator Glacier. It's moving down in this direction before taking this elbow out toward the ocean. You see that the coherence peak has a quite complicated pattern over Aviator Glacier. The correlation peak, or the coherence, is quite dark along the margins where it's being de-correlated because of the shear stresses that build up. But there's some other pattern that's being superimposed on the surface of the ice sheet of the glacier in the interior here. Not quite sure what that is! It looks a lot like some of the melt features you see in LANDSAT data over parts of the Antarctic Peninsula. But what the exact physical interpretation of the signature is remains something of a mystery.


In summary, I want to show you this picture of South Pole Station. This is the runway that supports South Pole Station, the station itself, the old South Pole Station now I guess is located right here.

But it's kind of an interesting image in that what you can also see in this imagery but you can't see when actually on the surface is the remnants of the old IGY station. This station is now buried but because the radar does penetrate into the surface, you can see the evidence of the old runway that supported the old station which is now buried but is still visible on this RADARSAT imagery.

In terms of the project itself, there are measurable, scientifically interesting changes occurring in Antarctica within a 3 year period between AMM1 and MAMM. There's changes in the ice sheet and the coastal sea ice. Are these changes related? You know, those changes that I noted in the Antarctic Peninsula? And, perhaps most importantly, RAMS places these changes in context of the continental ice sheet. So we can see whether or not a change, spectacular although it may be, is a regional effect, or can be interpreted as a continental-, and perhaps global-, scale effect.




PRISM © 2002, 2003 is brought to you by
NSF Logo
National Science Foundation University of Kansas
NASA logo
KTEC logo
Kansas Technology Enterprise Corporation
KU Jayhawk logo
University of Kansas