Skip to the content

IOP A community website from IOP Publishing

environmentalresearchweb blog

« Energy, land and power | Main | Nuclear waste moved off the agenda »

The ASTER global digital elevation model

ASTER, the Advanced Spaceborne Thermal Emission and Reflection radiometer, has been looking at the Earth since it was launched in 1999 on a satellite called Terra. A joint Japanese-American venture, ASTER is the foundation for all sorts of environmental monitoring efforts, but in particular for the GLIMS initiative. GLIMS stands for Global Land Ice Measurements from Space, and while we are doing acronyms I may as well tell you that DEM is short for digital elevation model.

Glaciologists need good maps, which nowadays means DEMs derived from air photos or satellite images, but we have a special interest in up-to-date maps because nearly all of our glaciers are getting smaller. This includes getting thinner. An accurate map of surface elevation tells us the state of the glacier at the date of the image. If we have an accurate map or DEM from some earlier date, we can subtract the newer DEM from the older to measure the change in elevation.

There are complications. Many older maps turn out to be much less accurate than modern ones. Often, air photos of glaciers taken in the golden age of mapping, roughly from the 1940s to the 1960s, show featureless white expanses. Snow just isn’t photogenic. Surprisingly often, there simply aren’t any accurate maps from the olden days.

Nevertheless, GLIMS was conceived in the 1990s as a way of bringing our knowledge of glacier surface elevations up to date. This is where ASTER comes in. It has two cameras, one looking straight down and one looking backwards. The two cameras see the same patch of surface, but several seconds apart because the satellite is moving. So ASTER can see in stereo, and can work out the surface elevation using trigonometry.

Participants in GLIMS have been documenting glacier changes with images from ASTER and other satellite sensors. ASTER, though still working, is beginning to feel its age, but something very notable happened in June of this year: the ASTER Global DEM was released. This staggering product contains more than 250 billion land-surface elevations, every 30 metres over nearly the whole world.

ASTER DEMs are composites from images obtained between 1999 and 2008, so they may prove not to be well adapted for measuring glacier elevation changes. But there is one way in which they are sure to excel: the mapping of regions that have never been mapped accurately before.

The ASTER GDEM had a predecessor in the form of SRTM, the Shuttle Radar Topography Mission of February 2000. Interpreting radar imagery presents problems of its own, but SRTM DEMs have already been put to good use in monitoring glacier elevation changes. One problem is that the shuttle orbit limits coverage to between 60°N and 60°S, so that most of the world’s glaciers are not covered. Aggravatingly, however, it was decided that the South Sandwich Islands, the only significant land between 56°S and 60°S, weren’t worth the trouble.

They are more than ordinarily troublesome as far as conventional mapping is concerned: windy, cliff-girt and ice-covered, therefore nearly impossible to land on by boat, helicopter or any other conveyance; and too far away to justify sending a plane equipped for photogrammetry in the faint hope that they won’t be cloud-covered.


Montagu Island, the largest (101 km<sup>2</sup>) of the South Sandwich Islands. Exposed land is green and brown. Ice is purple and blue, with the glacier margin in red.
Montagu Island, the largest (101 km2) of the South Sandwich Islands. Exposed land is green and brown. Ice is purple and blue, with the glacier margin in red.

The ASTER GDEM team doesn’t think the South Sandwich Islands are too much trouble, even though they had to contend with clouds, ice floes that look like land, and, in the case of the island in the map, a continuing volcanic eruption. You can see one of their problems in the map: the south coast is greyed out because the GDEM masked off the surrounding ocean by relying on the best previous estimate of Montagu Island’s position, which was 4 km too far to the northeast. Less obviously, the largest of the pale blue patches is the summit of Mount Belinda. It rises only to 1070 m, a full 300 m lower than previously believed on the basis of a shipboard measurement made in 1930.

1930 was the date of the first of only three “serious” visits to Montagu Island. You can see the top of Mount Belinda from sea level, if it isn’t cloudy. But even on the clearest of days you can’t see the rest of the summit plateau, on which no human eye has ever gazed. So this map is exploration in the modern mode, that is, from orbit. Thanks to ASTER, it is the first ever contour map of Montagu – including the bits we have never seen before.

Chances are you will never need to think about Montagu Island again. But it holds 50 times as much glacier ice as the iconic and much better known Kilimanjaro, and it is a part of the global glaciological picture that doesn’t deserve to be left out. If Montagu Island should happen to cross your path in the future, remember that you first saw it contoured courtesy of the ASTER GDEM.

TrackBack

TrackBack URL for this entry:
http://www.iop.org/mt4/mt-tb.cgi/3468

Comments (10)

  • 1 dave richards November 18, 2009 6:01 PM

    nice, but do the ASTER GDEM data provide an aquisition date, i.e. are the data a 'snapshot' for a fixed time or time period (e.g. 1999-2008)?

  • 2 Graham Cogley November 18, 2009 6:31 PM

    Dave - No, the GDEM is not accompanied by acquisition dates, so that you have to assume that the date is “December 1999 to June 2008”. The uncertainty is thus about 3 years, which reduces its usefulness for comparisons with recent maps (say post-1980s) but not necessarily for comparisons with maps from the 1950s or 1960s (or earlier).

    The number of stereo image pairs used for each pixel is given in an auxiliary file, but the patterns are very complex. Decisions about which pairs to use must have been made by the processing algorithm on a pixel-by-pixel basis. I think this means that it would not be practical to provide a list of dates.

    Graham.

  • 3 Anonymous November 18, 2009 6:56 PM

    Thanks,
    it's a shame that the ASTER data aren't fixed to a specific date, say 2008, as comparing these to the SRTM data aquired in 2000 would provide a comparatively easy way of estimating change in volume of a lot of glaciers (those between 60°N and 56°S) during this period