VII. GLACIERS AND ICE SHEETS

SAR Interferometry and Surface Change Detection

Ice sheets, ice streams, floating ice shelves, and mountain glaciers together constitute the cryosphere, an integral and dynamic part of Earth's land-ocean-atmosphere system. The cryosphere is sensitive to changes in mean ambient temperature and precipitation on a wide range of time scales, thus various components of the cryosphere are senstitive monitors of climate change. While some components (e.g., the polar ice sheets) respond very slowly to climate change, with response times of hundreds to thousands of years, other components (e.g., mountain glaciers) respond rapidly, on the order of seasons to decades. This wide range of response complicates the definition and implementation of appropriate monitoring strategies. One feature common to virtually all glaciers and ice sheets is that they occur in remote areas, making satellite observations crucial.

Although the polar regions are remote, they play a critical role in the Earth's climate, ocean, and atmosphere system. For example, it is widely known that global warming could melt the Antarctic or Greenland ice sheets and lead to a rise in sea level, with major impact on all coastal communities. Less well known is the fact that sea level could rise very quickly due to accleration of ice discharge into the oceans. The West Antarctic ice sheet is believed to have an unstable character and holds sufficient ice volume to raise sea level by 6 m (Alley and Whillans, 1991). Consideration of dynamic response times (Oerlemans, 1989) suggests that this could happen catastrophically, perhaps over time scales a short as 100 years.

Topographic data are basic to glacier and ice sheet research, because surface topography directly reflects the driving and resisting forces that affect ice motion (e.g., gravitational acceleration in the direction of surface gradient and bedrock topography, respectively). Unfortunately, no high-resolution topographic maps exist of Greenland, Antarctica, and most mountain glaciers. While lower resolution coverage exists, it is inadequate for many applications. Much work is needed to acquire high-quality observations and to make the data available to the scientific community in a standardized format. Clearly, satellite techniques have many advantages for acquiring complete high-accuracy topographic data in the polar regions. Images by themselves, whether optical or SAR, are very useful for visualizing features, but quantitative elevation data are needed for most scientific applications.

As soon as possible, we must determine whether total ice mass is stable, shrinking, or growing. High-resolution elevation data are critical to determining total ice volume and ice extent and, when repeated after one or several years, allow an estimate of the rate and nature of changes. Monitoring of the margins may be important in early detection of changes (van der Veen, 1991). SAR interferometry has proven capable of detecting the grounding line that separates floating from grounded ice in Antarctica, and should be very sensitive to any changes (Goldstein et al., 1993; also see cover image). If height data are used to assess changes, it is important that basic elevation data be as accurate as possible (at least 1-m accuracy, preferably 20 cm or better) to enable accurate assessment of the rate of change after several years of measurements.

High-resolution elevation data are also important to the study of mountain glaciers. Because they are much smaller in volume than polar ice sheets, mountain glaciers are much more responsive to climate change on short (10-100 year) time scales. Mountain glaciers probably play a major role in sea-level change (Meier, 1984, 1990), so repeated global measurement of glacier topography is an important long-term goal. At present, uncertainties in the mass balance of mountain glaciers are sufficiently large that quantitative modeling of their role in sea-level rise is difficult. However, available data are good enough to indicate that mountain glacier wastage plays a major (probably dominant) role in sea-level rise. The following is summarized from a recent National Academy of Sciences report (Sea Level Change, NAS, 1990).

The current rate of sea-level rise is estimated to be about 1.5 to 2.0 mm/yr, with large contributions from the melting of mountain glaciers (~0.5 mm/yr) and thermal expansion of the oceans. The Antarctic and Greenland ice sheets apparently contribute little to current sea-level rise, but this could change in the near future. Available data for several mountain glaciers suggest that they are wasting at a rate sufficient to cause their disappearance within several hundred years, and that the maximum melt rate has not yet occurred, assuming present trends continue.

It is important to realize that there are several weaknesses in available data. First, the number of mountain glaciers monitored is too small to be representative, and there are good arguments to suggest that local variability is large. Secondly, due to their remote location, the data base is particularly weak in the three regions containing the largest number of mountain glaciers (the mountains bordering the Gulf of Alaska, Central Asia, and Patagonia). Finally, much of the data for ice mass change actually are derived from length change, and the two quantities are only weakly correlated. Changes in elevation are more directly related to mass change.

All three weaknesses can be remedied by space-based elevation measurements. Since seasonal changes are large, monitoring elevation on a seasonal basis would separate seasonal from longer term trends. Meier (1990) suggests that the magnitude of seasonal change may be correlated with longer term trends. Thus, a short (2- to 3-year) space mission could yield data useful for evaluating longer term trends, even in the absence of follow-on missions.

Coverage Issues

One advantage of remote-sensing technology is that it enables study of large regions of our planet that we know very little about. One weakness in the current ensemble of satellite data is that many satellite missions do not cover polar regions, or provide only partial coverage. Satellite missions typically leave a gap of 8deg. latitude or more at the poles. For example, ERS-1 coverage extends to 82deg., while Geosat and Seasat extend only to 72deg.. Radarsat is planned to have a polar phase, but no altimetry measurements are planned. Clearly there is a need for satellite missions that are better configured for polar studies.

Satellite-based SAR interferometry has two important roles to play in polar studies. First, SAR interferometry can provide complete high-resolution high-accuracy topographic data. Second, repeat-pass interferometry can be used to measure ice flow and assess other changes. The cover image (from Goldstein et al., 1993) shows the first direct measurement of ice flow velocity from space without ground control.

Much work remains before the full potential of this technique can be exploited. Pilot studies involving SAR interferometry over snow and ice are required. Specific questions must be answered: How precise are SAR interferometry measurements of height and ice velocity over snow and bare ice? What are the effects of scattering and penetration on the fidelity of interferograms, and how does this depend on radar frequency? What is the range of correlation/decorrelation behavior over the range of microenvironments characterizing the world's glaciers and ice sheets. For example, the data presented by Goldstein et al. (1993) on the Rutford ice stream was one of 12 examples tried; only two cases provided meaningful results. Presumably, changes in snow/ice surface conditions, including snowfall, blowing snow, melting, and other temperature-related changes, affect the scattering and reflectivity of the surface. For mountain glaciers, high relief, high spatial variability, and rapid temporal variation may complicate the acquisition and interpretation of SAR interferometry data.

High-resolution topographic data can provide important secondary information including surface roughness and scattering properties. Using spatial statistics, this information can be exploited to characterize morphological prototypes, describe scaling properties, and derive classification maps (e.g., Herzfeld 1993; Herzfeld et al., 1993). Repeated mapping missions, even in the absence of correlation, can be used to assess large-scale changes through the differencing of elevation measurements. Examples include the advance/retreat of glaciers, location of ice shelves (Partington et al., 1987), thickening and thinning of glaciers, iceberg calving, ice-ocean interaction, motion of sea ice, iceberg tracking, land subsidence, surges, and glacier velocity.

Recommendations:

(1) Pursue research aimed at better understanding of the temporal decorrelation of SAR interferometry in polar regions and temperate mountain glaciers.

(2) Pursue a space-based SAR interferometry mission, ensuring polar coverage, and acquire an accurate high-resolution global DEM. The mission should include a laser altimeter. If feasible, use SAR interfero-metry data from this mission to compute a global velocity field for the Earth's glaciers and ice caps, and to compute changes in glacier and ice sheet elevation on seasonal to longer time scales.