T. A. Scambos
The great ice sheets of the earth, Greenland and Antarctica, remain among the most poorly mapped areas of the planet, in part due to the fact that their ice surface morphology is subtle and not easily detected with the naked eye. However, enhanced satellite imagery can reveal these subtle features, and has provided a means of rapidly improving our knowledge of various ice flow features within the interiors of the ice sheets. Two techniques, data cumulation and photoclinometry, have recently been applied to areas of interest on both the great ice sheets using AVHRR satellite data, yielding a qualitative and quantitative improvement in the characterization of the surface morphology.
Data cumulation combines several scenes of the same area, taken under similar lighting conditions, into a single scene with enhanced spatial and radiometric resolution. The process depends on accurate coregistration of the individual scenes to a subpixel level, which is accomplished by correlating the grey-scale values and interpolating a best-correlated coregistration vector. The images are resampled to a smaller pixel size and combined using the registration vector, and then the average pixel value at each of the new, smaller pixels is determined. Using AVHRR images, data cumulation can improve spatial resolution to approximately 700 meters (pixel size of AVHRR is ~1100 meters at nadir). Radiometric resolution is also improved since several brightness measurements are made at each pixel location. This translates into more detail in image maps of surface features, since subtle surface slope changes are made visible in the combined image. Data cumulated scenes of the Siple Coast of Antarctica and northeast Greenland show a significant qualitative improvement in surface detail.
Photoclinometry, sometimes referred to as shape-from-shading, uses digital image data as a quantitative map of surface slope, and seeks to extract surface topography from the grey-scale data. In this technique, the key component is a knowledge of the slope-to-image-pixel-brightness relationship, called the photometric function. Over most of the ice sheets, some knowledge of the surface topography exists from radar-altimetry-based DEMs. However, the spatial resolution of the DEMs is in general too coarse (typically 5 to 20 km) to reveal important flow features. In this application of photoclinometry, this initial coarse knowledge of the surface slope is used to develop the photometric function for AVHRR scenes. The scenes are filtered to the same spatial resolution as the DEMs and then mean brightness of the image data is compared to regional slope to derive the function. This is then applied to the 1100 meter pixels to create a much more-detailed surface topographic map, resulting in a quantitative improvement in knowledge of surface morphology. A demonstration of this technique is presented from a section of the recently-identified ice stream in northeast Greenland.
Contact Information:
Ted A. Scambos
National Snow and Ice Data Center (NSIDC)
Campus Box 449
University of Colorado
Boulder, Colorado 80309-0449
Telephone: (303) 492-1113
email: teds@icehouse.colorado.edu