= 
DERIVATION OF SEA ICE MOTION FROM SEASAT 
SAR IMAGES USING IMAGE PROCESSING TECHNIQUES 
by 
Pat S.. CHAVEZ, JR., 
Paul G. TELEKI 
William CAMPBELL 
U.S. Geological Survey 
Flagstaff, Arizona 
Reston, Virginia 
Tacoma, Washington 
Two Seasat L-band (23.5 cm wavelength) radar images recorded three days 
apart and digitally correlated by the Jet Propulsion Laboratory into image 
format, were used to extract information on the movement of sea ice within a 
30 km? area in the Beaufort Sea. Pixel locations of ice floes that could be 
identified on both images were selected; twenty "control" points were used in 
the final analysis. Several land (i.e., non-moving) points were used to 
geometrically register the two images together. 
The pixel location differences between the images were used to generate 
two vector files, one of which represents the direction of ice movement 
(THETA), the other the magnitude of ice displacement (R). The THETA and R 
vectors, calculated for the 20 control points were converted into raster-image 
format, such that the spatial resolution of the 30 km? area was represented by 
à 500 by 500 pixel array. A THETA and R value was assigned to each of the 500 
by 500 pixels in the array. Spatial filtering techniques were used for 
digital interpolation of the THETA and R values; the size and number of 
filters used were based upon the spacing and dynamic range of the original 
twenty control points. The interpolated results were then displayed using 
both black and white and color techniques. 
611 
UE re e nin