undetected by the OMW. We did not 

 expect to observe the growler(s), however. 

 The Wide-2 beam characteristics, while 

 beneficial for iceberg reconnaissance in 

 terms of areal coverage and high incidence 

 angles, are not conducive to detecting 

 contacts 15m or smaller. We observed 

 one additional contact in the OMW output 

 than was observed visually. It is thought 

 that this may have been a ship that was 

 observed in the RADARSAT scene, but 

 had moved out of the bounds of the scene 

 area by the time the aircraft arrived. There 

 were five ships that were observed at the 

 edge of the RADARSAT footprint area, 

 which could have been within the area 1.5 

 to 4.5 hours earlier. Though unlikely on 

 account of the outstanding visibility, this 

 result does present the possibility that an 

 iceberg may have been detected by 

 RADARSAT that was undetected by visible 

 means. 



We extracted the pixel values from 

 the RADARSAT scene and plotted them 

 for each iceberg as mesh wire plots, to 

 demonstrate the diffuse scattering 

 characteristic of icebergs (Figure 3). We 

 did the same with the ship targets from the 

 scene (Figure 4) to show the qualitative 

 differences between the ship's radar return 

 and those from the icebergs. With each 

 iceberg plot, we included a photograph of 

 the iceberg that correlated to the particular 

 RADARSAT target. One can observe that 

 some of the morphological features of the 

 icebergs are captured in the RADARSAT 

 return. More work is required to discuss 

 this more quantitatively. 



Discussion 



Despite the difficulty in arranging the 

 logistics, the one successful underflight 

 demonstrates that RADARSAT beam- 

 mode W-2 is appropriate for determining 

 the locations of point targets in the IIP 

 Operations Area. Further study is needed 



to develop the expertise to quickly and 

 accurately classify targets as ice/non-ice. 

 Work is currently underway by a number of 

 agencies, including the Canadian Ice 

 Service, the Danish Meteorological 

 Institute, and the Canadian Centre for Cold 

 Ocean Research and Engineering (C- 

 CORE), as well as some private 

 companies such as Satlantic, Inc. and 

 Space Imaging, Inc., to develop algorithms 

 to quickly and accurately filter targets 

 based on their radar return. 



There was no clear relationship 

 between iceberg size and DN, or 

 "brightness". In fact, there should be no 

 strong relationship, as the radar 

 backscatter from a point target is more 

 dependent upon the shape of the object 

 than upon the size (Raney, 1994). This 

 could be observed in the comparison 

 between the mesh plots of the RADARSAT 

 targets and the photograph of the 

 corresponding iceberg. The brightest 

 target was iceberg #16/Target A, with a 

 maximum DN of 54970. This iceberg, a 

 drydock shape with several pinnacles 

 surrounding a melt pool, likely presented a 

 highly efficient discrete reflector that 

 caused a high return of the incident 

 radiation in this part of the iceberg. 

 However, it is possible that this extremely 

 high brightness value is an artifact of the 

 speckle suppression procedures. In 

 pursuing a qualitative target decision as 

 either iceberg or ship, however, this artifact 

 is not of great consequence. 



The RADARSAT signature for 

 Iceberg #1 had a maximum brightness of 

 7285, relatively weak compared to the 

 other targets. This contact remained 

 undetected by the OMW and was very 

 difficult to discern, even with a priori 

 knowledge of its location. This may be due 

 to its peculiar shape (three small pinnacles 

 connected to a large underwater mass) or 

 to its increased incidence angle (as a result 



60 



