probability of C-CORE to correctly classify a target as an iceberg or ship in MANICE. The 

 ENVISAT targets not confirmed by reconnaissance represent false targets reported in MAINCE. 



Based on this small data set. the overall POD was 50%, and the overall POC was 72% (Table 2). 

 The POD in IS7 was 100%; however, only one ship was seen in this IS mode, and it was 

 incorrectly classified as an iceberg. The IS5 acquisition on 15 May compared to the IS4 

 acquisition on 21 May shows that both were consistent with the overall 50% POD, but the IS4 

 acquisition had a much higher POC (Table 2). Environmental conditions on both days were 

 relatively the same, with winds at 5-10 knots and seas at 2-3 meters on 15 May and winds at 10- 

 15 knots and seas at 1-2 meters on 21 May. Therefore, environmental conditions would have had 

 little effect on the POC numbers between the different IS modes. A conclusion could be drawn 

 that IS4 has a better POC than IS5 since the number of icebergs confirmed in both data sets was 

 relatively the same; however, the number of ships in the area must be considered. The acquisition 

 on 21 May was in an iceberg-dense region of the ocean where there was little vessel traffic, thus 

 likely affecting the POC number on that day. 



Regardless of the target density, each acquisition had an average of 1 3 false targets, with no more 

 than 14 and no less than 1 1 for any given day during the validation efforts (Table 2). False targets 

 are of a concern to IIP, especially if they are limit setters. Ice Patrol does not set limits around 

 radar targets, which are targets that cannot be positively identified. Although a false target could 

 potentially provide a safer Limit of All Known Ice (LAKI), it may unnecessarily cost the mariners 

 who heed IIP's LAKI valuable time on their transits. 



Conclusion 



The POD and POC results from the validation effort did not meet HP's thresholds for operational 

 use. However, IIP plans to continue validation efforts during the 2005 ice season, focusing 

 primarily on IS6 and IS7 acquisitions. In addition, C-CORE has recently completed changes to 

 the classification algorithms, which will be implemented in the IDS prior to the 2005 ice season. 

 This is expected to improve overall probability of classification. 



Insufficient data was collected and confirmed in the 2004 ice season to make a determination on 

 the POD and POC in the recommended IS6 and IS7 modes. Moreover, POC decreases in areas 

 of mixed target density, as seen when comparing the results of 15 May with those of 21 May. 

 Because HP's area of most concern is in the transatlantic shipping lanes, where both icebergs and 

 ships are prevalent and it is necessary to distinguish between them. Ice Patrol does not intend to 

 use ENVISAT imagery to determine the LAKI. However, HP could potentially use this data to 

 assess the iceberg-feeder population in an area with less shipping traffic if POD numbers increase 

 with the IS6 and IS7 acquisitions and improvements are made to the algorithm to increase the 

 POC. Probability of detection and POC numbers must increase before IIP considers using this 

 information to update the IIP iceberg database in BAPS. If POD and POC numbers increased to 

 75% and 90%, respectively, IIP would consider strategic use of this data. An example of strategic 

 use for this service would be to target a specific location, such as the Flemish Pass, by obtaining 

 repeated looks in the same area to determine the population of potential limit-setting icebergs. 



The required lead time necessary to direct ENVISAT acquisitions proved very difficult to work 

 with because predicting iceberg location two weeks into the future in the dynamic North Atlantic 

 is extremely challenging. This two-week lead time cannot be used tactically to determine limit- 

 setting icebergs because of the day-to-day changes in the LAKI. However, the near real-fime 



53 



