654 



The overall picture of Arctic ice thickness has been developed 

 by several investigators from a variety of sources. One such 

 portrayal is shown in Figure 18, depicting the mean winter ice 

 thickness developed from some of the early Navy data (adapted 

 from publicly available data). 



The Navy also holds some of the acoustic ice draft data in 

 processed forms. For its own purposes the Navy has developed 

 classified and unclassified databases of under ice features. For 

 instance, ice information was needed in acoustic models that 

 were developed to predict sound propagation loss in the Arctic 

 region. Since 1989 the Navy ha,s analyzed the DIPS data from 

 14 of the 20 cruises on which the DIPS system was deployed. 

 As a result of these analyses, 1 statistical parameters including 

 mean ice draft, rms ice draft, and standard deviation of the mean 

 ice draft, as well as a variety of parameters describing ridging, 

 were calculated for each nautical mile when ice was overhead. 

 These data are available in two different databases. 



The first database remains classified because it includes po- 

 sition and timing details for individual cruises as well as the 

 specific data extracted from individual cruises. In thaidatabase, 

 results are generally binned into 60 x 60 nm regions. The values 

 are interpolated for regions where data are not available, and a 

 clear distinction is made between real information and 

 interpolated values. The second unclassified database isentitled 

 Ice Profile Database V3. 1 . The data upon which V3. 1 is based 

 are identical to the classified database. In addition, the 

 60x60 nm geographic binning is retained. However, the data 

 are grouped into two seasons, fall and spring, and the results of 

 all the data obtained for each 60x60 nm block are averaged. 

 Values are interpolated for blocks where data are unavailable, 

 and information is not provided to allow the user to distinguish 

 real average values from interpolated values. 



In its present format the Ice Profile Database V3. 1 cannot be 

 exploited by the scientific community. Because of the global 

 averaging, ii is impossible to look at temporal variations in ice 

 conditions, potentially the most interesting aspect that could be 

 studied. Also, the grouping of data into spring and fall does 

 not provide adequate temporal resolution to allow sorting of 

 changes with time (although it may not be necessary to know 

 time and place exactly). Another problem with V3.I is that 



Figure 18. Mean Ice Draft from Variovs Early 

 Submarine Cruises 



Over the Arctic region, ice thichiess romposttei have ban constructed 

 from previottsly released submarine acoustic and a variety of supporting 

 data. 



This figure shows one such depiction of mean winter ice thichiesses. Not 

 surprisingly the mean is driven by the thick multiyear ice causing the peak 

 at .im seen in Figure 1 7. 



Although the distribution in Figure 17 results from measurements in one 

 part of the Beaufort sea. aiui Figure 18 shows a long term winter mean for 

 the entire Arctic region, the two depictions are quite consistent. 



measured data cannot be distinguished from interpolated 

 numbers that are filling data voids. 



The Profile Database could be made useful if the data were only 

 averaged within individual cruises and within each 60 x 60 nm 

 box (30x30 nm boxes would be preferable). Further, the box 

 averages should be coupled with a year and a time of year. If 

 these changes were made, scientists probably could perform 

 more detailed analyses of ice patterns and trends. 



The practical accessibility of the .inalog data is limited by its 

 present format. If these data are ever to be analyzed and thereby 

 made available for further studies, now is the time. The early 

 charts are nearly 40 years old. Yet, all the information required 

 for an adequate analysis still exists at the Arctic Submarine 

 Laboratory. 



With regard to the analog data, we expect that current state-of- 

 the-art image analysis techniques could prove extremely useful 

 in speeding the processing of the analog data. If it were decided 



26 



