ice extent (Figure 4). This result was expected 

 a priori, although the good level of correlation 

 was unexpected. 



Because the season length distribution 

 is nearly normal (Figure 5) and correlates well 

 with another environmental index, the season 

 length average (140 days) and standard de- 

 viation (35 days) are meaningful. Therefore, I 

 have defined an "average" ice season length 

 as the mean plus or minus one standard 

 deviation (Table 5). 



CONCLUSION 



Despite concerns over changing tech- 

 nology impacting the iceberg counts, high- 

 quality sea-ice data has validated most of the 

 recent IIP iceberg populations. With one ex- 

 ception, the SLAR year's populations do not 

 appear to be significantly overcounted. Con- 

 versely, the pre-SLAR year's data does not 

 appear to be undercounted. 



Based on the sea-ice correlation, I pre- 

 sented iceberg population severity classes 

 (Table 4). Similarly, the iceberg season length 

 correlated well with the sea-ice extents and, as 

 such, seems to be a believable index. Based 

 on the near-normal distribution, ice season 

 length severity classes are presented in 

 Table 5. 



52 



