The submarine record of Arctic sea ice draft: what is it telling us? (INVITED)

There are numerous signs of a reduction in the Arctic sea ice cover in the 1990s. One very compelling record is that of satellite-observed ice extent in summer. Another record of great import to the question of a changing Arctic Ocean comes from U.S. and British Navy submarine upward-looking sonars. Since 1958 these data have been accumulating at a rate of about one and a half cruises per year. Of these only a third have been processed, and only data from 18 cruises are publicly available in any form. So what we can glean from these data today will eventually be much improved upon.

The end-of-summer draft appears to have declined from about 3 m in 1960s to about 2 m in the 1990s (Rothrock et al., 1999). The winter draft seems to be reduced as well, but not as strongly. The decrease clearly comes in the proportion of ice thicker than 4 m and not in the fraction of open water.

In the collective record analyzed so far the mean draft is 2.95 m. The variance is 1.68 m^2. Together, a mean annual cycle, an interannual trend, and a mean regional variation account for about half of this variance. These dependences on season and region are not new or surprising, but by isolating them one can estimate the interannual component of change which is intriguing. The remaining half of the variance is due to variability in ice advection and in thermal forcing and to error.

Being for operational purposes, these data leave something to be desired, but it is believed that errors can be reduced to between 10 and 30 cm. We will briefly review the errors that can arise from (1) the correction to zero draft in segments of open water, (2) the digitization of chart records and (3) the different footprint seen by the sonar from different depths. All of these can bias the draft data.