16 November 2004

Senate Committee Testimony: Observed Changes in Arctic Sea Ice Cover and Projections for the Future

On 16 November, the U.S. Senate Committee on Commerce, Science and Transportation held a hearing about Global Climate Change to hear testimony on the assessment recently released by the Arctic Council and the International Arctic Sciences Committee. NSIDC's Mark Serreze participated in the first of two panels, presenting information about findings in the Arctic. Serreze highlighted changes in sea ice concentration, and the acceleration in sea ice losses, noting in particular the record or near-record lows of the last three years. The text of his testimony appears below.


The dominant feature of the Arctic Ocean is its floating sea ice cover. This thin, variable layer strongly controls the heat balance of the planet, and changes in size with the seasons and year by year. On average, the ice covers about 14 million square kilometers in winter and about half this area in summer. The ice typically ranges from 1-5 m in thickness, depending on the region, season and year. Records for the past several decades document significant shrinking of the ice cover, with extreme retreat in the past three years. There is some evidence of attendant thinning of the sea ice. The observed decline in Arctic sea ice is fundamentally in accord with climate model projections of continued ice losses through the 21 st century.


1) Over the past 30 years, the annual average extent of Arctic sea has decreased by about 8%. Losses for this period are much larger in late summer to early autumn (15- 20%).

The most accurate assessments of Arctic sea ice extent are for late 1978 onwards. These are based on NASA satellite data (so-called “passive microwave” imagery). Other forms of satellite data (back to the early 1970s), ship reports and aircraft reconnaissance can extend the record back to the beginning of the 20th century. It appears the decline in sea ice began around 1960, with the late summer and early autumn losses again standing out. The last three years (2002, 2003, and 2004) have seen extreme sea ice reductions.

2) There is evidence that the sea ice cover has also become thinner.

There are no reliable methods to monitor sea ice thickness from satellites. The best information comes from upward-looking sonar carried by submarines operating under the ice. Comparisons between sonar records collected during 1958-1976 with more recent data (1993-1997) indicate that between the two periods, mean ice thickness at the end of summer decreased by over a meter for much of the central Arctic Ocean. Subsequent analyses, based on both observations and state-of-the art sea ice models, give further evidence for thinning from the late 1980s through 1997, but with some recent recovery.

3) The observed changes in Arctic sea ice extent and thickness are best explained from a combination of climate warming and changes in the circulation of the sea ice.

Changes in temperature over the Arctic Ocean have been assessed using data from Russian manned ice camps (1950-1991), arrays of drifting buoys (1979 onwards) and satellite remote sensing (1981 onwards). While each analysis yields somewhat different results due to differences in data type and record length, they convincingly point to spring/summer warming and a longer summer melt season. Starting around 1970, the so-called North Atlantic (or Arctic) Oscillation (NAO), a large-scale pattern of variability in the atmospheric circulation, began to shift towards its “positive mode.” While this is known to have contributed to recent Arctic warming, the primary impact on the sea ice is through changes in the surface winds, which alter the circulation of the ice cover. There is ample evidence that through various mechanisms, these altered wind patterns have helped to reduce the extent and thickness of the ice. The NAO has regressed to a more neutral state in the past five years, yet the ice cover has continued to decline, as seen in the extreme losses of 2002, 2003 and 2004.

4) The sea ice reductions are in fundamental agreement with model projections.

Global climate models used in the Arctic Climate Impact Assessment and other investigations point to continued sea ice losses through the 21 st century in response to greenhouse gas warming. There are disparities between different models in the projected rates and spatial patterns of change. Ice loss is nevertheless a universal feature of the projections.

5) Attribution of the observed Arctic sea ice decline to greenhouse gas warming is complicated by variability in the atmospheric circulation.

Building on Point #3, the decline in Arctic sea ice extent has been “boosted” by changes in the atmospheric circulation associated with the NAO. The NAO has and will always contribute to “natural” variability of the Arctic climate system, which complicates detection of a greenhouse signal. However, while the past five years have seen this atmospheric pattern return to a more normal state, the sea ice cover is still declining. There is also growing evidence that through various mechanisms, the effects of greenhouse warming may favor the positive mode of the North Atlantic Oscillation that fosters sea ice losses.


The most reasonable assessment is that the Arctic sea ice cover is beginning to respond to the effects of greenhouse gas warming. This assessment is based on: (a) observational evidence, (b) increasing confidence in projections from state-of-the-art global climate models, (c) ice core records indicating that carbon dioxide levels in the atmosphere are now the highest of the past 400,000 years, (d) other “proxy” records (such as from tree ring analyses) indicating that recent climate warming is moving outside of the bounds of natural variability over the past 1000 years. The particularly strong natural variability in the Arctic climate system lends some uncertainty to this assessment. Another source of uncertainty is the extent to which greenhouse gas loading and stratospheric ozone losses in the Arctic alter major patterns of atmospheric variability such as the North Atlantic Oscillation.

Mark C. Serreze, PhD
National Snow and Ice Data Center
University of Colorado , Boulder

Return to News Room