NSIDC recently switched the baseline against which we analyze Arctic sea ice extent. Previously, we relied on a baseline that coincided with the beginning of the satellite period and stretched 20 years, from 1979 to 2000. The new baseline runs from 1981 to 2010, covering 30 years. Why did we make such a change?
Switching to a 30-year baseline allows us to be consistent with other climate monitoring agencies, which commonly use a 30-year time period for conducting analyses. This new baseline also helps account for the wider variations observed in Arctic sea ice extent.
“The new baseline does obscure the recent downward trend to some degree because the low values don’t look as low,” said Walt Meier, an NSIDC research scientist. However, the new baseline includes wider variations as a function of that declining trend and the recent record low Arctic sea ice extents.
What does changing the baseline period mean?
When scientists are looking at Arctic sea ice, they need to define how much variation is considered normal. A baseline provides a reference range to compare how past and current measurements change, or deviate, over time, and allows scientists to identify trends that fall outside of that normal range.
According to NSIDC research scientist Walt Meier, NSIDC uses what is called a standard deviation, which measures how much the data tends to fall above or below the average. Two standard deviations, what NSIDC uses, encompass roughly 95% of the data. The remaining 5% that fall outside the deviation range are considered outliers, highlighting data that are unusual.
“An outlier could be an error in the data, may highlight a very rare occurrence, or could indicate a trend,” Meier said. “In this case, we conduct quality control to make sure that the data are not in error. So for Arctic sea ice, we actually see the strong trend since the year 2000, which has made Arctic sea ice values in summer to be greater than two standard deviations from the 1979 to 2000 average.”
Has the sea ice decline become the new normal?
Changing to 1981 to 2010 includes the low values after 2000, causing the standard deviation to become wider. “Now, here is the key thing,” Meier said. “Calculating the standard deviation assumes that spread of the data is roughly balanced on each side of the average. The same amount of data should be above the average as below the average throughout the time period, and their distance from the average should be consistent as well.”
However, there is clearly a declining trend in the Arctic sea ice extent, meaning that more data has fallen below the average toward the end of the time period. The declining trend reduces the 1981 to 2010 average compared to 1979 to 2000, but increases the amount of variability, i.e., the standard deviation. This new range of deviation changes only 2.5% on the high end (compared to 1979 to 2000), but 20.9% on the low end, which accommodates the downward trend.
Does changing the baseline obscure the decline in sea ice?
Meier cautions that the standard deviation for the 1981 to 2010 period can somewhat obscure the sharp downward trend in ice extent since the early 2000s, which was better highlighted by comparison to the 1979 to 2000 baseline. But with a 30-year comparison period, we can average out year-to-year changes and other short-term variations such as El Nino and the Arctic Oscillation. As a result, this longer time series allows us to more confidently see the climate signal. While the downward trend in Arctic sea ice does not look as visually extreme in the graph, scientifically it is still quite important. In 2012, summer minimum sea ice extent was less than half seen in any year during the 20th century, and that loss resulted in a great deal of ocean and atmosphere warming that is being transmitted into the rest of the Earth system.
Changing this baseline doesn’t mean we don’t need to worry about declining sea ice.
“The bottom line is that the average period and standard deviations are convenient references for conveying information on the changes in sea ice,” Meier said. “Whatever reference is used, the essential scientific conclusions are the same: there is still a very strong decline in Arctic sea ice, particularly during summer, and small increases in Antarctic sea ice.”
Related Icelights posts
Related NSIDC content