San José State University
Thayer Watkins
Silicon Valley
& Tornado Alley

The Variability of the Area of Arctic
Sea Ice Throughout the Year

The record of the extent of arctic sea ice in recent years shown below indicates that while the sea ice declined to a minimum in 2007 it has been increasing since then. The graph displays the mean levels throughout the year and curves that represent the mean plus and minus one standard deviation. The standard deviation used is a constant, indicating that is computed from all of the data rather than for each part of the year.

Background information on the matter of the extent of arctic sea ice at the end of the summer see Arctic Sea Ice in the Summer. The purpose of this webpage is compute the standard deviation of the sea ice extent throughout the year. The data used is the arctic sea ice extent made available from the website of the IARC-JAXA Information System (IJIS) at the International Arctic Research Center in corporation with the Japan Aerospace Exploration Agency (JAXA).

The graph provided by IARC-JAXA, shown below, indicates the variation in the sea ice extent is much larger at the times of the minimum and maximum that it is the other times of the year.

The downloadable data from IARC-JAXA has some normal problems. There is missing data for some days. If these occurred at ends of data period those days were left out of the computation of the mean and variance. If the missing data points were within the data period an interpolated figure was used based upon the adjacent years. The two data points for February 29th (2004 and 2008) were simply left out of the analysis.

The mean and standard deviation for each day were computed and displayed below.

The graph of the standard deviations displays the expected seasonal variaton, but the abrupt change indicates an abnormal problem in the data. An investigation reveals the source of the problem is in the data for 2005. The graph of the data for that shows an abrupt disappearance of millions of square kilometers of sea on night of October 31st, Halloween night. This is thermodynamically impossible.

Something went wrong with the sensor or the data transmission process. The anomalous data for for 2005 must be treated as missing data and the means and standard deviations recomputed. The data for the whole month of November of 2005 appears to be invalid. The data for this month was replaced with the interpolated values based upon the corresponding days in 2004 and 2006.

The recomputed means and standard deviations based upon the revisions in the 2005 data are shown below.

The profile of standard deviations still seems a bit implausible. Apparently there is some change in the data collection from October to November. Nevertheless that profile is used to compute confidence limits at the 95 percent level for the area of arctice sea ice throughout the year. These limits, which are approximately the mean levels plus or minus two standard deviation units, are shown in the graph below.

These limits show that at the time of the minimum ice in September there is a much wider confidence interval than in say May. Likewise the confidence interval is wider at the time of the maximum area in March.

The graph below shows the area of arctic sea in 2007 along with the 95 percent level confidence limits.

The minimum ice in 2007 was a record low, but as the above graph shows the area for 2007 did not fall below the lower confidence limit. In other words, the sea ice area in 2007 was not significantly different from the mean at the 95 percent level of confidence. Since 2007 the arctic sea ice has been rising and in 2010, so far, is near the mean.

To be continued.

HOME PAGE OF applet-magic
HOME PAGE OF Thayer Watkins