We have been looking at sea ice trends below the arctic circle at the request of Anthony Watts. It is a curiosity of his that he’s been asking Walt Meier of the NSIDC to consider for some time. I am a fan of the NSIDC because their data access is excellent and they answer questions very quickly and reasonably. This post is from the daily sea ice data as presented by the NSIDC on their FTP site. It is several gigabytes so if you are serious, fileZilla is a good free software to facilitate download.

From the video’s produced, it is apparent that a lot of noisy data exists at the extreme lower edge of detection. This data results in sea ice being detected in isolated squares of warm latitudes with no chance of having actual sea ice. The effect is visible in this video showing both poles through the history of satellite ice data.

You can see the great lakes around Michigan sparkle year round with sea ice detection noise yet we know that the beaches in July and August are 90F and and the only ice you will find would be in someones drink.bbSo the false detection at the low end of the microwave sensor range is a known factor. Plotting the sea ice area outside of the circle above, we can see that the sea ice never quite hits zero. It gets pretty close though.

It is interesting that the minimum value has a shift at about 1998. Fluctuations in the minimums don’t seem to have much trend so I assume the effect is instrument related. There are a number of different instruments on different satellites which have been combined to create this trend.

There is a statistically signficant trend in the sea ice outside of the arctic circle. This ice is completely melting as expected every year so the trend we see is a result of reduced formation. I’m curious now what percentage of this new formation is in the open sea vs landlocked lakes but that will be a subject for a future post.

The Arctic trend shown next is comprised of everything above the arctic circle.

The trend is also significantly negative.

For confirmation of the above, these results differ very little from the UIUC cryosphere page. UIUC does infill the pole hole with estimated data whereas I simply leave it out. The pole hole is the region around the pole where the satellite instruments do not reach. This region changed size early in the record leaving some difficult choices as to how to handle the newly available data. I simply used the large mask throughout the record when creating trends. My trend may be slightly more negative than theirs due to the difference but the fractional differences are very small.

The purpose behind some of this work was to determine what percentage of the above trend is seasonal ice unrelated to polar cap melt. Taking 152000km^2/decade of seasonal ice out of the 518000km^2 total, that amounts to 29% of the melt trend is due to ice which, in the last 34 years, is completely seasonal in nature. Arctic circle ice also melts every year so it is a mix of seasonal and perennial (multi-year) ice. The ratio of the seasonal ice 152000 to the mixed seasonal/multiyear ice 366000 is 41% – not sure what use that is but it is interesting to consider that the multi-year ice loss is quite a bit less than these graphs show.

This next graph requires some interpretation. It is a ratio of the seasonal ice area outside the Arctic circle to the ice inside the Arctic circle.

In looking at this plot, I read it by observing the annual peaks only. There is a visibly evident trend in the peak values each year. This means that the peak seasonal ice is decreasing at a higher rate than the perennial Arctic peak ice. This seems to be a confirmation of gradual warming processes controlling the peak amount as we would expect the southerly ice to show the effects first. One problem with this graph is that it reads near zero during the time of greatest melting so we really only have good information at the peaks.

More work needs to be done. The next thing I want to do is look at landlocked ice to see if there are trends in satellite detection ability. After that, I have some new ideas to isolate whether the 2007 and 2010 arctic minimums were localized effects caused by ocean currents or if they were larger in scope.

My focus on this now is because the data is interesting and extensive and I haven’t seen much work done on regional effects in blogland. I am very much skeptical that we should be worried about any of this. If you add up all the sea ice in the world, we have a heck of a lot of it at any given time. About 19million Km^2 on average. If you take the global anomaly and offset it by the average amount, it gives a good idea what the sea ice death spiral is working out to be.

This data was compiled from the daily Ease grid files presented by the NSIDC. Code for this post his here. ice code Save it and change the extension to R as WordPress won’t allow upload of text or R files. It is written in several sections: functions, Northern hemisphere ice, Southern hemisphere ice, plotting calls etc. Authors of the various parts include RomanM, RyanO, Nic, Steve McIntyre and myself. Nearly all of their work has been modified so many times by me they may not recognize it but still deserve credit for the good parts. Any errors are my own.



