After concerns over lackluster ice recovery in October, November is seeing ice roaring back.  The image above shows the last 3 weeks adding 3 M km2 of sea ice.  (The metric 1 Wadham = 1 M km2 comes from the professor’s predictions of an ice-free Arctic, meaning less than 1 M km2 extent) The Russian shelf seas on the left filled with ice early on.  On the CanAm side, Beaufort at the bottom center is iced over, Canadian Archipelago (center right) is frozen, and Baffin Bay is filling from the north down.  Hudson Bay (far right) first grew fast ice around the edges, and is now half iced over.  A background post is reprinted below, showing that in just 23 days, 2020 has added 3.1 M km2, 50% more than an average 30-day November.

Some years ago reading a thread on global warming at WUWT, I was struck by one person’s comment: “I’m an actuary with limited knowledge of climate metrics, but it seems to me if you want to understand temperature changes, you should analyze the changes, not the temperatures.” That rang bells for me, and I applied that insight in a series of Temperature Trend Analysis studies of surface station temperature records. Those posts are available under this heading. Climate Compilation Part I Temperatures

This post seeks to understand Arctic Sea Ice fluctuations using a similar approach: Focusing on the rates of extent changes rather than the usual study of the ice extents themselves. Fortunately, Sea Ice Index (SII) from NOAA provides a suitable dataset for this project. As many know, SII relies on satellite passive microwave sensors to produce charts of Arctic Ice extents going back to 1979.  The current Version 3 has become more closely aligned with MASIE, the modern form of Naval ice charting in support of Arctic navigation. The SII User Guide is here.

There are statistical analyses available, and the one of interest (table below) is called Sea Ice Index Rates of Change (here). As indicated by the title, this spreadsheet consists not of monthly extents, but changes of extents from the previous month. Specifically, a monthly value is calculated by subtracting the average of the last five days of the previous month from this month’s average of final five days. So the value presents the amount of ice gained or lost during the present month.

These monthly rates of change have been compiled into a baseline for the period 1980 to 2010, which shows the fluctuations of Arctic ice extents over the course of a calendar year. Below is a graph of those averages of monthly changes during the baseline period. Those familiar with Arctic Ice studies will not be surprised at the sine wave form. December end is a relatively neutral point in the cycle, midway between the September Minimum and March Maximum.

The graph makes evident the six spring/summer months of melting and the six autumn/winter months of freezing.  Note that June-August produce the bulk of losses, while October-December show the bulk of gains. Also the peak and valley months of March and September show very little change in extent from beginning to end.

The table of monthly data reveals the variability of ice extents over the last 4 decades.

The values in January show changes from the end of the previous December, and by summing twelve consecutive months we can calculate an annual rate of change for the years 1979 to 2019.

As many know, there has been a decline of Arctic ice extent over these 40 years, averaging 40k km2 per year. But year over year, the changes shift constantly between gains and losses.

Moreover, it seems random as to which months are determinative for a given year. For example, much ado has been printed about October 2020 being slower than expected to refreeze and add ice extents. As it happens in this dataset, October has the highest rate of adding ice. The table below shows the variety of monthly rates in the record as anomalies from the 1980-2010 baseline. In this exhibit a red cell is a negative anomaly (less than baseline for that month) and blue is positive (higher than baseline).

Note that the  +/ –  rate anomalies are distributed all across the grid, sequences of different months in different years, with gains and losses offsetting one another.  Yes, October 2020 recorded a lower than average gain, but higher than 2016. The loss in July 2020 was the largest of the year, during the hot Siberian summer.  The bottom line presents the average anomalies for each month over the period 1979-2020.  Note the rates of gains and losses mostly offset, and the average of all months in the bottom right cell is virtually zero.

A final observation: The graph below shows the Yearend Arctic Ice Extents for the last 30 years.

Year-end Arctic ice extents (last 5 days of December) show three distinct regimes: 1989-1998, 1998-2010, 2010-2019. The average year-end extent 1989-2010 is 13.4M km2. In the last decade, 2009 was 13.0M km2, and ten years later, 2019 was 12.8M km2. So for all the the fluctuations, the net loss was 200k km2, or 1.5%. Talk of an Arctic ice death spiral is fanciful.

These data show a noisy, highly variable natural phenomenon. Clearly, unpredictable factors are in play, principally water structure and circulation, atmospheric circulation regimes, and also incursions and storms. And in the longer view, today’s extents are not unusual.

 

Illustration by Eleanor Lutz shows Earth’s seasonal climate changes. If played in full screen, the four corners present views from top, bottom and sides. It is a visual representation of scientific datasets measuring Arctic ice extents.