A Tale of Two Indices




Sorry to be serious on April 1.  I am not a fan of ice charts restricted to one month, for reasons illustrated in the post Ice House of Mirrors (some humor there in honor of this day.) But March monthly average sets the baseline for the year’s melt season, and so there is considerable attention and significance attached to the month just concluded.

Here is a chart showing March 2016 compared to the previous ten Marches according to two different indices of Sea Ice Extent: MASIE (Multisensor Analyzed Sea Ice Extent) produced by the National Ice Center and SII (Sea Ice Index) produced by NOAA (both accessed at NSIDC).

MASIE and SII March

It is evident that the March annual maximum is trending slightly upward in MASIE and slightly downward in SII. Note that the indices were quite similar the first five years. Then since 2010, SII has declined quite strongly.

Note on Sea Ice Resolution:

Northern Hemisphere Spatial Coverage

Sea Ice Index from NOAA is based on 25 km cells and 15% ice coverage. That means if a grid cell 25X25, or 625 km2 is estimated to have at least 15% ice, then 625 km2 is added to the total extent. In the mapping details, grid cells vary between 382 to 664 km2 with latitudes.  And the satellites’ Field of View (FOV) is actually an ellipsoid ranging from 486 to 3330 km2 depending on the channel and frequency.  More info is here.

MASIE is based on 4 km cells and 40% ice coverage. Thus, for MASIE estimates, if a grid cell is deemed to have at least 40% ice, then 16 km2 is added to the total extent.

The significantly higher resolution in MASIE means that any error in detecting ice cover at the threshold level affects only 16 km2 in the MASIE total, versus at least 600 km2 variation in SII.  A few dozen SII cells falling below the 15% threshold is reported as a sizeable loss of ice in the Arctic.

Putting NOAA Reports in Context

With the background above, we can interpret NOAA`s meaning when they report (here) that 2016 winter ice extent is the smallest on record. That refers to the annual maximum daily extent they observed on March 24. Climatology usually uses the March average to indicate the year’s maximum (given the volatility of daily readings). As we can see, 2016 March average was higher than 2015, virtually tied with 2006 and just below 2011. SII showed March ice to be 364k km2 less than MASIE.

NOAA: “The extent in 2016 was 431,000 square miles (1.12 million square kilometers) below the 1981–2010 average, which is like carving away an area of ice the combined size of Texas, Oklahoma, Arkansas, and most of Louisiana.”

In other words, 2016 is below average by about the size of 2000 grid cells in their system, or 1.5% out of 136,192.  

For more background on the two datasets see here.

2016 in Perspective

As an example, consider how this March compares in the two indices.

In the graph MASIE shows 2016 starting the month at average extent, then declining and then recovering.  2016 ended below average in extent and comparable to 2015.  Meanwhile SII showed much less extent, rising to a late maximum and then declining sharply to be 400k km2 less at day 91.


Results for the first quarter of the year show the large differences between the two indices. SII portrays this winter as abnormally low, while MASIE shows an average year, slightly higher than 2015.

Looking at the extents in the various seas making up the NH Sea Ice, it is clear that all are typical, with three exceptions.  Compared to 2015, Barents and Baffin Bay are down (Barents lost 100k km2 in the last five days), while Okhotsk surplus more than offsets the losses elsewhere.


As the divergence of SII increases, it becomes less clear what it is really measuring.

The tables below give the reported ice extents in M km2:

Month 2016 2016 MASIE SII SII Deficit
Averages MASIE SII 2016-2015 2016-2015 SII-MASIE
Jan 13.922 13.472 -0.019 -0.131 -0.450
Feb 14.804 14.210 0.121 -0.199 -0.593
Mar 14.769 14.405 0.101 0.038 -0.364



New Russian Nuclear Icebreaker “Yamal”, sharing the name of the infamous hockey stick tree.



  1. Canadian Climate Guy · April 1, 2016

    Reblogged this on Canadian Climate Guy.


  2. joekano76 · April 1, 2016

    Reblogged this on TheFlippinTruth.


  3. Pingback: A Tale Of Two Indices | NOT A LOT OF PEOPLE KNOW THAT
  4. Jack Broughton · April 2, 2016

    Seems to demonstrate two truths:
    1. It is not an easy and readily reproducible measurement, as you comment, seems like about 3 % uncertainty in the values before you start to analyse them?
    2. Nature is showing its natural variability and a long-term mean with standard deviations would be more valid than regression analysis.

    However, the AGW believers are desperately searching for any trend that they can claim is consistent with AGW.


  5. daveburton · April 3, 2016

    Very informative article, thank you for educating me! I’ve added a link to this article on my sealevel.info web site, and I’ve added your blog to my list of recommended blogs.


  6. Taylor Pohlman · April 5, 2016

    While the difference between MASIE and the regular NSIDC is only around 3% or so in the winter ( which might almost be forgivable, where it gets ridiculous is in September, where it’s not unusual to see the difference between the two at well over 10%, with MASIE being higher, even though it’s threshold, as you note, is much higher than NSIDC. I’m told that the algorithms are badly fooled by ice surface meltwater. In the days DMI published their 30% numbers, it was a good check/correlation with MASIE.
    The weird thing is that the famous “minimum” contest is based on NSIDC, not MASIE, so winning is mostly about guessing how bad NSDIC screws up the calculation vs. trying to figure out how much ice is really left. Imagine if we only knew the speed of light to 10% accuracy – and this is called science?


    • Ron Clutz · April 9, 2016

      Thanks Taylor. It is good to know others are keeping an eye on this and have an historical context. Some years ago I was taken aback by the quote from Cavalieri et al. “We find a baseline difference in integrated ice concentration coverage north of 45N of 3.85% ± 0.73% during November to May (ice chart concentrations are larger). In summer, the difference between the two sources of data rises to a maximum of 23% peaking in early August, equivalent to ice coverage the size of Greenland.“

      Also: “Compared with the Canadian regional ice charts, the NASA Team algorithm underestimates the total ice-covered area by 20.4% to 33.5% during ice melt in the summer and by 7.6% to 43.5% during ice growth in the late fall.”


  7. Pethefin · April 9, 2016

    An example of the challenges in satellite monitoring of the sea ice was provided today by NSDIC:


    • Ron Clutz · April 9, 2016

      petafin, perhaps your are referring to the NH ice extent reported by SII from NOAA. Here are the last three days:
      2016 4 6 14.016
      2016 4 7 15.065
      2016 4 8 14.320
      Could April 7 be off by 1M km2?


      • Ron Clutz · April 9, 2016

        Also the addition of 300k km2 yesterday aligns SII with MASIE for April 8 2016.


      • Pethefin · April 11, 2016

        Actually I was referring to the fact that the sea-ice image claims that the entire Baltic Sea is covered with ice all the down to the Danish straits. And there’s additional ice here and there and everywhere, look for example at the are around Iceland. WUWT has a post on the problems today.


      • Ron Clutz · April 11, 2016

        Yes, I posted there on two spikes of 1M km2 shown by SII on April 7 and 10. Something is wrong, and MASIE shows no such thing.


  8. uwe.roland.gross · June 23, 2019

    Reblogged this on Climate- Science.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s