Choosing a Base to compare ASI lost on PIOMAS Volume

[Juan C. García]

The Excel file with all the information is at the following link:

https://www.dropbox.com/s/vngwlttpwvz7on9/Piomas_ASI_Lost.xlsx?dl=0

These data and graphs could help you to choose:

[Juan C. García]

From my opinion, it is important to discuss which base does NSIDC, IPCC, PIOMAS, etc. choose to compare the monthly values that they make public.

It is kind of official that we should use a 30-year time frame, to say that it is an example of “Climate”. That is, climate is an average of weather. So, to choose a base that represents climate, we should have a 30-year time frame.

The problem is that the ASI volume and extent had being changing, decade after decade. So, a 30-year time frame hides the melting that we had on the ASI.

My personal choice, if I have not restrictions choosing, will be the 1980-89 decade, because it shows that we have been losing ASI, decade by decade.

A second choice will be the 1979-2000 base, because it is a longer time frame (22 years). But in a way, I refuse to accept that we should choose time frames like 1981-2010 or 1979-2015, because they hide to much of the Artic sea ice that has been melting all this years.

What do you think?

[P-maker]

Since WMO and its predecessor decided that standard normal periods should be 1931-60, 1961-90 and 1991-2020, and since “Standard Normals” should continue to have some predictive power (in the absence of reliable forecasts), let me suggest that we are now so far into the new standard normal period 1991-2020, that it is timely to start working with provisory standard normals for the period 1991-2016. Although the remaining four years may still change the mean and extremes a fair bit, we still have 26 years of recent data to rely on.

Thus, I do not accept the premises given in this vote.

Please add the option of voting for "Provisory Standard Normal Period 1991-2016".

[Juan C. García]

Please add the option of voting for "Provisory Standard Normal Period 1991-2016".

Done!

[Buddy]

I would like to see it compared to BOTH (1) 1980 - 1989, AND (2) 2000 - 2009

[budmantis]

Please add the option of voting for "Provisory Standard Normal Period 1991-2016".

Done!

I prefer the 1991-2016 time frame, with the 1979-2000 as my second choice.

[Richard Rathbone]

NSIDC don't use a baseline for their monthly values. They just stick to Novembers in November, Decembers in December ... Seems like an appropriate method to me.

PIOMAS has the problem that the seasonal pattern has changed, so their anomaly plot has become misleading. For showing how the seasonal pattern has changed, decadal averages work pretty well.

[Juan C. García]

This is the graph of the 6 base options in which we are voting, plus the 2001-2010 base option that Buddy was asking for. The lower bases values are on the new options, that is, the "Provisory Standard Normal Period 1991-2016" Base and the 2001-2010 base.

[Juan C. García]

The effect of choosing a base with lower values, can be seen in different ways. By example, we can compare the September 2012 and 2016 ASI Volume, against the different bases:
______________________________________________________
                                                2012         2016
Sep. Vol. (1,000 km³)             3.787        4.529
______________________________________________________
Base        Avg. Sep. Vol.
                (1,000 km³)     ASI Lost 2012     ASI Lost 2016

Year 1979     16.911               77.6%               73.2%
1980-1989   14.824               74.5%               69.4%
1979-2000   13.914               72.8%               67.4%
1981-2010   12.005               68.5%               62.3%
1979-2015   11.350               66.6%               60.1%
1991-2016     9.443               59.9%               52.0%
2001-2010     8.728               56.6%               48.1%
______________________________________________________

When the base includes the 2000-2016 years, it will have an average with lower value. As a result, it will show that there is less Arctic Sea Ice Volume lost.

From my point of view, the 1979-2000 base reflects the reality better than the 1981-2010 or the 1991-2016 options.

P.D. Remember than on this poll, you can vote for 2 options (not just one).

[kiwichick16]

the only valid period offered is 1981-2010......as that is the only 30 year period on offer

....otherwise I would have suggested  1979-2008

as I'm sure most people know  30 years is the accepted time period for validating changes in climate

[Buddy]

This is the graph of the 6 base options in which we are voting

Sweet graph.   Visually....I love that.  For me....it's nice to see where those two "ten year bands" are (the old, high value band of 1980 - 1989 when we had more ice.....and the newer, low level ice band of 2000 - 2009 when we had less ice).  That gives me a "visual" of reasonable values....and any year can be quickly compared to either the old band or the newer band.

[P-maker]

Buddy:

”Sweet graph.”

– I agree!

Nice to see three pairs or buddies going down. I miss an independent “buddy” for 1979 at the top, and I miss a new challenge at the bottom (2016?). I am sure 2017 sea ice volume will lick the bottom line all the way to extinction.

Suggest a revised colour scale going from ice –blue in 1979 (+ earlier buddy) through hot-red in 2016 (+ upcoming buddy).

[Juan C. García]

NSIDC don't use a baseline for their monthly values. They just stick to Novembers in November, Decembers in December ... Seems like an appropriate method to me.

PIOMAS has the problem that the seasonal pattern has changed, so their anomaly plot has become misleading. For showing how the seasonal pattern has changed, decadal averages work pretty well.

Hi, Richard Rathbone.
I excuse myself because I am not answering your comment until now.
In my original comment, I mention:

From my opinion, it is important to discuss which base does NSIDC, IPCC, PIOMAS, etc. choose to compare the monthly values that they make public.

The true is that the IPCC does not make comparisons monthly. They usually use the data of the NSIDC, and the NSIDC does make comparisons to the monthly values, as it does on the daily values (originally with de 1979-2000 base, now with the 1981-2010 base).

By example, on October 2016 they wrote:

Arctic sea ice extent during September 2016 averaged 4.72 million square kilometers (1.82 million square miles), the fifth lowest in the satellite record. Average September extent was 1.09 million square kilometers (421,000 square miles) above the record low set in 2012, and 1.82 million square kilometers (703,000 square miles) below the 1981 to 2010 long-term average.
http://nsidc.org/arcticseaicenews/2016/10/

The difference on base is important. We can have different numbers depending of the values that we use to compare (daily values, 5-day average or the monthly average that NSIDC makes public). Taking the NSIDC monthly averages, the difference on the bases will be the following:

                     Avg                 Avg
Month     1979-2000    1981-2010     Difference
                (10⁶ km²)     (10⁶ km²)        (10³ km²)
Jan             14.95          14.64               307
Feb            15.75          15.46               292
Mar            15.85          15.59               254
Apr            15.09          14.85               249
May           13.66          13.45               206
Jun            12.22          11.96               266
Jul             10.20            9.78               412
Aug             7.73            7.28               451
Sep             7.06            6.54               522
Oct              9.32            8.94               384
Nov           11.35          11.03               324
Dec           13.42          13.12               299

As we can see, the difference can go from 206k km² in May, to 522k km² in September.

The questions we should ask ourselves are, does these differences are related to natural variability of the Arctic Sea Ice? Or these differences are the result of the melting of ASI, caused by the anthropogenic emissions of greenhouse gases?

From my point of view, if the answer is that the cause are the GHG, then we should stay with the 1979-2000 base, because the 1981-2010 takes out around 500k km2 on melting, and then we reduce the reported impact that has already happened on the ASI.

Does anybody have a good reason to keep the 1981-2010 base, other than “tradition” or “because climate should be measure on 30-years intervals”?

Can anyone imagen how will look the following Charctic, if 2016 is compared to the 1979-2000 baseline, instead of the 1981-2010 baseline? Surely, 2016 will be completely under the -2 standard deviation.

[Juan C. García]

Only two more days to vote on this poll. Remember that you can choose two options, not just one!

[crandles]

From my point of view, if the answer is that the cause are the GHG, then we should stay with the 1979-2000 base, because the 1981-2010 takes out around 500k km2 on melting, and then we reduce the reported impact that has already happened on the ASI.

Does anybody have a good reason to keep the 1981-2010 base, other than “tradition” or “because climate should be measure on 30-years intervals”?

30 year average seems about appropriate for general climate because atmosphere is so chaotic. Sea ice seems less chaotic and therefore a 20 year average or maybe even less may be more suitable for sea ice.

However, I think we should note that the purpose of a 30 year average is to be able to detect if there is a change. That is no longer in any serious doubt nor is whether GHGs are the primary cause. So having got to this point, the question is no longer whether there is a change but how and how fast it is changing. For this purpose, showing 10 or even 5 year averages may be more appropriate in order to see if there are patterns e.g. of gaps getting wider to indicate acceleration.

.

Selected 2 most popular options but probably lean towards 79-2000 (or maybe 81-2000). Don't think the comparison period should change each year.

[jai mitchell]

I absolutely agree Crandles.  The default selection of a 30-year baseline for globally averaged temperatures, subject to a +/- 35% residual swing on any given year (extreme variability) DOES NOT mean that a 30, 20 or even a 10-year baseline is necessary for 'all things climate'

The break in the trend can be seen clearly in the mid 1990's and again in the early 2000's, The communication of climate change going forward must NEVER AGAIN fall into the trap where a group, fearful of criticism, opts to include a later baseline because it is more easily defendable.  Since this is the desired results of the actual attacks on climate science, to only consider impacts compared to periods after the observed Global Warming effects have been realized, thereby changing the conceptualization of damage loss estimates and total anthropogenic impact (in the fossil fuel industry world, how much they need to cut back (if at all))

The earliest decade should be the baseline in this case.  The methodology is not significantly different between periods and so the baseline average is a robust value, in any case.

[marcel_g]

I selected 80-89, because I agree with what jai and crandles are saying here about not needing a 30 year baseline and shorter averages are pretty robust, especially in the earlier periods. I think 79-2000 is also a good baseline, because it doesn't have as much change baked into as the longer ones. The changes have been happening so fast that I think a 30 year baseline is somewhat misleading.

[Juan C. García]

Hi, Neven and DungeonMaster:

I feel that there are not yet enough votes on this poll. Also, I want that the members that have already voted, can change their vote if they want so. So, can you change the date that this poll ends, to another 10 days, perhaps? And at the same time, give the chance to change their vote.

Thanks!

[DungeonMaster]

Per author's request, period extended and changes allowed!

I also voted ...for "other choice": ask Tamino!

[nukefix]

I voted other and would prefer:

30-year period 1979-2008

[Pmt111500]

1981-2010 for me. Mark 0,5 and 1 sigma levels so 2007 can easily be seen going way outside it, and better see how far the amounts are from normals in recent years.

[Tealight]

I voted 1981-2010 and other. 1981-2010 because its the baseline NSIDC uses and other because I feel an average of 30 years doesn't reflect the state and change of the sea ice at all. It's a bit like the average speed in a drag race. You could say the car moved at an average of 100mph, but it only spend 0.1s at this speed and most of the time was below and above that number.

On my website I chose the baseline 2007-2016 because its a ten year period where the ice conditions (multiyear to firstyear ice) are relatively constant. We will also never go back to the conditions before 2007 so anything before 2007 is ancient history.

[Juan C. García]

Per author's request, period extended and changes allowed!

I also voted ...for "other choice": ask Tamino!

Thank you for the changes, Dungeon Master.
I am waiting for the December Piomas figures. Seems that we will not have them until monday. 

[Juan C. García]

This is the table with 1979 as the baseline. Of course, the percentage of Arctic Sea Ice Volume (ASIV) lost is the biggest on this baseline.

Some interesting points, with the 1979 baseline:

• There are four months on which 2016 is the lowest on record: April, May, November and December.
• December 2016 is the first December in which we have lost more than half of the ASIV.
• 2016 has become the second year lowest on record, with 46.4% ASIV lost. The first one is still 2012, with 46.7% ASIV lost (whole year average).

On September loses:

• 2012 is the lowest on record, with more than ¾ (77.6%) of ASIV lost. The second is 2011 with 73.5% and the third is 2016 with 73.2%.
• The 2010-16 ASIV lost is striking:  69.7%. If we round the number (as PIOMAS usually does) we have 70% ASIV lost! And it can increase when we have the full decade (2010-19).
• For NSIDC, the second year with September lowest on record is 2007. But the PIOMAS figures tell another story. 2007 is the seventh lowest on record. The first six lowest years on record are: 2012, 2011, 2016, 2010, 2013 and 2015.

PD: We should also be concern about the 2010-16 July to November average: 61.7% ASIV lost (not shown on the graph). From my point of view, this can mean that when we will have an ice-free Arctic, it could happen on more than one month.