IPCC figure 12.28 baselines?

[EgalSust]

Dear Forum,

(maybe this should be moved to the Stupid Questions -topic?)

As a reaction to the recent interest in the global sea ice area graphic by Wipneus it came to my mind to check what the IPCC said about Antarctic Sea Ice and how they represented the longer term development. I came across this figure from the Working Group One contribution to the Fifth Assesment Report (2013):



It seems to be that the IPCC has just given the observations different baselines than the models. Or am I misunderstanding something? Opinions? Also: if the baselines are indeed different, as it seems, how would a corrected figure look like?

ps here is the caption and source to the report:

Figure 12.28 |
Changes in sea ice extent as simulated by CMIP5 models over the second half of the 20th century and the whole 21st century under RCP2.6, RCP4.5, RCP6.0 and RCP8.5 for (a) Northern Hemisphere February, (b) Northern Hemisphere September, (c) Southern Hemisphere February and (d) Southern Hemisphere September. The solid curves show the multi-model means and the shading denotes the 5 to 95% range of the ensemble. The vertical line marks the end of CMIP5 historical climate change simulations. One ensemble member per model is taken into account in the analysis. Sea ice extent is defined as the total ocean area where sea ice concentration exceeds 15% and is calculated on the original model grids. Changes are relative to the reference period 1986–2005. The number of models available for each RCP is given in the legend. Also plotted (solid green curves) are the satellite data of Comiso and Nishio (2008, updated 2012) over 1979–2012.

http://ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter12_FINAL.pdf
page 1088 (pdf page 60)

[Peter Ellis]

You're misunderstanding.  As the caption says, all the lines are baselined relative to the reference period 1986–2005.  If you look at the green curve (observations) you'll see that the average over that period is zero (i.e. baseline), same as all the other curves.

[EgalSust]

Yes but on top of every image it gives a different numerical value for the averages for observations and models.

[EgalSust]

In other words, in every image the zero-value is different for observations and models.

Like this:

a) NH Feb
Zero for observations: 15.5
Zero for models: 15.9

b) NH Sep
Zero for observations: 7.1
Zero for models: 6.6

c) SH Feb
Zero for observations: 3.3
Zero for models: 3.0

d) SH Sep
Zero for observations: 19.0
Zero for models: 17.8

All zero's are 1986-2005 averages in millions of square kilometers.

[Peter Ellis]

Ah, I see.  Yes, that reflects the fact that the observations and models have a different absolute baseline and so they need to be brought in line with each other to see if they are following the same pattern.

It does not mean either the model or data is wrong, or that one is more accurate than the other.  e.g. microwave ice sensing data is known to miss very thin ice at some times of year, whereas the model includes this (usually however at a coarse resolution in terms of thickness "bands" of ice).