Pat Michaels Misleads his Readers on IPCC Draft

As a skeptic, one of the more important goals to me in our society is to make sure that pseudoscience and scientific falsehoods are adequately responded to and shut down.  Evolution and cosmogeny are the more commonly known ones in the atheist/free-thought community, because of the very strong creationist push against them, the very strong push to legislate intelligent design into our schools.  For me, even more important, and I lament less popularly told, is accurate science on climate change.

Forbes has taken upon itself to do just the opposite of educating its readers.  The outlet published about a "story" (written by Pat Michaels) regarding the leaked IPCC AR5 second draft, a story started at WattsUpWithThat? (I will not link to that website) claiming that Figure 1.4 of the draft (below) shows that the models from the IPCC have overestimated the warming that we have actually observed.

(Figure 1.4 from IPCC AR5, second draft)

The different colored bars are the IPCC's First Assessment Report (FAR, 1990), Second Assessment Report (SAR, 1995), Third Assessment Report (TAR, 2001), and Fourth Assessment Report (AR4, 2007).  Each report used climate models of different complexity (generally increasing in time, as computing power is wont to do) and a given input scenario to predict temperature increase based on the current physical understanding of the Earth's climate.  AR4, at least, 'hindcasted' the first decade or so using real greenhouse gas/solar/volcanic data (etc.), and the true 'projection' starts in 2000.

In this graph, each projection was shifted to have the same value in 1990.  We can see that the observations since 1990 (more specifically, since mid-2000s) appear to have gone below the projections from each scenario collection - indeed, why?

The uncritical, like Pat Michaels, say that it's because the models are wrong.  Michaels also thinks that the IPCC will remove this Figure, because apparently there was some change made during AR4 to a draft report Figure that he didn't like.

What happened in the drafting of AR4, we probably won't know.  Michaels probably won't either, since he didn't appear to really do much research into the matter.  This pattern also continues into AR5.  We have already been shown, for instance, that the models do match the observations (such as here, here, and here).  We already do know, for instance, that ocean heat content (a much BIGGER number than atmospheric heat content) continues to rise unabated.

(Ocean heat content data from Levitus et al. (2012))

So, that should leave us wondering, what is up with Figure 1.4 from AR5?  Well, maybe we should actually look at the Figure in context.  The leaked draft is available online and to maintain at least a bit of dignity I won't link to it, but if you want to find it you probably can.  For starters, the caption reads:

"Figure 1.4: [PLACEHOLDER FOR FINAL DRAFT: Observational datasets will be updated as soon as they become  available] Estimated changes in the observed globally and annually averaged surface temperature (in °C) since 1990 compared with the range of projections from the previous IPCC assessments. Values are aligned to match the average observed value at 1990. Observed global annual temperature change, relative to 1961–1990, is shown as black squares  (NASA (updated from Hansen et al., 2010; data available at http://data.giss.nasa.gov/gistemp/); NOAA (updated from Smith et al., 2008; data available at http://www.ncdc.noaa.gov/cmb-faq/anomalies.html#grid); and the UK Hadley Centre (Morice et al., 2012; data available at http://www.metoffice.gov.uk/hadobs/hadcrut4/) reanalyses). Whiskers indicate the 90% uncertainty range of the Morice et al. (2012) dataset from measurement and sampling, bias and coverage (see Appendix for methods). The coloured shading shows the projected range of global annual mean near surface temperature change from 1990 to 2015 for models used in FAR (Scenario D and business-as-usual), SAR (IS92c/1.5 and IS92e/4.5), TAR (full range of TAR Figure 9.13(b) based on the GFDL_R15_a and DOE PCM parameter settings), and AR4 (A1B and A1T). The 90% uncertainty estimate due to observational uncertainty and internal variability based on the HadCRUT4 temperature data for 1951-1980 is depicted by the grey shading. Moreover, the publication years of the assessment reports and the scenario design are shown."

This actually isn't that interesting for the discussion, but we get some more details on how the Figure was made and some basic context behind it.  This description is available with the Figure, but the Figure was added to the end of the draft: the draft does not place each Figure where it goes in the report, but has them collected at the end of the PDF.  We need to go a level deeper, straight to the source of the discussion on this Figure, Chapter 1.3.1.  The section has this to say about the Figure (my emphasis):

"Even though the projections from the models were never intended to be predictions over such a short time scale, the observations through 2010 generally fall well within the projections made in all of the past assessments. Note that before TAR the climate models did not include natural forcing, and even in AR4 some models did not have volcanic and solar forcing, and some also did not have aerosols. The projections are all scaled to give the same value for 1990. The scenarios considered for the projections from the earlier reports (FAR, SAR) had a much simpler basis than the SRES scenarios used in the later assessments. In addition, the scenarios were designed to span a broad range of plausible futures, but are not aimed at predicting the most likely outcome. There are several additional points to consider about Figure 1.4: (1) the model projections account for different emissions scenarios but do not fully account for natural variability; (2) the AR4 results for 1990–2000 account for the Mt. Pinatubo volcanic eruption, while the earlier assessments do not; (3) the TAR and AR4 results are based on MAGICC, a simple climate model that attempts to represent the results from more complex models, rather than the actual results from the full three-dimensional climate models; and (4) the bars on the side represent the range of results for the scenarios at the end of the time period and are not error bars. The AR4 model results that include effects of the 1991 Mt. Pinatubo eruption agree better with the observed temperatures than the previous assessments that did not include those effects. Analyses by Rahmstorf et al.(2012; submitted) show that accounting for ENSO events and solar cycle changes would enhance the comparison with the AR4 and earlier projections. In summary, the globally-averaged surface temperatures are well within the uncertainty range of all previous IPCC projections, and generally are in the middle of the scenario ranges. However, natural variability is likely the dominating effect in evaluating these early times in the scenario evaluations as noted by Hawkins and Sutton (2009)."

We see no discussion in Michaels' article about this section, and the reason is clear: it undermines his message.

The scenarios for each model are predictions of what CO2 will be, what aerosols will be (though only some models could handle that as it said), what other greenhouse gases will be, what sun activity may be, so on.  They have nothing to do with the models.  If the input is incorrect, then the output will be incorrect too.  It does not tell you if the model is wrong, it tells you that your scenario is wrong.  The hindcast for the AR4 models matches observations quite nicely, and that should indicate to us at least that the models do a pretty good job of taking accurate input and giving you the Earth's temperature, because they're based on our physical understanding of the Earth's climate.

What the section tells us is that the scenarios did not properly account for natural variability, such as ENSO (El Nino - Southern Oscillation), solar activity, and aerosol radiative forcing.  We do know that these played a large role in the last decade:

• 1998 was a very strong El Nino year, while 2011 and 2012 were La Nina years (El Nino causes surface warming, La Nina causes surface cooling);
• there was a prolonged solar minimum during 2008/2009, so again there is a cooling bias on the end of the time series;
• aerosol pollution (aerosols reflect and scatter incoming sunlight, so they cool the planet) somewhat increased over the past couple years, likely due to China's booming economy, which thrives off of dirty burning of coal.

Kauffman et al. (2011) helps to explain this in more detail.

The scenarios don't include natural variation much, so when we actually do take natural variability into account and remove it from the observations, as Rahmstorf et al. (2012) do, how do the observations look compared to the climate model runs?  Well:

(Figure 1 from Rahmstorf et al. (2012))

Would you imagine that.

Now one thing that Michaels might be right about is that the IPCC could indeed change this graphic before the final draft is released, and it's clear why: people like Michaels are adamant on taking it out of context and lying about it.  But it wouldn't be because Michaels was ever right about, well, anything.

This blog post will herald in my time as an author for the blog, and foreshadow the topic of many of my posts, which will be bringing into the spotlight the pseudoscience surrounding climate change "skepticism."