Understanding Climate Risk

Science, policy and decision-making

Sunday Age 10 Questions on Climate Change – the final two

leave a comment »

The Sunday Age – OurSay readers questions on climate change are down to the last two:

  1. ”THE claim ‘the science is settled’ is plainly false due to the many problems with the AGW [anthropogenic global warming] hypothesis (e.g. global temperatures have not risen since 1998 despite rising CO2 levels; alarmism is based on flawed models that do not reflect empirical measurements.)” 
  2. ”Why is the Australian public asked to swallow the ‘carbon dioxide is a dangerous climate-changing pollution’ crap when science shows no observed relationship between global climate and atmospheric carbon dioxide? There is no physical evidence showing a relationship between temperature and CO2, only computer models which give different answers according to whatever assumption data you put in. But there is a very close relationship between temperature and solar activity … Why, when thousands of respected scientists signed a petition saying they don’t agree there is a problem, are we being forced to give up billions in tax dollars to waste on trying to stop carbon dioxide emissions?” 

Michael Bachelard sums up the evidence with a couple of quotes from me. Although it is a long article for a newspaper, a bit more detail won’t hurt. These questions are two perennials that never die, no matter how often they are answered.

The claim “the science is settled” is a phrase that originated with the Merchants of Doubt. They created a straw man that can be beaten up on cue to shout out “but the science isn’t settled – it’s never settled! Galileo!” I describe its history in this post, and William Connelley has more on its origins and use here.

Global temperatures have not risen since 1998 despite rising CO2 levels

BEST decadal land temperature record comparison

BEST decadal land temperature record comparison

Bachelard links to the Berkeley Earth Surface Temperature Project (BEST). This project began by questioning current methods of averaging temperature, only to discover what climatologists have known for decades: quality control and many records produce a reliable average over space and time.

On the right is decadal land surface temperature from BEST and three other records. All use different methods of averaging. The BEST project also uses a separate set of stations, rather than the reference stations relied on by the other groups. No conspiracy there – move along folks.

Bachelard refers to my take on recent warming:

Jones’s theory is that temperatures in 1998 represented a ”step change” in global warming, rather than part of a smooth trend, and that we are now operating in a new, warmer paradigm.

Instead of analysing mean global temperature as a smooth trend I use step and trend analysis. The complex energy interactions in the climate system do not produce a smooth increase in atmospheric warming with superimposed natural climate variability. Variability in the ocean affects the transport of energy from the shallow ocean to the deep ocean, and from the shallow ocean into the atmosphere. Because this transport is nonlinear, we should not expect gradual warming under climate change. The following two charts show step and trend for global warming (land plus ocean) for the GISS and HadCRU data sets.

Step and trend analysis for mean atmospheric global warming anomalies for the GISS (top) and HadCRU (bottom) time series. Note different scales. The thin black line denotes annual anomalies, the dotted line is a quadratic (non-linear) least squares fit, the step changes are measured using the Maronna & Yohai bivariate test and Rodionov STARS test and the dashed lines are intervening linear trends. Significance and non-significance (NS) for the linear trends are given. The step change increases in red are calculated by both tests as the shift in mean between the before and after period - they do not measure the actual shift in that year.

These charts are pretty busy, but show the annual time series, a simple quadratic curve and segmented trends separated by steps. Both time series show the same timing of significant shifts using two tests. Shifts occur in 1930, 1997-1980 and 1997, with each measured change being within 0.03°C of the other for both tests. Pre and post 1997 temperature increases by 0.3°C, a change that occurred in most regions of the world. Without making any claims about the significance of the linear trends (though they are shown), these charts show positive trends both before and after 1997.

By concentrating on temperature since 1998, so-called skeptics conveniently ignore the largest single warming event for decades. In 1997-98 global temperature spiked during the “El Niño of the century”, then never settled all the way back. Using a relatively brief slowdown in warming to disprove AGW is like claiming that the global financial crisis proves that economic growth doesn’t exist.

Two key pieces of research from this year show that brief trends mean little for the CO2 temperature relationship. Santer et al. in Geophysical Research Letters show that at least 17 years worth of record is required to establish a trend. Meehl et al. in Nature Climate Change show that over the past decade most of the energy increase in the earth system was probably going into the deep ocean, and that such periods of hiatus in atmospheric warming are relatively common in climate models.

Alarmism is based on flawed models that do not reflect empirical measurements

The print version of the article contains a chart based on Hansen et al. (2006). The original is shown below right.

Global surface temperature computed for scenarios A, B, and C, compared with two analyses of observational data. The 0.5°C and 1°C temperature levels, relative to 1951–1980, were estimated to be maximum global temperatures in the Holocene and the prior interglacial period, respectively.

Climate models reproduce emission scenarios that change mean global temperature in a realistic manner. They contain most of the major climate features though do not always reproduce them particularly well.

The empirical good/models bad trope is an example of the naturalistic fallacy: no matter what definition of good is proposed, one can always ask “But is that a good definition of what is good?”

The models-are-bad line of reasoning falls over because:

  • Scientific models are tools. Rejecting models as less reliable than observations overlooks the fact that all observations are interpreted using theoretical models of what those observations represent. That is, scientific models and empirical data are both interpreted via “theory”. Rejecting one for the other is the naturalistic fallacy.
  • The preface paradox says that some of the propositions in this book will be fallible. And so with models. The question is whether a model is useful and what it can be used for.
  • This raises the question of model reliability. Model reliability needs to be assessed within a theoretical framework – as science, rather than comparing like with like, which is inductive thinking.

Model success is also measured through explanatory power. A model is successful if it produces results that it was not designed for. If a climate model has the right physical relationships, we should see similar shifts in global temperature to those in the observations. To test that, I’ve chosen a single warming sequence at random from over 75 runs in a directory without looking for a “good” or “likely” result. The MPI Echam model running the A2 emission scenario with observed forcing to 2000 is the one selected. I have not analysed this previously.

Step and trend analysis for the MPI Echam4 A2 Emission scenario Run3 simulation mean global warming anomalies. The black line is annual temperature, and the dashed lines linear trends separated by statistically significant step changes.

The result on the right shows that although the model has less climate variability than in the real world, it produces a series of step changes in warming from 1990, before making the transition into step-and-trend from mid century. Note the slight cooling trend from 2016-2032!

Ok, you clever people reading this recognise that this finding is inductive. It can’t be confirmed until the physical causes of these shifts are understood (I’ve got some ideas). However, the models do produce physically realistic step changes in warming that have not been fully assessed by the modelling community. Part of the problem lies in the analysis and communication of climate output as smooth curves. Smooth curves are useful for asking “How much will climate change?” but it’s not the same as asking “How will the climate change?”

Having used up quite a bit of space on addressing the first question, I won’t grace the second- the science shows no relationship between CO2 and climate – with a response, save to say you can read Michael Bachelard’s answers in The Age. The assertion that models results are based on input assumptions is false. Climate models do not operate on input assumptions beyond the forcing scenarios they use. Their performance depends on how well the physical system is represented.

Many objections to the science based on the temperature record originate because warming is not a smooth upward curve. However, if warming has been represented as a smooth curve, but behaves more like a staircase, then the risks of climate change are being understated rather than overstated.

Edited for clarity, November 7, 16:30


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 )

Twitter picture

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

Facebook photo

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

Connecting to %s

%d bloggers like this: