If the science is right, why are computer models struggling to get the facts right?

Yes I trust the scientists, yes I agree we should take preventative measures and yes I agree we are warming. What I find difficult to trust is a panel who find it a good idea to play around with data to make things fit. Dishonesty at such a high level will always create a lack of trust in the entire idea. Some say that the report has been watered down while others say it is over the top. Which is it? Why are some scientists back pedalling from the report? If the theory is so rock solid wouldn’t they all be sticking together?

What I know about climate change only scratches the surface of the subject but I’m currently hearing that the information may be flawed. What do I do?

So the Old man replied:

First, I think we need to be clear about what it is we think global climate models do and don’t do, and what they can and can’t do. One of the difficulties is that we often have the impression that, since these are expensive, sophisticated pieces of technology programmed by bona fide geniuses, then they should bl**dy work! More seriously, mainly thanks to the way that the media tends to report the output of model studies and research, we are given the impression that the modellers are presenting clear-cut and precise projections which have a real correlative at a specific point and time. This is not how modellers see their output, and they would probably say that it was wrong to imagine that this is what they are trying to do.

Problems arise when actual events (such as the 2007 Summer sea ice melt) occur which fall outside the projections of the models. This might be because an exceptional combination of factors (with a low combined probability) have occurred, in which case it would not be surprising if the models ‘missed’ it, or because the models themselves (which are simplifications of the real world) did not have the relevant information or programmes to calculate such an event. But this also falls into the category of having unrealistic expectations of what GCMs do and how they work.

Where you have a point is that it does look like the GCMs as a whole are not capturing the rate of change as it is currently (apprently) occurring. I would correct this to say that the current rates of change are at the upper bounds of model projections but (on average) within the bounds of acceptable margins of error.

Your worry seems to be about the IPCC. Is it right to claim that they are ‘playing around with data to make things fit’? The simple answer to this is no. Generally, such claims come from sources which wish to discredit the work of the IPCC, for whatever reason. That science involves reanalysing data and making corrections to previous observations is a good thing, not a bad one, so long as the result is that the new information is more accurate and more reliable than the old information. But some unscrupulous people take advantage of the common misunderstanding that this somehow demonstrates a flaw in the fundamental science or processes, to score political points. I cannot necessarily convince you that the IPCC is not dishonest, but the only claims I have seen to this effect have come from very specific and controversial sources. Given the credentials, experience and expertise of the people involved in putting it together, I am strongly convinced that these people would not knowingly engage in an act of deliberate scientific deception; such a thing would run counter to all of their training and principles and is, given the implied need for hundreds of them to conspire to deceive collecitvely, extremely unlikely.

In answer to your ‘how many’ question: in my research, 17% thought the IPCC was overstating the impact of CO2, 18% thought it was understating it, and 65% thought the report was pretty much bang on. Since there is almost certainly an underlying tendency in most science to be conservative in its conclusions, I would tend towards the view that the IPCC, if it is in error, is likely to be too conservative about its projections, rather than the other way around. Your point at the top about the GCMs apparently not capturing the rate of change would be illustrative of this tendency.

I do not know of any scientists who are distancing themselves from the scientific basis of the report. There are some who have concerns about the third section, in particular, but much of this is because this is less about what the science shows, and more about what we should do about it; as such, it is intrinsically less ‘scientific’ than section 1, and is therefore open to a range of interpretations and opinions which may not have been fully represented, in these scientists opinions.

I should say that ‘the theory’ – by which I presume you mean ‘the AGW theory’ – is not something which is in question in the IPCC’s work, as such. Right at the outset, the summary reports state that this is, in scientific terms, a ‘given’. In my research, not a single scientist claimed to believe that AGW was not happening, though a small number (fewer than 5%) did express the opinion that much of the recent warming is likely to be natural.

I think that covers the main points… 


  Doubting Thomas came back with:

…There are changes happening but I can’t see anything that convinces me that it is not primarily a natural cycle…

To which he got this response, which I will probably get slammed for by real scientists…

Being a weathery sort of person, and thus being aware of the natural variability in both weather and climate, it is no surprise that you are inclined to consider this as a real possibility. You probably know a bit about NWP, too, be honest…

This feeling is so often at the core of various people’s doubts that I think it is worth thinking about. What follows is an exercise in mental visualisation for chronologically advanced. I hope at least some of you are old enough to get the idea…

Rolf Harris.

Long before he recorded ‘Stairway to Heaven’ or held a puppy’s paw while its n*7s were removed, Dear Rolf had his own, quite popular, television variety programme on the BBC. The highlight of each programme came at the end, when, armed with a huge canvas (four metres by three) and a set of decorator’s paint rollers, Rolf would paint a picture live. The first strokes were always big, bold blocks of virulent orange and blue, red and so forth, across, along and down the canvas. ‘Can you see what it is, yet?’, he would ask the audience. No, Rolf, not a clue. The painting would continue, along with some irritating humming and pseudo-aboriginal grunting, a splash here, a dab there; the canvas was getting quite full. ‘Can you tell what it is, yet?’ asks Rolf, knowing that the answer will still be ‘No.’ At the last minute, Rolf would add just a few more lines and dabs in critical places, the subject of the picture becoming suddenly apparent to us: ‘Oh; it’s a kangaroo crossing the Sydney Harbour Bridge! How clever…’ And we, the audience, berate ourselves for not seeing it sooner. And applaud Rolf for his ingenuity.

This is nothing like climate modelling. I’m using it as a metaphor. Bear with me. (Can you tell what I’m getting at, yet?)

When we start doing ‘detection and attribution’ of climate change, we put in the background first; the basic physics of Solar radiation, the basic chemistry of the atmosphere, the proper proportion of ocean to land surface, and so forth. We have nothing which does more than vaguely resemble an idealised and simplified world. Then we add, here and there, the details of interactions and processes, the Coriolis effect, this feature and that feature, until we reach a stage where the background is all in place, but we still can’t really see the picture. All of these are natural components and physical/chemical interactions. (How do we know what goes where? In the case of climate modelling, the background is the known science and the observed historical data.)

At last, every known natural variable is incorporated into the picture, with all that is understood about the various interactions of elements added in. Then, in the case of climate models, we look at the past. If the timeline is sent backwards, from a starting point of, say, 1850, and the model is run, do we get an accurate replication of temperature change since that time? At first, things look good, but by the 20th century, the observed temperature and the model replication are clearly diverging. But all known natural variables have been included; how can this be? We tweak the models, change the assumptions about the physics and so forth, to the limits of possibility. But, however hard we try, the models will not replicate late twentieth century temperature change.

So we add the final dabs and touches, the critical lines and bits of information; the human alterations to the physics and chemistry, the emissions and land use changes, the effects of deforestation and wildfires, power plants and industry, agriculture and urbanisation, and so on. And, Wow! Suddenly the picture becomes clear: we can see the kangaroo! The climate models now replicate historic temperature changes in a pattern which bears a close resemblance (from a distance) to historic events, captured by the data. The detail is a bit fuzzy in places, and the match isn’t perfect, but the divergence of temperature which existed before has now disappeared.

As things stand, no known combination of (physically) possible natural interactions can account for the changes in climate which have happened in modern times. On the other hand, the combination of natural and human interactions, using the known science, can provide a reasonably good match to these changes; good enough for scientists to be satisfied that the change in climate can only be explained by such a combination. Now, C-Bob (and others) have a point: we know that we don’t know everything about this complex system, therefore the possibility always exists that something has been missed, overlooked, underestimated or misunderstood. There is a theoretically possibility that the models are wrong. Knowing this, other scientists look at the various variables and uncertainties, and apply probabilistic formulae to the problem. Then, they express their results in terms of the probability of this being correct. As things stand, both the model replications and the probability of error are both strongly indicating that the answer they have come up with – AGW, in short – is very likely, nearly certain, to be the best and correct answer.

In short; no combination of natural forcings can account for recent climate change. The combination of natural forcings and human forcings can.

Can you see what it is, yet?