Sympathy for the climate models?

There was a comment on one of my previous posts suggesting that I should stop being “sympathetic” towards climate models since, without criticism, climate scientists will never put any effort into correcting the problems with their models. The suggestion being that they could do better, but it suits them that their models predict higher global surface temperatures than is observed.

So, am I sympathetic to the climate models? I guess I should acknowledge that I probably am. One reason is simply that a lot of what I do involves complex modelling (although not as complex as climate modelling) and so I have some sense of how complicated these models can be. It seems highly unlikely that they’re getting it wrong simply because it suits them to do so. Another reason is related. If a large number of well-trained, experienced scientists think that these models have some merit (despite the known problems) then they’re probably right. The chance that some educated lay people know better than those who actually work with these models is probably small. This doesn’t mean that the scientists are defintely right, simply that it is unlikely that they’re wrong.

However, none of the reasons above is actually a particularly scientifically credible reason for being sympathetic towards climate models. I do, however, have a more justifiable reason for thinking these models haven’t actually failed, as many skeptics would have us believe. The main reason for believing these models have failed is that they’re predicting higher global surface temperatures than has been observed. The basic claim being that the prediction is wrong, therefore the model has failed.

These, however, are not simply models for predicting global surface temperatures, they’re climate models. They do much more than simply predict global surface temperatures. If I understand the basics (and I’m happy to be corrected if I don’t) what they’re actually doing is modelling the transfer of energy through the climate system. One of the fundamental reasons why the evidence for global warming is strong is that there is a measured energy imbalance at the top of the atmosphere. This imbalance is measured to be between 0.5 and 1 Wm-2. It’s my understanding that climate models produce top of the atmosphere energy imbalances that are consistent with these observations.

If so, then climate models are actually get the fundamentals correct. They are correctly modelling the increase of energy in the climate system. So, how significant is it that recently they’ve been overestimating global surface temperatures. Well, it’s clear that about 90% of the excess energy goes into the oceans, leaving 10% or so to heat the land and atmosphere. I don’t know precisely what fraction actually heats the surface, but lets assume a few percent. It’s clear, therefore, that a small error in the amount of energy going into the oceans can have a big impact on the estimated rise in the surface temperatures. To assume that a model has failed simply because something that only involves a few percent of the energy being modelled is maybe not quite consistent with observations, seems a little extreme; especially as the models are probably correctly modelling the evolution of the total energy.

There is, however, a little more to this. There are two parameters that are typically of interest. One is the Transient Climate Response (TCR) and the other is the Equilibrium Climate Sensitivity (ECS). The TCR is the change in surface temperature at the instant at which CO2 has doubled. The ECS is the change in surface temperature required to return the system to equilibrium after CO2 has doubled. It is a little more complicated than this, but this is essentially what these parameters are.

Since climate models are currently over-prediciting the rise in global surface temperature, this could indicate that the TCR is lower than previously expected. There is, in fact, some evidence that this may be the case, but it is still uncertain. However, if climate models are correctly modelling the change in total energy in the climate system, then their estimates of the ECS is likely to still be quite accurate. This could mean that we have more time (i.e., the TCR is lower than expected) but the ultimate rise in surface temperature will be as climate models have predicted. I don’t know about you, but this doesn’t really make me feel any less concerned.

So, that’s the real reason why I am sympathetic towards climate models. They may currently be over-predicting the rise in global surface temperatures, but there’s no evidence that they’re not correctly modelling the evolution of the total energy in the system (feel free to correct me if this is wrong). At best, this means we have more time than we previously thought, but – despite this – the end result will still be the same.

Advertisements
This entry was posted in Climate change, Global warming and tagged , , , , , , , , , . Bookmark the permalink.

12 Responses to Sympathy for the climate models?

  1. Sou says:

    Good points there. I’m not convinced the models are overestimating surface temperature. Bear in mind that the surface temperature is still within the model envelope. Not only that but we need to wait a bit longer before we can jump in and say the modelled surface temperature is “wrong”. Weather is happening in the meantime. Sometimes the surface temperature meanders and sometimes it races. It’s been meandering the last few years but it hasn’t been dropping despite a quiet sun and various La Ninas. The hottest year on record was only a couple of years ago in 2010.

    I realise the UK Met has revised down a tad its projections for the next five years but IMO anything can happen. What would be a concern would be if it meandered a couple more years then shot up swiftly like in the 1990s. Will we cope? What will it do to extreme weather? That’s bad enough at the moment and we’re supposed to be in a ‘flat time’. I think we’ll suffer quite a bit the next time the surface temperature races up.

  2. A good point. I should have mentioned that. As you say, the measured surface temperatures are indeed within the range of most models and so we can’t really say that the models have got it wrong. Even I’ve been caught out by the misinformation presented by those who claim to be “skeptics” 🙂

  3. reasonablemadness says:

    I don’t think, that one can even tell so far, that the models are over-predicting current temperatures or that one can say that global warming – even on the earth surface – has slowed down or stalled. That is just one of the deniers talking points, because it is easy to oversell something, that is absolutely nothing special.

    They sell the idea, that the temperature curve is leveling of, that it has stalled and that there is no warming since 15 years. And if you show the temperature curve with a certain, cherry-picked rolling mean interval, it seems at the first glance – and especially to nearly everybody that is not familiar with climate time series analysis – that the temperature stalled or declined and that might be evidence, that global warming is over:

    http://woodfortrees.org/plot/hadcrut3vgl/from:1982/to:2012/mean:145/plot/hadcrut3vgl/from:1987/to:2004/trend

    This idea sells, because you can see the decline with your own eyes quite easily, and a picture can tell more than a thousand word. And once that picture is in your head together with a misleading comment, it is quiet difficult to get that picture out of the head again.

    Because what people like Anthony Watts or Lucia don’t show you and don’t tell you (and what most followers in the denialosphere don’t know, because they don’t hear it from them) is the following:

    This “decline” is nothing special. The temperature curve is full of nearly identical situations in the past *during* the current warming, but the warming did not stop.

    I assume most people have seen the temperature curve of the recent warming from the 1970s onward till today, but the current flattening sticks into the eye. So, why is that, when there are other parts of the curve that are identical? I assume it has psychologically reasons, because the other “declining” parts are in the middle of the curve, followed by further warming, so you see these declines in context and it appears therefore just as a minor wiggle in an overall upward graph. But for the current “decline” the graph has not yet a further warming that can be shown. So your eyes only see the flattening at the end, which makes it look more pronounced and important and it easy to fool oneself into thinking, that this flattening may go further or temperatures will even starting to decline.

    But if you look at the other similar parts of the current warming in the same way, it is nothing special anymore:

    Decline in the late 1970s:
    http://woodfortrees.org/plot/hadcrut3vgl/from:1964/to:1987/mean:160

    Decline in the 1980s:
    http://woodfortrees.org/plot/hadcrut3vgl/from:1973.6/to:1994.1/mean:160

    Decline in the mid 1990s:
    http://woodfortrees.org/plot/gistemp/from:1984/to:2001.2/mean:120

    The current decline since the mid 2000s:
    http://woodfortrees.org/plot/hadcrut3vgl/from:1986/to:2012/mean:145

    Or if you look a them all in one picture, you would have trouble to tell which time period is which:

    So the whole idea, that global warming stopped or model projections are over-predicting, is IMHO nonsense, as we have several examples in the past, where it was nearly exactly the same situation and the warming didn’t stop there either.

    One can see that also, if he looks at the trends over different time spans:

    The 15-year time span from 1980-1995 shows a temperature trend of 0,7°C per century.
    The 15-year time span from 1997-2012 shows a temperature trend of 0,9°C per century.

    So even when looking at temperature trends, the current period shows nothing exceptional and similar things happened in the 1990s and the warming didn’t stop there either.

    It is also clear, that this whole “it didn’t warm since 15 years” scheme is nonsense, when one looks at the uncertainties in the trends. All 15-year time periods since 1950 have an uncertainty of +/- 1,5-1,7°C per century. The temperature trend from 1970-2000 (which was even in denier eyes a warming period) is also only about 1,6°C per century. So the uncertainty for 15 year time periods is about as big as the noise in the data aka short-term variability (El Nino, etc).
    So it is no wonder, that with so much noise a 15-year time period may show all sorts of trends between slight cooling and intense warming. You cant therefore draw reliable conclusions from such a small time period. It just doesn’t make sense. Only when looking at longer timescales the real temperature trend will overcome the noise.
    And as long as the system is not in balance (which it is not, because we can observe the imbalance), there is simply no reason to believe, that the current “pause” is anything more than all the other “pauses” which we can see in the temperature record. The most probable thing to expect is, that the warming goes on as before and not that some magic ocean cycle that nobody has any proof of is the cause of all that and that the earth will now cool again.

  4. Indeed, I agree. A fair amount of what I’ve written before has made this case. What motivated this post was partly a comment by someone that I’m sympathetic to the models and also comments by Dana Nuccitelli (with which I agree) that really the oceans dominate and so fixating on surface temperatures is unfounded. So what I was trying to get across was that even if there is a discrepancy, it’s in something that is only influenced by a few percent of the energy that these models are considering and that it can change substantially if something like the energy going into the oceans was to change very slightly (in a relative sense).

    However, I do agree with you completely that looking at short time intervals when considering the temperature anomaly data is silly and that there is nothing especially surprising about the evolution of the temperature anomaly data at the moment.

  5. Gavin's Pussycat says:

    Actually they are not even just climate models, they are general circulation models. The basic machinery is the same as in numerical weather models, and those perform pretty nicely. Well enough to make folks come back for more…

    There is so much more about climate than global mean temperatures. There’s the diurnal cycle, the cycle of the seasons, snow on mountain tops, storms, ENSO, the jet streams, climatic zones, Hadley cells, the intertropical convergence belt… looking at a model run visualization, you couldn’t tell that it isn’t satellite imagery! That, my friends, is called success.

    …and anyway, you don’t have to be perfect, just better than the competition 🙂

  6. Well, I suspected my description may well have been a little simplistic. You seem, at least, to have confirmed my general view they are remarkably complex and to assume that they’ve failed because one – possibly minor – aspect appears not to be matching observations (for some reasonably short period in a climate sense) is somewhat extreme 🙂

  7. ECS excludes slow feedbacks which are difficult to model. ESS gives total end warming which, according to paleoclimate, is likely to be double ECS.

    The models also have a lot of trouble managing the energy transfer to ice sheets, which may be one more reason they have overshot short term warming. The ice sheet response has also been far faster than expected.

    The models will improve as a more complete data set becomes available. And I think they’ll probably be a lot more exact in 10 to 20 years time.

  8. reasonablemadness says:

    Not only has the ice sheet response been much faster than the models have shown, also sea ice has decreased a lot more, than what was predicted by the models. The heat needed for melting all that ice is the real world must in the models therefore have gone somewhere else, e.g. it might at least partially have gone into rising surface temperatures in the model runs. But as I said before, the current “pause” is not even special in context of the previous temperature curve since the 1970s. It could also just be that more heat is going at the moment into the ocean, than in years which showed stronger surface warming.

    And surely you are right that models in 10 to 20 years time will be much more exact than current models. But we can’t wait that long and do nothing till then. Every year that passes by, either the consequences in the future will be getting worse, or we will have to cut back our emissions much faster and harder in the future to make up for the lost time. The best thing to do would therefore be, to act now, so that we can transition into a sustainable energy system on a much smoother path. In fact, the best thing were, we would have started that 10 years ago or at least when the AR4 was published.

  9. Gavin's Pussycat says:

    As I said, you only have to do better than the competition… which is dead easy, as nobody has even tried to build physics-based models behaving ‘denialistically’, showing little or no global warming when being hit with the CO2 sledge. If anyone has ever tried, I figure they gave up when noticing how utterly weirdly such a model would behave — and not just where global mean temps are concerned. You couldn’t even plug it into a NWP code and expect it to perform serviceably.

  10. Oh I entirely agree that we don’t have enough time to perfect the models before acting. Paleoclimate makes it pretty clear that ESS is at least double ECS and the human forcing is much more rapid than anything yet experienced. So ESS is likely even greater for AGW than even paleoclimate would indicate.

  11. Gavin's Pussycat says:

    And this sums it up

  12. Yes, I saw that. Very good.

Comments are closed.