Update Be sure to come back on Sunday and see my wrap-up column.
A reporter from Nature who could not be brought to understand science was more important than fallacy or politics at least came out of his fog to ask a fellow scientist Kevin Trenberth to critique our “Why models run hot” paper. See this post for the details on this and other climate reporters. Here is Trenberth’s critique as given to me by the reporter (so God only knows if it’s accurate), followed by my response. Please at least read the penultimate response.
Very misleading opening:
IPCC does not make predictions: they have made scenario dependent projections and early projections in FAR were of just unrealistic GHG scenarios. The paper completely misrepresents IPCC in this regard.
Einstein said one should make a model as simple as possible but not simpler. There exist simple models, such as the MAGICC, that deal with a crude ocean as well as the land and response. These have proven useful in IPCC for interpolating global mean temperature between different scenarios. But such models have no hydrological cycle and are toys.
This model is even simpler. It has no recognition of land vs ocean and its distribution, or the atmosphere. It has no clouds or water vapor. It assumes a linear response to a forcing. Many of the past variations in climate are related to Milankovitch changes: the changes in the orbit of the Earth around the sun. There may be no change in radiative forcing but there can still be very large changes in climate: ice ages even. This is because of the distribution of the incoming radiation throughout the year and how it affects ice.
This model does not even handle that case. Yet it is applied over the past 800K years.
The model is then used to play toy games with a justification of some choices from IPCC.
A number of exercises are gone through to select some parameters but it is not easy to see what the tuning is to. Section 7 sets this up and ludicrously concludes there is no warming in the pipeline which is totally at odds with the heat capacity and response of the ocean. It also seems to be assumed for the scenarios (cf 8.3.2).
So there are a lot of “what if” statements without justification. For instance it is concluded in 8.4 that 74% of the warming since 1850 is anthropogenic whereas it seems likely that the value is greater than the observed value, because natural variability has recently suppressed warming at the surface.
It then goes on to take away another 0.6K because that is in the pipeline?
The paper ignores all of the literature related to the recent hiatus in warming related to small effects from missing forcings (mainly volcanoes) and natural variability, especially PDO and consequential burying heat in the ocean. Or that 2014 is warmest on record.
 “IPCC does not make predictions: they have made scenario dependent projections…”
This is false. Scenarios are projections are forecasts are predictions. All projection-slash-forecasts have the same form. They say, given this set of conditions, here is what the future will look like. If the conditions do not hold, then the forecast is not valid. The IPCC basically releases several forecasts, each with different conditions. To judge the efficacy of the forecasts, all we do is look for the conditions that obtained and then measure the forecast’s goodness.
As I’ve pointed out many times, simply saying that next year will be like last year beats the IPCC forecasts. In technical parlance, climate models don’t have (forecast) persistence skill (no matter how well they might fit or backcast past data). And that can only mean that the models on which the IPCC relies are busted, that the science is flawed in some way. The burden of proof is on the IPCC to discover why.
Side note: those who attempt to evade the force of a busted forecast, which logically implies an incorrect model or theory, often attempt refuge behind the “scenario” label. But it is a thin disguise.
 “[The Monckton et al. model] has no recognition of land vs ocean and its distribution, or the atmosphere…Many of the past variations in climate are related to Milankovitch changes: the changes in the orbit of the Earth around the sun…Yet it is applied over the past 800K years…The model is then used to play toy games with a justification of some choices from IPCC.”
Somehow, even though we go to great pains in the paper to admit the model is “irreducibly simple” and does not even pretend to capture all the physics of the climate, that it is simple is given as a criticism. We say, Our model is simple to which Trenberth replies Their model is simple! I’ve stared at this hard, and I guess the only thing I can take from it is that Trenberth agrees with us—Trenberth agrees with us—that our model is simple. Well then.
 “A number of exercises are gone through to select some parameters but it is not easy to see what the tuning is to…ludicrously concludes there is no warming in the pipeline which is totally at odds with the heat capacity and response of the ocean.”
There are two mistakes here, a trivial and an important one. I’ll save the important one because Trenberth makes the same mistake thrice (ocean heat capacity here, later “natural variability”, finally “hiatus”). The trivial mistake is where Trenberth claims he could not “see what the tuning is to” etc., possibly because he did not read the paper carefully, the most charitable explanation. I commend to him Section 5 “How does the model represent different conditions?”, which begins with the words “The simple model has only five tunable parameters…” And to Section 6 “Calibration against climate-sensitivity projections in AR4”, which begins with the words “To establish that the model generates climate sensitivities sufficiently close to IPCC’s value…” And to Section 7 “Calibration against observed temperature change since 1850”, which… Oh, you get the idea.
Trenberth was being lazy.
 “…a lot of ‘what if’ statements without justification…concluded in 8.4 that 74% of the warming since 1850 is anthropogenic…”
The “without justification” quip is more laziness on Trenberth’s part. The whole paper is nothing but justifications about why this or that will happen conditional on our simple model. Trenberth is thus complaining that we use our model to make statements about our model. Subtracting the bluster, what we have here is Trenberth agreeing with us again.
He would have a good complaint were he to say something like, “Their model implies X, but Y is true, therefore I reject their model.” And he would be right in rejecting our model, too. Why, that would be the same criticism scientists like myself make when rejecting IPCC models.
And, lo, Trenberth does try his hand at this excellent rebuttal, as we see next.
 “The paper ignores all of the literature related to the recent hiatus in warming related to small effects from missing forcings (mainly volcanoes) and natural variability, especially PDO and consequential burying heat in the ocean.”
Also recall his “heat capacity…of the ocean” and “natural variability” critiques. What Trenberth thinks is his most damning criticism is instead glaring proof that Trenberth, and many other scientists, have lost their way. I have pointed out, time and again, that you must not say “hiatus” or “natural variability” or anything else like that. “Natural variations” do not and cannot explain the “pause” (I beg you will read the link).
A physicist sets out to model the climate. To do so, he must incorporate whatever physics he thinks are meaningful or probative to why the climate does what it does. Make sense? The physicist then releases a forecast (or “scenario”) conditional on that model. If the observations and forecast do not match, the model is busted. Something is wrong with it. It is not right. It is wrong. It is in error. It is a bad model. I’m not sure how I can be clearer.
Trenberth and pals have released forecasts conditional on various models all of which fail badly. These models did not correctly capture the observations. They are therefore wrong. These models purported to explain the climate, and the climate just is “natural variability“, it just is heat capacity of the oceans, it just is cloudiness, it just is land use, it just is everything the climate is.
If you somehow reject this obviously true proposition, what do you think the climate is that these climatologists have been modeling?
What is happening is Trenberth is blaming the observations for failing to conform to his model. It is reality that is in error, not his theory. This is a special form of insanity encapsulated by the aphorism the love of theory is the root of all evil.
In our paper, we tried to show how a vastly simpler model than the kind Trenberth touts explains temperature better than more complex models. We know—as in know—that the complex models have something wrong with them. We know this because their forecasts do not match observations. What we did was to suggest a plausible explanation why this is so. Hey. We might be wrong. Assume we are. Assume our guess is invalid, our paper worthless, and that instead something else is wrong with Trenberth-style climate models.
Then it is still true that something else is wrong with Trenberth-style climate models! Would we call this, oh I don’t know, a travesty? His models do not suddenly become correct because we made a mistake. What a silly thing that is to imply.
And anyway, our model, simple as it is, sure does look better when compared against reality, no? Not that our model should be used for much except as a clue to climatologists where to look for their mistakes. Mistakes mistakes mistakes.
 “…2014 is warmest on record.”
Oh dear, oh dear. Oh no. Trenberth couldn’t have possibly meant to deceive a poor reporter with a statement he knew to be false. Could he? That would be unethical. And Trenberth is a scientist. Or maybe Trenberth, an expert in his field, just didn’t know that ridiculous claim was false? That also can’t be, because if it was, then he couldn’t be much of an expert. Let’s be nice and call this a prolonged typo.