We’ve had our eye on the tropical troposphere for a long time. More than five years ago we made a video on it explaining that the lack of warming in that part of the atmosphere was a big problem for climate models. While some defenders of the orthodoxy insist that the absence of warming 10 km up in the air above the equator doesn’t matter because nobody lives there and all the warming is at the surface where people do live, that pseudo-defence misses the point. Which is that the models tell us that if greenhouse gases drive the climate there should be warming up there but there isn’t, regardless of whether the readings are taken by satellites or weather balloons. OK, say the faithful remnant, the models are wrong but they’re still right. See, they only over-predict warming in the tropical troposphere because they over-predict warming at the tropical surface. So they have the actual data wrong but the theoretical physics right, because as you go up through the atmospheric layers the warming rate increases as it should, and the only reason they over-predict warming aloft is they overpredict it alow. Which would be feeble even if true because actual data are kind of important to the story. But apparently it isn’t true anyway. According to a new study in Scientific Reports they exaggerate the amplification too. So even if the surface warming were forced to be correct inside the models they’d generate too much warming aloft. And the authors have a theory as to why. Technical “why”. The polemical “why” being, alas, all too obvious.
The author list includes some familiar names if in some cases from a surprising profession: University of Alabama climatologists John Christy and Roy Spencer, but also University of Guelph and Michigan State University economists Ross McKitrick and Timothy Vogelsang. So why economists? Because answering the question requires statistical tools developed in the econometric literature.
To follow the issue requires keeping track of many things that sound alike. For instance tropics (the lowest latitudes) and troposphere (the lowest layer of the atmosphere), especially as here we are concerned with the troposphere over the tropics.
We are also concerned with warming trends at the surface and at every layer of the atmosphere as you go up through the troposphere, in both models and the real world. Models say that the warming aloft should be larger than at the surface by some fixed ratio, generally assumed to be about 1.4:1. So if greenhouse gases cause 0.1 degrees C warming per decade at the surface, and as you go higher the warming gets amplified by a factor of about 1.4, you’d expect to see about 0.14 degrees C per decade in the mid-troposphere, if the surface warming were 0.2C you’d expect 0.28 in the mid-troposphere, if 0.3 then 0.42 and so on… if the models are right.
It has already been shown that they’re not in one key way. Actual observations show less amplification, with a ratio more like 1.1 than 1.4. The question is whether that difference is “statistically significant” or not. The world being a complicated place, mathematicians and especially those in the statistical branch of mathematics have devoted a lot of time and though to the question of how to separate a meaningful difference from random fluctuation, aka “signal” from “noise.” And of course if that difference in ratios is within the range of random fluctuation, the modelers can use the feeble dodge that while they’re wrong they’re not “significantly” wrong and maybe the gap is merely due to “noise” in the complicated climate system.
Oddly enough, considering how old the science of statistical analysis is and how important in many real-world areas like insurance, the technique for figuring out whether two trend ratios are significantly different was only recently developed by Professor Vogelsang. He was working in the econometric field, but as this kind of robust mathematical technique works on data generally not just economic data, he now applies it to the tropical troposphere problem in this study. And the result of his elegant method is anything but pretty.
The authors compared trend ratios (warming aloft to warming at the surface) between observations and models, more specifically from five different observational data sets versus those from 39 different climate models. In the vast majority of cases, individual models show too much amplification of warming at a statistically significant level, and so do the models taken together as a group. So the excuse for excess warming in models, already feeble, has the added disadvantage of being wrong. It’s not the case that the models get the abstract physics of amplification right and merely bungle the real-world surface temperature. They get both wrong.
The authors then discuss why it might be happening, again from a technical point of view. And it’s not simply the case that some models’ Equilibrium Climate Sensitivity, their belief in the absolute amount of warming in degrees caused by any given relative doubling of atmospheric CO2, is too high, although that error is certainly a problem. But here, the study claims, something more specific is going wrong: the models inject too much moisture in the tropical troposphere in response to warming, compared to what happens the real atmosphere.
It’s a revealing flaw because water vapour feedback is a key driver of warming in models. What far too many alarmists and journalists do not grasp about the allegedly “simple” physics of the alleged “greenhouse effect” is that CO2 by itself does not warm very much. H2O is a much stronger greenhouse gas than CO2, and a key assumption rather than demonstration of the orthodox “Enhanced CO2 Effect” hypothesis is that the small amount of warming caused by CO2 directly leads the atmosphere to absorb more H2O thus amplifying the effect. And if it’s not true as a general rule, the whole hypothesis falls apart.
It’s apparently not true in this specific case. And therefore, as an example of the general issue, by allowing too much water vapour to accumulate the models overpredict the tropical troposphere response to a little carbon dioxide-induced warming. So they have the physics wrong as well as the facts. Not a good look.
Of course if the study had shown that only a few models got the amplification rate wrong while the majority got it right, we would chalk it up to minor errors in the weakest models. But in this case, as in the tropical troposphere studies more generally, almost all the models get it wrong, and the only ones that look approximately correct have low sensitivity.
If we confined our attention to those models we’d conclude climate warming is not a big problem and not worth sacrificing our economic future over. Which, in fact, is what we conclude.
