(Please also see my edit close to the end, since I missed an updated version of the post that corrects for the very thing I complain about here! Well, sort of corrects.)
The Dee Dee Way
I got into an argument with someone recently on NOAA's Facebook page about how we know that CO2 is a greenhouse gas. Long answer short: because physics, and the models we based off of those physics are really, really good at predicting observations. In my search to find a good explanation though, I stumbled across a blog post from Tallbloke, guest-written by one Ed Caryl, a lifelong electronics expert who has had some experience on hurricane- and typhoon-tracking aircraft. So, not quite a climate guy, but a respectable CV nonetheless.
True to his CV, Caryl starts right off the bat by claiming that the expected radiative forcing from CO2 is a fate in the anthropogenic global warming religion analogous to the Christian Hell, and yada yada about sinners. If that wasn't pleasant enough, he goes on to present what he thinks is evidence, using the MODTRAN model held near-and-dear to serious people for its accuracy, that an increase in CO2 will not lead to any increase in radiative forcing. Apparently literally everyone who has ever used this model just missed that!
He shows this evidence through a series of charts, chiefly this one:
What this graph shows is the observed upward radiation as you increase CO2, and as you increase "humidity". Caryl thinks our present-day "humidity" values are at roughly 50% globally, and that the CO2 concentrations are at 400ppmV. This is not a particularly bad assumption (once we figure out what humidity actually means), but we have to keep in mind one very specific detail: that's only at the surface.
What is this humidity variable that Caryl is displaying anyway? Is it absolute, or is it relative humidity? He doesn't specify, but it would seem that—given his description for the remaining parameters in the model runs, he thinks this is relative humidity. And he thinks that when he selected the "Holding Fixed: Relative Humidity" option, the results he got would be for constant relative humidity in the atmosphere.
No Dee Dee! That's not what that button does. What that button does is attempt to scale the vapor pressure by the change in equilibrium vapor pressure as ground temperature offset changes. That's exactly what the Help page for the model says. It's only relevant when that offset is non-zero; but Caryl's offset is at zero!
But it gets so much worse: relative humidity is never constant with height. Never ever ever ever. In fact because there's virtually no water above the tropopause, relative humidity drops off to zero with height. It all precipitates before then.
But it's even WORSE: Caryl apparently doesn't know what the Water Vapor Scale factor is. He apparently thinks it is relative humidity percentage—no Dee Dee! That is certainly not the case. Why on Earth do you think it is called a Scale parameter? Because at 1 it is representative of roughly present-day values, and it proportionally scales the vapor pressure values when you modify the numbers. How do I know this? Because you can actually print out the values from the model itself! There's literally a button called "Show Raw Model Output" to check these things! Shall we see what that button actually does?
My parameter choice (at first) shall be US Standard Atmosphere, Scale Factors of 1 for each of Ozone and Water Vapor, and CO2 at 400ppmV and Methane at 1.8ppmV and Tropospheric Ozone at 30 ppbV, with a Offset Temperature of zero. For my next run, I set the Water Vapor Scale to 2. Here's the pressure profiles for each gas in the model for each run, respectively, and please pardon the enormous graphics but it's so you can all see:
Can anyone see what the pattern here is? Here's a hint: all of the water vapor pressures (fifth column in each table) in the second run are twice those in the first. And what happens to relative humidity when you double the amount of water vapor in the air? You end up doubling the relative humidity (and the output caps all >100% values at 100%). When Caryl points to his curve that says "50% humidity", he's actually pointing to a curve where the relative humidity is 100% for the first 16 kilometers of the atmosphere—in other words, basically the entire troposphere. But that's not the funny part: since he multiplied the water vapor content by 50, the water vapor pressure at the surface is about 300mB. That's a third water. You would drown if the atmosphere was that wet.
The Dexter Way
Now what buttons should we actually press, if we truly want to figure out the change in radiative forcing due to a doubling of CO2? Well, we want to look up (so looking at downwelling radiation). We also want to choose realistic greenhouse gas levels, like the ones I did above; Scale Factors of 1; and a surface temperature offset of zero. We have to realize that the water vapor profile in the atmosphere is already where it should be. You can go check that in the actual output if you'd like: the relative humidity value for a Scale Factor of 1, in that first model run I did, was approximately 50% for the lowest levels of the troposphere.
So when we double CO2? My first run was set at 280ppmV, and the second at 560ppmV, which represent pre-Industrial levels and then double that. The observed downward radiation for the run with less CO2 was 259.144 W/m^2. For the second run, it's at 262.41 W/m^2, which is a difference of 3.27 W/m^2.
This is somewhat less than the value given by Myhre et al. 1998 (and whose figures the IPCC cites), which was 3.7 W/m^2. I am not sure why there would be this distinction; perhaps slightly different water vapor values? Something to do with clouds? But this is a cursory blog post, not an in-depth analysis, so I'm not supposed to draw too much from it. We probably can conclude though that Caryl certainly has drawn quite a lot from quite a wrong series of blind button-mashing, and that his conclusions about MODTRAN and about cooling are not for serious scientists.
Edit: Dee Dee Attempts Dexter
I am rather silly and saw that Tallbloke included a link to an updated post, but I didn't read that in full. It was also pointed out in the comments as well! So this shouldn't actually be new to the people there. But, allow me to set the record a bit straighter, and give Caryl a fair consideration.
In his new post, he redoes the calculations with new Scale Factors ranging between 0 and 0.8. Of course, he still labels these values as "Relative Humidity" values, which is false as can be seen above. But whatever, labels are labels: what should we take of his interpretation? He says:
"As you can see, only in the driest desert on earth will increasing CO2 increase the back radiation very much. Water vapor is by far the strongest greenhouse gas, but it is nearly saturated above 10% humidity. At 10% humidity and above, no amount of CO2 does very much."
sigh No, no that is not true. Since he still does not know what the Scale Factor is, he still does not know what curve in his Figure 6 is actually a 10% relative humidity curve (again, where in the vertical profile is RH ~10%?). A relative humidity of roughly 50% is not a bad estimate, for surface levels; but that's achieved with a Scale Factor of 1, not 0.8 or 0.1.
He also calls this a "CAGW fail", where CAGW stands for "catastrophic" anthropogenic global warming, and why does he say this? Because CO2 doesn't seem to have much of an impact at high relative humidity levels (read: high Scaling Factors), where he says "high" is a Scale Factor of ~0.10 (he thinks it's a RH of 10%). But... did he even look at the numbers? When you double CO2 with a Scale Factor of 8%, as he did for 200-400ppmV, we get downwelling radiation of 219.141 – 213.96 = 5.181 W/m^2. This is a lot more than the forcing that the figures from Myhre predict. Because it's still not a physically-representative atmosphere!
It's rather clear that Caryl didn't actually learn what any of those buttons did, even though it was plainly pointed out to him. It's a rather big shame that such clearly incorrect material was able to get past the review capabilities of a skeptic blog twice in a row. Of course, a shame, but not a surprise.