Modeling Unreality

Guest Post by Willis Eschenbach

In the comments to a previous post of mine, Bob Wentworth made an interesting point. He said that it’s not enough to propound my new theory of how the climate works. I also have to show that my theory that emergent climate phenomena control the climate is not included in the current climate models and theory. I’d not thought of that before, and it’s a very valid point.

So I decided to see how the models treat the question of how the oceans, and in particular the tropical oceans, respond to downwelling radiation at the surface. To do this, I looked at the correlation between downwelling radiation and sea surface temperature (SST). If the correlation is positive, it means that when the radiation goes up the temperature goes up. And if the correlation is negative, when the temperature goes up, the radiation actually goes down.

So below are the results from five different climate models, showing the response of the oceans to net downwelling radiation at the surface. The net solar radiation is the downwelling solar (shortwave, or “SW”) radiation less what is reflected upwards from the surface, plus the downwelling infrared radiation (longwave, or “LW”) from the clouds and the atmosphere. This is the so-called “greenhouse radiation”

What’s improperly but inalterably called “greenhouse radiation” starts out as energy absorbed by the atmosphere—absorbed solar energy, sensible and latent heat moved from the surface to the atmosphere, and radiation from the surface of the earth that is absorbed by the gases, including greenhouse gases (GHGs) in the atmosphere—water vapor, CO2, methane, and other minor gases. Once the radiation and the other energy is absorbed, it warms the atmosphere. And since anything that can absorb radiation also can emit radiation, those greenhouse gases radiate the absorbed solar, sensible, latent, and radiated heat in all directions.

This downwelling radiation resulting from the atmosphere is what is known as “greenhouse radiation”.

This “greenhouse radiation”, the downwelling radiation from the atmosphere, leaves the surface warmer than it would be if there were no greenhouse gases—if there were no GHGs, the upwelling surface radiation would go straight to space and be lost. But instead, the upwelling surface radiation is absorbed by the greenhouse gases in the atmosphere about half of it is returned to the surface.

(Important note: the above is all well-established science. The downwelling radiation from both the atmosphere and the clouds has been measured, not modeled or estimated, by thousands of scientists around the planet for decades. There’s an entire network of observing sites around the US called SURFRAD, which as the name implies do nothing but measure the radiation flows, both shortwave radiation (sunshine) and longwave radiation (thermal radiation) to and from the surface. Here’s a typical 24-hour measurement of the downwelling infrared (longwave) “greenhouse radiation” from one of the SURFRAD stations.

Figure 1. Downwelling longwave (thermal) radiation from the atmosphere, AKA “greenhouse radiation”, as measured at the Table Mountain SURFRAD station. SOURCE

Now, as I said, there’s no question that such downwelling radiation from the atmosphere is real and leaves the earth warmer than it would be without GHGs. As a result, I politely invite people who do not think that such downwelling radiation is real or that it leaves the earth warmer to take up that argument anywhere but on this thread. This thread is NOT a place to debate the existence of downwelling radiation from the atmosphere. It is a place to discuss the size and nature of the effect of that radiation. So let me be perfectly clear—I will delete any comments that claim that downwelling radiation from the atmosphere doesn’t exist or that it doesn’t leave the earth warmer than in its absence. I’m more than happy for you to debate the existence of downwelling radiation … but please, do it anywhere but on this thread, thanks. And please, don’t whine like a baby about how I’m the krool science police. It’s not gonna work, I’ll just delete that as well. With the caveats clearly stated, let me return to the discussion.)

So here are the results from 5 different climate models, showing the correlation between downwelling “greenhouse” radiation at the surface, and sea surface temperature.

Figures 2 a-e. Results from five climate models involved in CMIP5, the “Climate Model Intercomparison Project”. All of them have used the same data—”ts”, the surface temperature; “rlds”, longwave downwelling surface radiation; “rsds”, shortwave downwelling surface radiation, and “rsus”, shortwave upwelling surface radiation. The model results are all available from the World Climate Research Program. And there’s a list of the variables here.

There are several things of interest in these model results. First, they vary greatly in the amount of ocean that is negatively correlated with downwelling radiation. The MIROC model at the bottom (e) has almost no ocean with a negative correlation to radiation, while the NorESM model (d) has a much larger area.

Second, the strongest correlation is near the poles, with correlations between 0.8 to nearly 1.0.

Third, the average correlation in the tropics is quite varied—0.22, 0.25, 0.36, 0.45, and 0.45 for the various models. And the same is true about global average correlation.

So with that as prologue, here is the actual reality as determined from different observations.

Figures 3 a-c. Results from comparing CERES satellite-based radiation datasets with the Reynolds Optimally Interpolated (Reynolds OI) sea surface temperature (SST), Berkeley Earth SST, and CERES SST.

Some notes about these. First, despite using different datasets, unlike the models they are very close in all values

Second, the correlation near the poles is much smaller than that shown in all of the models.

Third, all of them show larger amounts of negative correlation in the tropics, as well as globally, than do any of the five models.

There’s another way to look at this same data. This is to look at a scatterplot of the longer-term (a couple of decades) averages of the gridcell surface temperatures versus the corresponding average amount of net surface radiation each gridcell receives. Here, for example, is the CERES radiation data versus the Reynolds SST data.

Figure 4. Scatterplot of temperature of the 43,350 1° latitude by 1° longitude oceanic gridcells versus the net downwelling surface gridcell radiation. The black/red line shows a LOWESS smooth of the data. The slope of the LOWESS smooth shows the change in surface temperature for each additional W/m2 of downwelling radiation. Values are longer-term (two-decade) averages of the gridcell variables.

This represents the long-term relationship between downwelling radiation and ocean temperature. The ocean has had hundreds of years to adjust itself to the average amount of downwelling radiation. Note that in the warmest parts of the ocean, the correlation between the radiation and temperature goes negative—as one goes up the other goes down. This is what we saw in the tropics in Figures 3 a-c.

Now, to compare this to other datasets, it’s not too meaningful to include the 43,350 individual data points. So first, let me compare the LOWESS smooth of the data in Figure 3 with the LOWESS smooths of the corresponding data for the other two observational sea surface temperature (SST) datasets, the Berkeley Earth SST data, and the CERES SST data.

Figure 4. LOWESS smooths of scatterplots of net downwelling radiation and sea surface temperatures, observational data.

Other than a small difference near the poles, close to the edge of the sea ice where the water is just above freezing, all three observational datasets are in good agreement. And again, at the warm end of the scale at the right, we see the correlation go negative in all the datasets.

Next, let me compare the LOWESS smooths of the models in the same fashion.

Figure 5. LOWESS smooths of scatterplots of net downwelling radiation and sea surface temperatures, computer model results. Note that only one of them, NorESM, goes negative at the warmest sea surface temperatures.

Again they are similar … but in this case the difference is in the warmest areas. As we saw in Figures 2 above, they differ greatly in the area of the warm ocean where the correlation between radiation and temperature goes negative.

And to close out this part of the discussion, Figure 6 below shows the data in Figure 5, with the observational data from Figure 4 overlaid on the top.

Figure 6. LOWESS smooths of scatterplots of net downwelling radiation and sea surface temperatures, computer model results plus CERES observational data.

Now, my theory about emergent climate phenomena says that at the warmest ocean temperatures, the action of thunderstorms will strongly cool the sea surface … as we see in the observational plots above.

Here’s further evidence that the thunderstorms strongly cool the surface at the highest temperatures. Figure 7 below shows the “net cloud radiative effect” (CRE). Clouds cool the surface by blocking the sun. They also warm the surface by absorbing upwelling longwave from the surface, about half of which is radiated back to the surface. The “net cloud radiative effect” is the sum of the warming and the cooling effects. Here’s the map of the surface net cloud radiative effect.

Figure 7. CERES surface net cloud radiative effect versus Reynolds sea surface temperature.

Note the great strength of cooling at warm sea surface temperatures, up to as much as ~ 70 W/m2 of cooling in certain locations. This is about half of the global ~ 160 W/m2 of average solar energy at the surface. This is what causes the negative correlation between radiation and temperature in the warmest parts of the ocean.

My theory also says that the increase in sea surface temperatures will be slower than it would be otherwise, due to the action of a variety of emergent phenomena acting to cool the surface. And we see that as well in Figure 5 above.

So … does this establish that my theory about emergent climate phenomena is true?

Nope. It’s more support, but its far from establishing it.

However, it does strongly suggest that emergent climate phenomena are not realistically included in the climate models.

Finally, Figures 2 a-e of this analysis reveals the huge differences between just these five climate models … so next time someone says the models are “physics-based”, you’ll know that they’re talking Hollywood.

What do I mean by “talking Hollywood”?

Well, it’s like when Hollywood says a movie is “based on a true story” …

My best wishes to all, stay well in these most curious times …

w.

Post Scriptum: I must confess that I am quite baffled by how mainstream climate scientists handle the whole subject of climate models. Clearly, as shown in Figures 2 a-e above, certain models are fairly close to at least some aspects of reality, while others are very far from reality. For example, the MIROC model shown in Figure 2 (e) is clearly missing some very important aspects of oceanic behavior, while the Norwegian model NorESM Figure 3 (d) gives much more realistic results.

But all of that gets ignored by the mainstream scientists. All of the results of the different climate models are given equal weight, the group is called an “ensemble”, and a simple average of all of their output is taken to be a valid result … say what?

If I ran the zoo, I would get the modelers together and devise some simple tests, something akin to the graphs and regional measurements in Figures 2 a-e above, but covering many other aspects of the real climate system. I would have a competition wherein we could evaluate and rank all of the models based on how well they passed those tests. Not only that, but I would use the tests to examine and elucidate the reasons why some models do so much better than others.

I would also use things like Figure 6 above to work to understand why all of the models that I tested are in one tight bunch, and all the observational datasets are in another tight bunch … what are the models missing?

I have no explanation for why the modelers deal with the models in this curious hands-off “everyone is equal” manner. However, as many folks are more than happy to remind me, I’m merely a fool without any credentials, just three lifetimes or so of personal experience at solving real-world problems … so I’m clearly unqualified to opine on really complex sciency things like climate models.

It does, however, remind me of modern education, where people want to get rid of the SAT and other tests and even get rid of grades so all the students can feel good about themselves.

Similarly, it appears that the scientists just want to give every model a “Participation Prize” so they don’t damage any of the modelers’ precious self-esteem … and sadly, this is what passes for “science” in the climate world.

My Usual Request: QUOTE THE EXACT WORDS THAT YOU ARE DISCUSSING! I can’t tell you how many times folks have twisted, misrepresented, or spun my words and then attacked me regarding their fantasy of what I said. Misunderstandings are the bane of the intarwebs.