by Frank Bosse
Increasing evidence of small aerosol forcing supports the importance of internal variability in explaining inter hemispheric differences in temperature variability.
In the CMIP5 climate models, we find two strong anthropogenic forcings influencing the climate: the warming greenhouse gases (GHG) and the cooling aerosols that reflect sunlight. The difference between the strength of these forcing agents is a key factor when estimating the sensitivity of the real climate system to the GHG, mostly important to carbon dioxide. In the last few weeks several papers have been published that provide important insights for narrowing the scope of the impact of aerosols on the real climate system.
A recent paper by Chung and Soden (thereafter – CS17) the authors examine the difference of the temperature anomalies (and also of precipitation) between the hemispheres of the earth. In the observations there are some shifts since the beginning of the 20th century and from about 1980 on we observe a steady warming trend of the NH versus the SH.
Fig.1: The interhemispheric temperature development in observations (black) and in CMIP5 models ( red, with interquartile range :green and blue). Source: Fig. 5 of CS17
The authors attempt to identify the source of model spread. They focus on the fact that anthropogenic aerosol emissions are mainly in the NH, and that as aerosols have a short lifetime. Hence, most aerosol forcing will also arise in the NH. That NH dominated aerosol forcing can be expected to cause greater cooling in the NH than in the SH, whereas GHG forcing is quite evenly distributed between the NH and SH.
Using climate models, they find that difference of the (NH-SH) temperatures is most likely explainable with a high forcing due to the aerosol-cloud interaction. Their conclusion:
“Models with larger cloud responses to aerosol forcing are found to better reproduce the observed interhemispheric temperature changes and tropical rain belt shifts over the twentieth century, suggesting that aerosol–cloud interactions will play a key role in determining future interhemispheric shifts in climate.”
The aerosol-cloud interaction in models produces a much stronger (negative) aerosol forcing due to additional dimming of the incoming sunlight. The physical hypothesis: clouds have an increased albedo and a longer lifetime when they are influenced by anthropogenic aerosols. A former paper questions if the conclusions of CS17 from the model-research are transferable to the real world. The author asserts in relation to observations in the real climate system:
“The fact that cloud albedo is not significantly larger, or even smaller, in the Northern Hemisphere is an indication that the aerosol is not a first-order factor for cloud properties.”
Two brand new papers shine a new light on the model-real world discrepancy at least in the field of the aerosol-cloud interactions. Mallavalle et al 2017 (for a summary see also this post ) tried to estimate the aerosol-cloud effect in the real world with the help of a volcano-eruption which produced an aerosol impact very similar to expected anthropogenic SO2 sources. An accompanying comment in “Nature” by the well known cloud and aerosol expert Bjorn Stevens captures the point in the title: “Clouds unfazed by haze”.
The hypothesized aerosol-cloud interaction is small. So observed aerosol forcing is substantially smaller than in most models.
Stevens clarifies:
“Until now, however, the biggest surprise has been how hard it is to find compelling physical evidence for strong aerosol forcing.”
He underlines the big problems of some climate models with this fact. They often compensate a high sensitivity versus GHG with a high negative forcing due to aerosol haze, thus enabling them to fit the observed temperature record during the “tuning period” .
One of the main pillars of this “compensating haze” is no longer available. Stevens concludes with:
“Unless this changes, in so far as aerosols are concerned, it seems that there is little to fear from clearing the air.”
That is, we won’t get a strong temperature increase when we globally reduce the air pollution in the future, we don’t have to fear this. The (negative) aerosol-forcing is not big enough.
When we return to CS17: The authors attempt to account for the observed interhemispheric temperature changes by a phenomena that is now known to be small
So, while the findings of CS17 are very likely true in relation to global climate models, their conclusion that “aerosol–cloud interactions will play a key role in determining future interhemispheric shifts in climate” in the real world is completely unjustified.
The approach of CS17 clarifies the problems of many climate models: They have to take into account some hypothetical forcing to explain the real world observations, in this case the record of the interhemispheric temperature difference (IHD). So let’s have a look at another explanation:
Fig.2: The IHD (regressed on the IPCC AR5 forcing data, blue) and the AMO (SST 70W…7W;25…60N, also forcing regressed, red). Both lines are regression residuals with an 11a-loess smoother.
The forcing regressed IHD has a very similar pattern as the AMO, which is a product of the climate systems internal variability.
In the case of the IHD there is also a forcing at work but not the aerosols, the land areas, which are mostly located in the extratropical NH, show more warming (not due to the lower heat capacity vs. the oceans in the SH but due to the lower available moisture over land than over the oceans). Moreover, an important land mass of the SH, the Antarctic, is decoupled from the rest of the climate system and there are also other phenomena involved in the Antarctic leading to a non- or very small warming there.
Accordingly, under every forced warming one would expect the NH to warm faster than the SH.
The rest (the IHD residuals shown in Fig.2) is internal variability. It comes very likely from the over time changing meridional heat transport of the real climate system, not from “red noise” as this current paper shows again.
The CMIP5 models struggle with the internal variability and some of them replace it with a hypothetical, in reality non-existent forcing. This leads to “running hot” ” as soon as aerosol emissions start rising more slowly than GHG concentrations. Even where models do produce AMO-like multidecadal fluctuations during their historical simulations, they are unlikely to be in phase with the real world AMO as it would be necessary when selecting a defined ”tuning period” . This leads to increasing positive deviations of the model-mean when it comes to an estimation of the future temperatures.
Moderation note: as with all guest posts please keep your comments civil and relevant.
