## Comments on S c h w a r t z et al. (2010)

Posted by Ari Jokimäki on January 24, 2010

**UPDATE**: *I misunderstood some points about this paper and made some false comments about it here. Therefore I wrote a new version. Use the new version because all the information in this version is not correct.*

Newly published paper by Schwartz *et al.* (2010) (abstract) has been claimed to show that theory of AGW is false or that “global warming has been cancelled”, etc. The claims are based on this statement in the paper’s introduction:

However, the observed increase of GMST over the industrial period is less than 40% of what would be expected from present best estimates of Earth’s climate sensitivity and the forcing (imposed change in energy balance, W m

^{-2}) by the observed increases in GHGs.

*(GMST = global mean surface temperature, GHG = greenhouse gases).*

In other words, they determine expected temperature rise from greenhouse gas forcing and the climate sensitivity, then look at the observed temperature rise and compare the two. Not surprisingly, they found that the two are different. I said “not surprisingly” because they only looked at greenhouse gas forcing and I know that there are other forcings at play and I’m also quite sure that some of those forcings work in different direction than greenhouse gases, aerosols for example. Just a simple example of the situation would be that if GHG’s would cause a warming of 2K and aerosols would cause cooling of 1K, then the resulting warming from these two would be 1K meaning that the observed warming would be only 50 % of the expected warming from GHG’s.

So, the 40 % number they give doesn’t represent the total overall situation, but it only represents the situation if only greenhouse gases are considered and the rest forcings are ignored. Now, Schwartz *et al.* know this because it is the subject of their paper to study what causes the difference, so Schwartz *et al.* are not claiming that observed temperature is less than 40 % of the expected all-forcing-temperature. Yet, it is the 40 % number that is the one they are now parroting all over the Internet as if it would represent the total overall situation.

It would be the same as if I would calculate that aerosols in the air would cause cooling of certain amount and then I would note that global temperature has been rising instead of expected cooling from aerosols. I would then say that I will now consider why there is such a difference but somebody else would just quote me on the observed temperatures not showing the expected cooling and would then spread that word as a proof that the theory of aerosols has been now shown wrong.

Well, at this point we are only in the introduction section of the Schwartz *et al.* and we already have handled most of the false claims circulating in the Internet about this. But Schwartz *et al.* do have things to say even beyond the introduction.

### Rest of the paper

Schwartz *et al.* are studying if the difference between the observed and expected greenhouse gas warming is due four main things:

– Natural variation in global temperature.

– Lack of attainment of equilibrium.

– Overestimate of climate sensitivity.

– Countervailing forcings over the industrial period.

They calculated that the expected warming from GHG’s would have been 2.1 K. They said that the observed temperature increase had been 0.8 K. That means that they are looking to find 2.1 K – 0.8 K = 1.3 K of cooling from the above mentioned four things.

**Natural variation in global temperature** can cause up to 0.2 K of cooling according to them. This is how they found it out:

We use variation in preindustrial global temperature as inferred from proxy records, mainly tree rings, ice cores, corals, and varved sediments to estimate the likely magnitude of any natural cooling over the 150-year interval of the instrumental record.

Proxies? Tree-rings??? Surely any self-respecting climate denier at last now will dump this paper as a heretic production. Well, seriously, I think that’s reasonable approach to get a rough idea. However, it’s also bad news for those who think that the global warming is from natural variability. According to Schwartz *et al.* observed = 0.8 K and natural variability = 0.2 K. That means the observed warming is 400 % of the expected maximum warming from natural variability – a worse result than the observed versus expected from GHG’s.

Note that natural variability can work for both directions, it can cause cooling or warming.

**Lack of attainment of equilibrium** is a fancy way of saying that there might be delays in the climate system so that not all the warming from GHG’s has yet been realised in surface temperature but is instead hiding somewhere. Ocean is the most obvious and important place to hide the warming from GHG’s. They determine that 0.37 W/(m^{2}) of the forcing could be hiding in the ocean, and they say that it corresponds to 22 % of the warming discrepancy, which would give about 0.5 K of cooling (I might have misinterpreted that though, they don’t express it very clearly).

Note that this effect works only to one direction, it has a cooling effect on global surface temperature.

**Overestimate of climate sensitivity** suggests that the climate sensitivity would be lower than the expected range. That would explain the discrepancy. They note that IPCC limit for very “unlikely” climate sensitivity is 1.5 K and they say that the observed warming would require the climate sensitivity to be even lower than that. That, however ignores the other factors causing the cooling mentioned above. The situation is presented in their Figure 2. There they present the observed warming as a horizontal line and they have added the natural variability as a horizontal band around the observed line. The expected warming from GHG’s is presented as an increasing line. One can see that when accounting for natural variability, the expected warming goes out of the band at climate sensitivity of about 1.7 K. That already is within IPCC very unlikely limits, and approaching the “likely” limit of 2.0 K.

However, they haven’t included the “Lack of attainment of equilibrium” value of 0.5 K discussed above. If we would include that, we would get a possible climate sensitivity of 2.2 K, well within the IPCC “likely” limits. This wouldn’t even include the aerosol forcing, which is likely to be substantially negative. With aerosol forcing of the size IPCC has determined to be the best estimate we would get even higher possible climate sensitivity (one that would agree quite well with IPCC limits), approaching 4 K. I have reconstructed some relevant parts of their Figure 2 and I have added the 0.5 K lines there as well. See the Figure 1 below.

*Figure 1. Reconstruction of the relevant parts of Schwartz et al. Figure 2 – the warming of Earth’s surface (X-axis) as a function of climate sensitivity (Y-axis). Expected increase of global mean surface temperature for GHG’s only (black), expected increase of global mean surface temperature for GHG’s and aerosol’s based on IPCC’s best estimate (green), observed increase of global mean surface temperature (blue thick line) and the possible effect of natural variability to that (blue thin lines), and observed increase of global mean surface temperature when ocean thermal sink has been accounted for (red thick line) and the possible effect of natural variability to that (red thin lines).*

**Countervailing forcings over the industrial period** also have an effect to the global temperature. Aerosol forcing we already discussed briefly above and it is the only forcing they are discussing here. Here too they discuss Figure 2 in a manner that is ignoring other factors. They say that with the IPCC best estimate aerosol forcing the warming *“would be compatible with the lower end of the IPCC “likely” range of climate sensitivity”*, but actually if we consider the natural variability and the warming wasted to the ocean, we can see from their Figure 2 that resulting climate sensitivity could easily be 4 K. Here are my estimates for the climate sensitivity (values in Kelvins) based on their reconstructed Figure 2 presented above as Figure 1:

GHG GHG + aero Observed 1.3 2.2 Obs & natural 1.0 - 1.6 1.7 - 2.7 Obs + ocean 2.1 3.5 Obs + ocean & nat 1.8 - 2.4 3.0 - 4.0

Schwartz *et al.* do make an important point about aerosol forcing, the fact that it has large uncertainty. But I’m not quite sure that’s exactly a new finding.

So, at this point it seems I’m disagreeing with them a little. In my opinion they are stressing the low end of their results and not considering the high end much. In fact the warming that goes to the ocean is quite certain component, so they definitely should have considered that in their Figure 2.

They then enter to a discussion about the methods of determining the climate sensitivity and possible actions for improving the aerosol forcing uncertainty.

### Conclusion

Claims in the Internet about Schwartz *et al.* are largely based on misunderstanding and not reading the paper beyond the abstract and/or introduction chapter. However, there is an apparent actual point in Schwartz *et al.* that other factors contributing to the difference of observed and expected warming are not enough suggesting that we have some forcings wrong or that climate sensitivity is somewhat smaller than we have thought.

To me it seems that Schwartz *et al.* are mistaken and their point seems to rise from the fact that they didn’t consider ocean thermal lag when they determined the whole situation. They considered the ocean thermal lag separately but did not include it to their discussion of Figure 2 describing the overall situation. When the ocean thermal lag is included, the results seem to agree well with the IPCC values and the best estimate of the climate sensitivity would be 3-4 K.

I’m also little disappointed of the lack of references to the preceeding studies on the matter. For example, Lean & Rind (2008) determined the relative sizes on forcings, finding no such problems as Schwartz *et al.* are suggesting.

**UPDATE** (January 25, 2010):

I’ll add one note. As I have been making my paperlists, I have read a lot of introduction sections of papers because there the existing research on the subject in question is given and also the references to the key papers on the subject. I was quite amazed when I had read the introduction section of this Schwartz *et al.* paper. There isn’t a single reference to peer-reviewed papers, but they only reference IPCC 4th assessment report once. I don’t recall seeing any other papers with so poor introduction section. Also note that Schwartz had problems before with the ocean’s role in his 2007 paper, see the comment section below (thanks to Paul Middents for pointing that out).

## Paul Middents said

Ari,

Very good summary and comments.

Didn’t Schwarz have some difficulty in a previous paper dealing with ocean related time constants?

Paul Middents

## Ari Jokimäki said

Yes, there seemed to be some problems with that in Schwartz (2007), here’s Schwartz’s reply to the critics:

Good point, thanks. It puts this new one to even stranger light. I added a note of it above with another note on the introduction section of this new one.

## AJ said

Ari,

That’s a nice discussion about this paper. While I agree with you on that parts of this paper (and its press release) is written in a little bit of a strange way, that encourages false interpretations of the actual content of the paper, I may have to disagree with you on that they wouldn’t have properly included the “lack of equilibrium” in their calculations.

They say the deviation from equilibrium is about 0.37 W/m2, which would correspond to some amount of cooling. In your calculations, you said it’s 0.5 K. However, the amount of cooling that this 0.37 W/m2 corresponds to, actually depends on climate sensitivity. If you’re saying it’s 0.5 K, then climate sensitivity should be 0.5 K / 0.37 W/m^2 = 1.3 K / (W/m2) (or 5 K per CO2 doubling) for this to be true.

The way that Schwartz et al. take this into account is by defining this effective forcing, which is the actual forcing minus this “lack of equilibrium”, and the climate sensitivity is then the observed temperature change divided by the effective forcing. In this way, the “lack of equilibrium” is taken in the units where it’s measured, i.e. in W/m^2, without the need to make an assumption of how many degrees this corresponds to. Therefore, I think this way is more consistent. If you want, it’s kind of equivalent to your way, if you let this amount of warming to which the 0.37 W/m2 corresponds to, depend on your result for the climate sensitivity.

For the numbers, Schwartz et al. give the best estimates:

Warming 0.8 K

GHG forcing 2.6 W/m2

Tropospheric ozone forcing 0.35 W/m2

Aerosol forcing -1.2 W/m2

Combined with the heating rate of 0.37 W/m2, these give (2.6+0.35-1.2-0.37) W/m2 = 1.38 W/m2 for the “effective” forcing.

With the observed warming, this gives 0.8 K / 1.38 W/m2 = 0.58 K/(W/m2) for the best estimate of climate sensitivity. Multiplying by 3.7 W/m2 per CO2 doubling, one finds that this is equivalent to 2.15 K per CO2 doubling.

So in a summary, at least I don’t see any trivial errors in their calculations. However, any claims that this paper would show that AGW has been cancelled are not at all supported by the contents of the paper. They actually provide a nice discussion of what this “low” sensitivity would mean in practice: It means that we could keep on emitting CO2 at the current rate for only a little more than 30 years in order to avoid overshooting the 2 K warming target. This paper will probably become one of the many that the deniers keep on citing with the hope that no one will actually read what the paper really says.

## Ari Jokimäki said

Thanks for the comments, AJ! I added a note above warning about my possible misunderstanding.

Ok, let’s see. My number was based on this statement in Schwartz

et al.(and like I said, I wasn’t sure if I had interpreted it correctly):“For sensitivity “TFrom there, I calculated that 22% * 2.1K = 0.5 K._{2x}= 3 K, the corresponding fraction of the warming discrepancy attributable to thermal disequilibrium is 22 ± 7 % (19 ± 6 %, 32 ± 10 % for “T_{2x}= 4.5 K, 2 K, respectively).”I see. So the yellow line in their Figure 2 is the representation of that, or is it? Why is it above the GHG expected? Hang on… they say in the beginning of the section 2 that doubling of CO

_{2}should have caused 2.1K of warming with climate sensitivity of 3K. Now, looking at Figure 2, it is the yellow “GHG equilibrium” line that crosses the 3K sensitivity at 2.1K warming. There’s one thing I misunderstood. Yet they say that black line is“expected present increase (black) in global mean surface temperature (GMST) above preindustrial temperature from the forcing by present (2005) incremental concentrations of long-lived greenhouse gases as a function of CO2 doubling temperature ΔT. Furthermore, as they say that for 3K sensitivity, the warming discrepancy from thermal disequilibrium is 22 %, which I have calculated corresponds to about 0.5 K, so we look at Figure 2 so that we look at the point where yellow line crosses the 3K sensitivity line and travel about 0.5 K down from that point so we should end up at about 1.6 K. I see no lines crossing the 3K sensitivity line at 1.6K warming._{2×}“I guess I still don’t understand. It seems that they are saying that with climate sensitivity of 3K, the expected warming is 2.1K and the ocean thermal lag should take care of about 0.5K at that particular sensitivity, so the Figure 2 should in my opinion have something crossing the point x, y = 3K, 1.6K.

Do you understand what that black line is there?

## AJ said

Ari,

My number was based on this statement in Schwartz et al. (and like I said, I wasn’t sure if I had interpreted it correctly): “For sensitivity “T2x = 3 K, the corresponding fraction of the warming discrepancy attributable to thermal disequilibrium is 22 ± 7 % (19 ± 6 %, 32 ± 10 % for “T2x = 4.5 K, 2 K, respectively).” From there, I calculated that 22% * 2.1K = 0.5 K.

I think there’s actually a small error in your calculation there. The warming discrepancy is not the 2.1 K expected from GHG forcing, but you have to subtract the observed 0.8 K from it. So the correct number would be more like 0.22 * (2.1 – 0.8) K = 0.3 K. Another way to come up with this number is to note that 3 K CO2 doubling sensitivity is 3 K / 3.7 W/m2 = 0.81 K/(W/m2), so 0.37*0.81 = 0.3 K.

Do you understand what that black line is there?

I think the yellow line is GHG forcing only and the black line is GHG forcing minus the 0.37 W/m2. If you imagine a vertical line where equilibrium climate sensitivity is 1 K/(W/m2) (or 3,7 K for CO2 doubling), you notice that the yellow line crosses this line at about 2.6 K , so therefore the yellow line corresponds to 2.6 W/m2 forcing, while the black line is about 0.37 K lower at that point, indicating it corresponds to (2.6-0.37) W/m2 forcing.

Anyway with a closer look there’s actually something about this figure that I’m not getting. Perhaps someone could help me out on understanding why is the upper red line so high up? If it’s suppose to be the expected warming from all the stated forcings, assuming aerosol contribution 0.6 W/m2, shouldn’t it cross this imaginary 1 K/(W/m2) line at 2.6+0.35-0.37-0.6 = 1.98 K ? Looking at the figure, it’s definitely above 2 K at that point, maybe something like 2.1 or 2.2 K.

## Ari Jokimäki said

Ah, yes, there’s a mistake. Warming discrepancy, not the whole warming.

I think you are correct. They have named these lines rather misleadingly, at least for me.

It seems to be about 0.6 K below the yellow line there, so perhaps that’s it. That would then mean that they didn’t include the ocean thermal lag to the red aerosol lines. I think it’s the same thing with the 1.2 and 2.4 lines, they should be lower.

## AJ said

I don’t know, to me it seems like the 1.2 line is pretty pretty much in the right ballpark with the thermal lag included (if I understood right, it should cross 1 W/m2 at 1.38 K) but with the 0.6 line there is something strange there. They seem to be too high, but if I subtract the thermal lag from what I think they should be, then it would seem they’re too low. So I can’t figure out what we (or they) are missing.

## Paul Middents said

James Annan has a pretty acidic take on this paper.

http://julesandjames.blogspot.com/2010/01/more-schwartz.html

Paul

## Ari Jokimäki said

Thanks for the link, Paul. Acidic indeed. 🙂

AJ, I made some measurements from the Figure 2, and here are the results:

Measurements have been made at 1K/(Wm

^{-2}) equilibrium climate sensitivity (3.7 K CO_{2}doubling temperature), given values are the warming at the given crossing point and the difference to the yellow line crossing point is given in parentheses.So you are correct that the red 0.6 K line seems to be in a little strange position, but it’s actually not that far from being 0.6 K off from the yellow line, which would mean it’s without the thermal lag. But the red 1.2 K line seems to be very close to being 1.2 K off from the yellow line, again suggesting that the thermal lag is missing from that. It’s the same thing with the red 2.4 K line, it’s about 2.4 K off from the yellow line. If I haven’t misunderstood again, red 0.6 K line should be 0.6 + 0.37 = 0.97 off from the yellow line (1.57 K for 1.2 K line and 2.77 K for 2.4 K line). The difference between the black line and red lines doesn’t bring up any familiar numbers. So, it seems that the thermal lag is indeed missing from the red lines in that figure.

If this feel correct, then I’ll proceed to write a new commentary on this paper, and I’ll leave this one as a document of my mistakes and the subsequent development.

## Ari Jokimäki said

Well, actually I think I’m overlooking the tropospheric ozone part. There’s the coincidence that tropospheric ozone is 0.35 W/m

^{2}and the thermal lag is 0.37 W/m^{2}. At the equilibrium sensitivity of 1 K/(Wm^{-2}) these correspond to 0.35 K and 0.37 K. Using the calculation AJ gave above, we get very close to values observed in the Figure 2 (difference to Figure 2 value in parentheses):Red 0.6 K line: 2.6 K + 0.35 K – 0.6 K – 0.37 K = 1.98 K (0.15 K)

Red 1.2 K line: 2.6 K + 0.35 K – 1.2 K – 0.37 K = 1.38 K (-0.02 K)

Red 2.4 K line: 2.6 K + 0.35 K – 2.4 K – 0.37 K = 0.18 K (0.15 K)

It’s strange that the 1.2 K line seems to be exactly in the correct position, but the 0.6 K and the 2.4 K lines are both 0.15 K too high.

Edited to add: This all just demonstrates that one should perform calculations thoroughly like AJ did before making any conclusions.

## AJ said

Ari, yeah, it seems like we’re converging towards the same conclusion: The red 1.2 line is about where it should be, but the 0.6 and 2.4 lines are a little bit off. I’ve no idea where this difference of 0.15 K (or actually W/m^2) could come from, except from a bug in the code that generated this figure. Such a small bug, that doesn’t really affect the conclusion of this paper, could have easily been left unnoticed by the authors and reviewers of the paper.

I do think that usually it’s a fair assumption that the peer-review system works at least reasonably well, and therefore it might be a good idea to be pretty careful when criticizing peer-reviewed papers. However, we’ve seen before that this is definitely not always the case (remember Lindzen & Choi 2009, etc), so a healthy amount of skepticism towards the peer-reviewed literature is probably not a bad thing. Even if some criticism turns out to be incorrect, I don’t think it’s such a terrible thing, if we can all learn in the process of correcting our mistakes.

## Ari Jokimäki said

Well, I’m converging with a little help from you, thanks. 🙂

Being careful in criticism is always a good advice, but in this case I just wasn’t paying enough attention so now I have to pay the price and rewrite the second half of this thing.

I apologize to Schwartz

et al.for accusing them of things they weren’t quilty of.## Ari Jokimäki said

I published the new version on this:

https://agwobserver.wordpress.com/2010/01/27/comments-on-schwartz-et-al-2010-version-2/

I added a note in the beginning of this version. I also edited the title of this one so that it wouldn’t turn up in web searches so much.