Thursday, July 16, 2009

Yet more on decadal prediction and record temperatures

There's an interesting post up on RC, but I'm not convinced by it. I commented on a predecessor of that work a few years ago, which I think is very interesting in terms of our understanding of the behaviour of complex dynamical systems but rather over-sold inasmuch as how relevant it is to quantitative prediction of the climate system.

To expand on that a little: the research outlined there points to qualitative changes in the climate's behaviour which (they think) might affect temperature trends, but their analysis cannot (that I'm aware of) determine the magnitude of these natural trends. Thus, while this process may help to explain the somewhat sharp corners in the temperature record, there is no implication that it actually explains a substantial portion of the observed trends. (As an aside, the sharp corners aren't really that sharp and hardly need explaining anyway, over and above the forced response and a spot of interannual variability). In fact, when they applied their analysis to a GCM in that 2007 research, they found the method worked just the same, and explained small changes in trend of about 0.05C/decade - roughly 1/4 - 1/3 of the recent forced trend, so nothing like enough to cause a cooling or even a halt in the forced warming. Of course the model might not be quite the same as the real climate, but in many respects they are pretty good.

So I'm a little surprised by Kyle Swanson's post in which he seems to be proposing that the climate has been flat-lining since 1998 due to this natural effect. For starters, global warming hasn't actually stopped, and the red line on his graph that I've copied below is just an arbitrarily-drawn line that has no possible connection to any linear fit that I can see.
The green trend line of 0.1C/decade is fitted to 1979-1997, and although he says "no cherry picking here" it is clearly biased low relative to other trends he could have used. I think 0.15C/decade is a more reasonable estimate of the forced trend, there is no evidence for any major change in this post 1998 and I don't expect to see any such major change. (I also queried the IPCC's prediction of an increased warming rate over the next 30 years back when they were crafting their report, as I see no evidence for that either).

So the lines are pretty much junk, I'm afraid (I'm not particularly bothered by their representation of Smith et al - I'd agree that a warming rate of 0.3C/decade doesn't seem very likely). But there's more...there seems to be an El Nino on the way. Following last year's strong La Nina episode, the Nino indices are trending positive, and the forecasts are more or less in agreement (though not universally so) in pointing to at least a moderate El Nino right through the coming winter. Eg see here for a recent summary and forecasts. Of course it might not be a bumper event and might not result in the temperature breaking the 1998 record (next year would probably be the one to look out for, as the temperature anomaly seems to lag the El Nino itself). But it would not surprise me to see Swanson's prognostication of no rise in temperatures proven wrong rather sooner than 2020.

Update
I see that Tamino has said some pretty similar things. But not identical, so go and read his post.

7 comments:

Roger said...

Hi James,
It's all interesting stuff, and I'd like to get a better handle on it by looking more closely at the overall energy budget of Earth. I've been reviewing estimates of ocean heat content over the years and see that estimates have been considerably lowered since 2000. Youare an expert in these matters, so could you explain to me why Syd Levitus' estimate for the rise between 1955 and 1994 dropped so much between his 2000 paper and his 2009 paper?

Thanks

James Annan said...

I don't have all the info to hand right now, but I recall there is a (now) well-documented problem with the drop speeds of various devices which resulted in a significant bias in a chunk of the record. Basically, they measure temperature as they sink, so you have to know how fast they fall to know the depth of the measurements. Correcting their drop speed has changed the linear trend by a bit and also smoothed the record, eliminating a large wiggle that had cause a lot of headaches, and bringing these measurements into line with others made in the same time and place by other methods.

The change in linear trend is not huge though.

Roger said...

Thanks for your reply James, please would you comment on another aspect which has been puzzling me.

According to my calcs, the ocean must store heat to depths well beyond the near surface mixed waters in order to account for the thermal expansion of the ocean observed by satellite altimetry. But how does the warm water get down there? Is it something to do with increased salinity of surface water due to evaporation which makes it dense enough to sink? Or are there underwater waves which mix the deeper waters? Or what?

Thanks for your time.

James Annan said...

You are right that the warm anomaly penetrates to some depth, and it is basically due to mechanical mixing processes. Tidal energy, wind-driven currents and surface wave energy are all dissipated in the ocean (eg through friction and internal wave breaking) which results in turbulence and mixing. There is also deep convection at high latitudes where surface water is rapidly mixed to great depths. The warming anomaly has been observed to below 700m, although a large majority of the warming is above this level.

Roger said...

James, thanks again, most helpful.

A further question if you have time. The ocean seems to have heat storing and heat emitting modes, manifested by el nino and la nina. In warming mode, the rising SST's radiate into the atmosphere and raise the air temperature, until the newly warmed air suppresses the oceanic emission. Looking at graphs of SST against lower tropospheric air temp during this period of solar quiet, the relationship seems quite clear and 'noise' free. The air temps seem to lag the SST temps by between 2 and 4 months or so.
http://www.woodfortrees.org/plot/uah/from:2005/scale:0.5/plot/hadsst2gl/from:2005/offset:-0.4/plot/hadsst2gl/from:2005/trend/offset:-0.4/plot/uah/from:2005/trend/scale:0.5
Is the lag due to the number of times the back radiation of the atmosphere bounces the heat back to the ocean surface and goes round the cycles of evaporation precipitation etc before the heat finally escapes to space? In other words not so much thermal inertia as a maze of interactions which take time to play out?

James Annan said...

I'm afraid I don't know anything about a possible phase lag, either its robustness or cause. I could guess that the ocean basically drives the atmosphere and the lag is due to the thermal inertia of the latter, but it's just a guess.

Hank Roberts said...

tidbits, might lead to something useful if followed:

http://www.argo.ucsd.edu/

https://www.sciencemag.org/cgi/content/full/318/5857/1765

https://www.sciencemag.org/cgi/content/full/sci;318/5857/1765/DC1