Monday, January 18, 2010

Fingerprinting climate change

I'll take up one of the questions from the question place, how do we 'fingerprint' current climate change as being from CO2, rather than from any of the many other things that affect climate?  Same questioner asked several other interesting questions, and they'll be topics for later posts.  One thing I've mentioned is that there are indeed many things which affect climate.  A few to start with are: CO2 (or, more generally, the non-water vapor greenhouse gases*), solar input, volcanic aerosols, and orbital variations.

I don't really like 'fingerprint' as the description, myself.  I think of the process more as the 'duck test' -- if it looks like a duck, walks like a duck, and quacks like a duck, it's probably a duck.  We start by finding out what would be true if a given candidate were driving climate change, the more implications the better, and then go look at what is going on in the atmosphere.  So I'll approach it that way.

Orbital cycles, we know very well.  They're slow -- taking 20,000 to 100,000 years to cycle.  The magnitude can be large -- about 5 C for global mean temperatures between the coldest parts of an ice age and the warmest parts of the period between ice ages.  But this gives us a rate of change for temperature of 0.5 degree per 1,000 years (5 degrees from the trough to the peak in 10,000 years -- the fastest orbital cycle). 

Volcanoes affect climate by dumping aerosols into the stratosphere.  Or, rather, if they affect climate, this is how they do it.  It takes a pretty major eruption to get junk in to the upper atmosphere.  Mt. St. Helens had no real effect on climate, for instance.  It takes something like Mount Pinatubo (1991) or El Chichon (1982) to have a significant effect.  The aerosols do two things to climate.  First, they warm the stratosphere.  The aerosols absorb solar energy, warming themselves and their surroundings.  Second, they cool the surface -- the energy they absorb in the upper atmosphere is not available to keep us warm at the surface.  But there's a different characteristic to these aerosols -- they fall out of the atmosphere quickly.  After 2-3 years, you can't see their effect on temperatures any more.  During the 2-3 years, you can see something like a 0.5 C cooling (Pinatubo's peak).  If you started your observation near the peak of a volcano's effect, you could see a very large warming -- 0.5 C in 2 years.  But it won't continue.

Solar output also varies.  For data, see, for instance, the NGDC archive or PMOD's site.  The latter also presents graphics that tries to remove the sensor to sensor offsets, so that you can get a proper overall picture.  One thing we'd expect is, when the solar output (total solar irradiance at the sites) is higher, the earth should be warmer.  And when it's lower, the earth should be colder.  We can't expect instantaneous response to the variations, for the same reasons that noon is not the hottest time of day, or that December 21st is not the coldest (northern hemisphere) or warmest (southern) day of the year.  The climate system takes time to respond.  But the cycle should be present.

When the sun is brighter, we expect both the stratosphere and the surface to be warmer.  Both see more energy coming from the sun.  Since the sun varies on about an 11 year cycle, we also expect solar-induced climate change to show an 11 year cycle.  Further, if you look at the solar output over the period of satellite record, we expect to see a slight cooling, if anything.  Solar output has a declining trend over this period.  We can also use the">simplest climate model to estimate just how much temperature should change over the 11 year cycle, or due to the slight declining trend.  For the 1 Watt per square meter variation during a solar cycle (but remember to divide by 4 because the earth is not a flat disk always pointed squarely at the sun), we expect a temperature change of 0.05 C.  The much smaller declining trend in solar output would contribute, then, some much smaller than 0.05 C trend over the satellite period.

If carbon dioxide were the source of a climate change, we expect some rather different things than for those other mechanisms.  As with the solar changes, we expect the surface to warm.  Unlike with the sun, however, we expect the stratosphere to cool.  We can repeat the response calculation for the greenhouse gases.  They've contributed an increase of about 2.25 Watts per square meter (which is for all square meters of the earth's surface).  So with this being the only consideration, we'd expect about 0.45 C warming since pre-industrial times. 

All these descriptions, I hope it is obvious, are simplified.  You'd have a lot more work to do to publish in the scientific literature on climate change attribution.  On the other hand, while simplified, they all point in the correct directions.  (Namely, you'll see these directions in that same literature).

So now let's look at some climate observations, and see which mechanisms look most like that.

1) Over the past century, global mean temperatures have warmed by about 0.75 C (1.3 F).
-- that rate is 7.5 C per thousand years
2) The surface has warmed but the stratosphere has cooled
-- for the latter, see, for example, Thorne et al., 2005 Revisiting radiosonde upper air temperatures from 1958 to 2002, J. Geophysical Research, 110, D18105, doi:10.1029/2004JD005753.

How does this match against the mechanisms?

Orbital cycles: The observed magnitude could be explained, but it is 15 times too fast.  For the same reason as with solar input, this also fails to explain the second observation -- stratospheric cooling.

Volcanoes: The magnitude is ok, but the time scale is far too long.  Recovering from a major volcanic eruption would show a cooling of the stratosphere (as the aerosols fell out) and warming of the surface.  And it could provide a 0.75 C warming, but it would take only a few years, not a century. 

Solar output: Warmer sun would warm the stratosphere, which is the opposite of what is seen.  The trend in solar output is even or down, but the observed temperatures are substantial warming (satellite estimates for the lower atmosphere being about 0.5 C over the 30 years of satellite temperature observations).

Greenhouse gases: Cool the stratosphere, which is observed.  Warm the surface, which is observed.  On their own, should warm the earth by about 0.45 C, and a greater warming than that is observed.  This, too, is expected as water vapor is expect to act as a feedback.

In other words, the greenhouse gas candidate looks pretty much like a duck, it walks like a duck, it quacks like a duck, and it's about the same size as a duck.  In contrast, none of the other candidates look even vaguely ducklike.  You'll certainly want to continue research -- find more characteristics of greenhouse gas climate change, make more precise predictions for them, and get enough (and good enough) observations to match against your prediction.

* I know that some will complain about my separating out water vapor.  The thing is, water vapor is a responder to the other things that affect climate, not something that drives the changes in the first place.  Water cycles very rapidly through the atmosphere -- days to weeks.  If there's a bit too much at the moment, water vapor turns in to clouds, and then in to rain or snow -- taking it back out of the atmosphere.  If there is too little, you get more evaporation and rain and snow, until there's enough.  It is these other factors (dry greenhouse gases, sun, orbit, aerosols, ...) which set what is 'enough'.


Dan Pangburn said...

Since 2000 the atmospheric carbon dioxide level has increased by an amount equal to 21% of the increase from 1800 to 2000. According to the average of the five reporting agencies (four since Climategate), the average global temperature has not changed much for several years and during the seven years from 2002 through 2008 the trend shows a DECREASE of 1.8°C/century. This measured SEPARATION between the increasing carbon dioxide level and not-increasing average global temperature is outside of the 'limits' of all of the predictions of the IPCC and 'consensus' of Climate Scientists. The separation has been increasing at an average rate of about 2% per year since 2000. It corroborates that, at the present CO2 level, atmospheric carbon dioxide increase has no significant influence on average global temperature.

Climate change is natural.

The on-going temperature decline trend was predicted.

All average global temperatures since 1895 are accurately predicted by a simple model. There was no need to consider any change to the level of CO2 or any other greenhouse gas.

The model, with an eye-opening graph, is presented in the October 16 pdf at (Replace all references to PDO with ESST for Effective Sea Surface Temperature).

carrot eater said...

Is a Fourier transform too mathy for this format? If the solar cycle had much of any impact on the global means, it should be apparent in a FT, no? (I should stop being lazy and just do it...) Considering the lag and especially the magnitude of interannual variability (something the above commenter would do well to study), I would expect the solar cycle to get lost in the wash.

On the other hand, the greenhouse gas forcing is persistent and increasing over time, so over long enough time periods, you are able to see its effect emerge from the noise of interannual variability. Six years of data, taken in isolation, is not a long enough period.

Penguindreams said...

Thanks for the opportunity to point out a number of elementary climate articles I've posted before. Many of your errors in your short comment are at this very elementary level.

Cherry Picking
What Cooling Trend?
What is climate
Deciding climate trends

And, an old note on how you and your pals at co2skeptics, renamed climaterealists are unreliable sources on climate. Which your short comment here serves to re-illustrate.

I made use of fourier transforms last summer, in discussing a very bad paper, Introductory time series analysis. I'll be doing so again, when I go back to the 800,000 year climate record from the ice cores. It's just, as you can tell from the introduction, that I'm not assuming that everyone can compute the integrals themselves and this makes for some extra writing.

ourchangingclimate said...


For the temperature change to GHG (0.45 deg for 2.25 W/m2) you exclude feedbacks, whereas for the others (orbital, solar, volcanic) you seem to include the effects of feedbacks on the eventual temperature rise. This complicates the comparison, or am I missing something?


Penguindreams said...

You're 2/3rds correct. The sensitivities I give for orbits and for volcanoes are taken from observations. So they definitely have all feedbacks involved.

The solar cycle's effects are estimated in the same way as the greenhouse gas effects, so neither has any feedbacks included. We can turn around, then, and ask how strong the feedbacks have to be in order for the sun or greenhouse gas to supply the observed warming. For greenhouse gas, it's an amplification of about 1.7, for the solar cycle, it's far greater than 15. (15 would be enough for the peak to trough solar cycle, but that wouldn't give us a climate trend. Since the trend is very small, the amplification has to be very large.)

The feedbacks would also have to be pretty strange for the sun to be the source. Surface has been warming the last 30 years, while the sun is cooling. It could be that there's a feedback that provides a warmer climate from a cooler sun. But someone will have to present the work to show that such a feedback really exists.

Dan: Simply re-advertising your article will not be posted. New content relevant to the topic is required, not mere repetition. Your link is already up, so anyone who is interested can see what you had to say there.

Frank O'Dwyer said...

Thank you very much for this, it really helped my understanding.