Thursday, October 30, 2014

Harde Is A Very Wet

There has been some discussion in the comments about Hermann Harde's papers purporting to show that carbon dioxide is the tiniest knob controlling the climate.  Well, bunnies know that Richard Alley differs, but Eli's point was that Harde overestimates the relative humidity at altitude and this obviously increases the role that water vapor plays in his calculations (make no mistake it plays a huge role in real life) and underestimates the effect that carbon dioxide plays in the greenhouse effect.

Just to be clear, Eli does agree with Pekka and Tom that the text dealing with how to calculate raditative transfer in the atmosphere is a fine introduction.  However, long years of gazing at complex spectra lead Eli to see clearly that the water vapor peaks in Harde's calculated emission spectra were way too high.

Some doubts have been expressed.

To put these at rest, first look at Figure 1 from Harde's paper in teh Open Journal of Atmospheric and Climate Change

Compare this with the relative humidity calculated from AIRS satellite data from the NASA Giovanni DAAC

Harde is a factor of 2-3 too high.

The interesting thing is why, and the clue to that can be found in the article which Ray Pierrehumbert wrote and which Eli pointed to in the comments to the original Wet Post, but not the part that he quotes here
So what really determines the water vapor content of the free troposphere? It is easiest to think about this problem in a Lagrangian sense, tracking the water content of an air parcel as it wanders about the atmosphere. The fluctuating water content of the parcel results from a balance between the rate at which water is added to the parcel against the rate at which water is removed. Water vapor is removed either by condensation or by diffusion into a neighboring drier air parcel. Let us suppose for the moment that diffusivity is so low that the latter mechanism is unimportant. In that case, water vapor is removed when the air parcel wanders into a region where the local saturation specific humidity is lower than the current specific humidity of the parcel, at which time the specific humidity is reset to the lower local saturation value and the balance is rained out. The net result is that the specific humidity of an initially saturated parcel after time τ is equal to the minimum encountered along the trajectory during that time. By definition, this is a non-increasing function of τ, though there will be long periods of time over which the minimum remains constant between those times at which new minima are encountered.
Any attempt to use equilibrium thermodynamics to calculate relative humidity in the troposphere outside the marine boundary layer (where it is saturated) is doomed. 


Pekka Pirilä said...

I didn't even try to check, whether Harde's humidity levels are too high in the upper troposphere. What we can see in the figures of Harde's paper seems to tell mostly about the influence of the water in the lowest part of the troposphere, where most of the water vapor is anyway. The emission from water vapor that reaches 12.5 km comes mainly from close to the surface.

My calculation was based directly on the standard atmosphere for Midlatitude Summer. That has also a lot of emission from the water vapor in the upwards radiation at 12.5 km, but over most wavelengths the influence of the vapor on the total upwards radiation including that from the surface is much less, because the layers, where most of the water is, are only little colder than the surface. Thus the main effect is that emission from the vapor compensates most of the absorption of radiation from the surface.

EliRabett said...

Pekka, Harde's calculations of relative humidity diverge sharply from measurements even at low altitudes close to the top of the boundary layer (2 km or so).

As to whether radiation from water vapor makes it out of the atmosphere from down low, remember that the concentration of water vapor is above that of CO2 until you get to about ~9 km, and you accept that radiation from CO2 only can make it into space from that level. Same for water vapor emission.

What is interesting from your and my POV is that pressure broadening driven overlaps in H2O are much less than in CO2 because of the wider spacing of the lines so H2O emission can make it into space more easily for that reason.

The Spectral Calculator is good at these things.

Anonymous said...

Relative humidity goes down in nature, but according to Harde it goes up. Next: anonymouse claims nature is a Clausius-Clapeyron denier.

Rib Smokin' Bunny

...and Then There's Physics said...

Maybe someone can explain something to this other bunny, who is struggling to follow some of this. How does this relate to lapse rate feedback? I had thought this was due to water vapour condensation releasing more energy (heating) at higher altitudes than at lower and - hence - changing the temperature profile so as to produce a negative feedback. I would have thought, however, that that would require that relative humidity increased with altitude (as per Harde) rather than decrease with altitude (as per the lower figure).

Pekka Pirilä said...

A partial answer to your question is that the moist adiabatic lapse rate applies to the moist rising air that forms clouds and leads to rain. The uplift of moist air in tropics is an essential component in the circulation and influences the temperature profile also elsewhere.

As Eli explained referring to Pierrehumbert, the situation is different elsewhere.

Pekka Pirilä said...

I made a few additional calculations and plots using exactly the same model as before but plotting different results. What I have concluded is that there isn't necessarily much difference between our results. The impression of a difference is largely due to the way the sharp peaks and valleys are plotted.

With suitable choices that overemphasize the sharp structure the results are rather similar to those of Harde, while using another choice they are similar to yours.

Including all the peaks and drawing the graph using lines that are much thicker than the actual structure results in something like Harde's plots as can be seen here Plotting values calculated as averages of 1.0 1/cm bands leads to something that resembles your results.

Both figures are drawn from the results of a single line-by-line calculation with grid spacing of 0.1 1/cm.

When a band model is used averaging is part of the original calculation. A line-by-line model calculates values exactly at the grid points. Some points happen to fall near the center of sharp peaks. When the line width is broader than the real structure, a misleading plot is produced. For averages over wider bands both methods give essentially the same results.

EliRabett said...

Pekka, the problem, of course, is how thick the lines are. At ultimate spectroscopic resolution the lines in a drawing are going to be much wider than the lines in the calculation creating the impression of an overlap when there is not one. If you point to the drawing and say, see how much more important water vapor is you are misleading. Eli had thought of that but then he saw Figure 1 above in one of the 2013 papers which clinched the argument of Harde being all wet for him.

Pekka Pirilä said...

We seem to agree on this. For those, who don't know, what the lines are like I show here a narrow (20 1/cm wide) band from a region, where water emission lines are more numerous than average.

I can also add that the moisture profile of the Midlatidude summer standard atmosphere that I used in my calculations are similar to those shown by Eli in the above post based on satellite data.

Anonymous said...

Good for Eli ( not something I type very often ) - he's spotted a real variance.

Do remember to think dynamically, however.

Where there is rising ( the magenta, cyan, and green in particular ), relative humidity is around 100% in the upper atmosphere.

Where there is subsidence ( orange and red ) relative humidity is very low in all but the lowest kilometer or two.

Anonymous said...

"How does this relate to lapse rate feedback?"


Remember, the "lapse rate feedback" is modeled to occur over time with global warming - what Eli is pointing out is not change over time but how the atmospheric relative humidity ( RH ) varies with height.

Consider that rising air and subsiding air must balance ( conservation of mass ).
Then consider that the area of rising air is small ( intense updrafts of thunderstorms ) compared to the area of subsidence ( less intense but large scale motion ).
Subsiding air warms ( in accordance with the equation of state ) but doesn't gain any water vapor, so the relative humidity is less than it was before subsiding.

Remember, too, that the "lapse rate feedback" is not being observed - to the contrary the upper troposphere is warming less than the lower troposphere. Whether this persists and exactly what the ramifications are will be interesting. My thought is that it is a serious error with convection in the models - they don't capture all the convection/precipitation because those events are sub-gridscale and highly non-linear.


...and Then There's Physics said...

Yes, good point. I forgot that lapse rate feedback will be a change over time.

If I understand the latter part of your comment properly, if they're getting the lapse rate feedback wrong, then the net water vapour feedback could be more positive than models currently suggest (water vaper + lapse rate, I guess).

BBD said...

Remember, too, that the "lapse rate feedback" is not being observed - to the contrary the upper troposphere is warming less than the lower troposphere.

Is this known or asserted?

Russell Seitz said...

Harde introduced Salby in Hamburg :

"as a highly recognized guest from abroad , he is one of the few specialists who really have a view over the whole area of the climate development. Despite of his long relationship with the most renowned climate institutes he has preserved his own critical and constructive reasoning , which in some parts is in real contradiction to the official expert opinion and which also faces the assessment of the IPCC

In your talk today you will present the most accurate discussion of climate to my opinion..."

Was he following orders ? …

Anonymous said...

"Harde is a factor of 2-3 too high."

Presumably that refers to the relative humidty curves (Harde's and the measured ones) and it's really not true that they differ by a factor of 2-3 over the entire range (they don't differ at 0km and the difference increases to about a factor of two at 4km for the tropics and midlats)

To find "how wet harde is" one has to multiply the saturated exponential by the rel humidity as a function of altitude for each case (harde's incorrect rh curves and the correct ones) and integrate the difference between the curves over the entire altitude range (in reality, the exponential drops very quickly, so the higher altitudes - -where the greatest difference in rel humidity between harde and reality occurs --add very little).

for the tropics and midlatitudes the overall (integrated) difference is about 25%.
For the poles, it's less.

But I would agree that the overall difference is significant.

- Jim

EliRabett said...

The differences at lower altitudes means little because the radiation is essentially trapped as is the case for CO2. What matters is the humidity at the altitude from which the water vapor emits to space, e.g. where the mean free path upwards exceeds the height of the atmosphere.

That, of course, is higher up (several km) so yes, the mistake in relative humidity makes a significant difference.

Since the absolute humidity is the saturated humidity which is a function of temperature, times the relative humidity, why yes, an error of x% in relative humidity is an error of x% in the absolute humidity.

Eli blogged about a similar issue here. You can judge, more or less, the altitude at which emission to space occurs by where the line meet the black body curves for the temperature at various altitudes.

As Pekka and Eli discussed, you should really use the actual linewidth at those altitudes for that, but the temperatures are fairly low as is the emission to space altitude.