2010-12-01 07:11:25Lapse Rate
Paul D


A technical question about convection/lapse rate.

Actually firstly, I'll make a confession...

I'm the person behind the Warmcast greenhouse effect simulator. John Cook knows about it, but probably didn't know I was responsible. I think one or two others have come across it:


It's very basic (although what do you compare it with???!), but it is supposed to be basic.
It's extremely loosely based on radiative-convective models, without the maths, pure science and currently without convection.
Before anyone says it has a lot of flaws, well yes I know that, do you really expect such a simple idea to model the atmosphere accurately??

OK got that out of the way...now.

Currently it doesn't include simulation of convection although i'm trying convection out as a simple algorithm.

Does anyone know how much impact convection has on the gradient of the atmospheric radiative lapse rate???

Please be gentle, if you think the simulator is absolute crud.

2010-12-02 12:05:13
Glenn Tamblyn


The Ville


Here is a link to the wiki article on Lapse Rate http://en.wikipedia.org/wiki/Adiabatic_lapse_rate

The gist of it seems to be that convection doesn't impact on the lapse rate which is purely Delta T with altitude. What matters is the temperature of the air mass and its relative humidity as this determines which of two lapse rate regimes apply, the dry or moist lapse rate.

Convection doesn't seem to be involved in the magnitude of the lapse rate at all


2010-12-02 20:44:33
Paul D


Hi Glenn, i'll have a look at that page.

But the reason I ask is that in my copy of 'Climate Modelling Primer' by McGuffie and Henderson-Sellers, there is a graph (Figure 4.2 to be exact) that shows convection has an impact on the troposphere lapse rate, because the energy is redistributed through a column of air.
The stratosphere has an inverse temperature gradient and is relatively stable because it is hotter at the top than the bottom. The troposphere is different because it warms up from the bottom.

My understanding is that radiative-convective models do a simple mathematical adjustment (instead of simulating the physical shifting of air) to take into account convection in the troposphere.

2010-12-02 21:09:50
Paul D


Actually the Science of Doom has a useful page:


It has a simple diagram showing the difference in the Troposphere when convection is included.

My question is, how much is the gradient changed??

SoD suggests the radiative gradient would be more than 10K/km
With convection its about 9.8K/km

The actual rate is 6.5K/km once humid and dry conditions are taken into account.

Maybe that is enough.

I'm actually contemplating the idea of upgrading my simulator to include Stratosphere simulation, but I think my brain might burst trying to do that. I have actually tried out convection in the existing programme, of course if you apply to much, the upper atmosphere ends up hotter then the bottom, which isn't what happens in the troposphere.

2010-12-14 21:56:56
Paul D


I have added a new page on the simulator blog explaining some differences between a real Earth atmosphere and the simulator:


I have also included a simple convection algorithm in the simulator which results in s slightly different temperature profile.

I think I might revise some of the texts. I'm finding as my knowledge improves, I'm not happy with previous descriptions.

Any comments welcome.

2010-12-15 03:01:27Simulator
Julian Brimelow

Radiation transfer is not my forte, but I really enjoyed playing with your simulator The Ville.  Great job IMHO...especially of one is presenting the concept to younger people or lay people.

 Which reminds me, would there be a way to speed things up a bit?  Right now one has to invest quite a bit of time to watch the experiment, and you might lose people before they get to see the conclusion.


Could you also maybe have a before and after vertical temperature profile, or one that is continuously updated on the right-hand side?

2010-12-15 06:05:46
Paul D


heh, heh.

I'll have a think about the speed thing!

It would require processing the data-array quicker and probably boosting the frame rate??

Another idea I had was to 'seed' the atmosphere so that it was pre warmed!

I had thought about producing a plot of temperature that could be printed out, with changes marked on it, such as changes in the ice extent or other things.

2010-12-15 06:14:26
Paul D


Actually I think there is a bug in the energy meter, which is a bit annoying. I thought I fixed that weeks ago!

Sometimes it doesn't work at all, although if the page is refreshed it usually fixes the problem.


2010-12-17 09:10:59

Sorry for jumping in so late.
In the real troposphere the lapse rate is determined by convection; it is the result of the adiabatic expansion of a lifting parcel of air. Radiative balance determines surface temperature and the height at which IR can escape to space, i.e. the energy balance.
2010-12-17 18:12:32
Paul D


Hi Riccardo.

Thanks for your input. My understanding of the subject is that there are two lapse rates that combine to give an 'environmental' lapse rate. eg. if you had a theoretical world where convection did not exist, then you would have a radiative lapse rate or temperature profile through the atmosphere.

If you then added convection to this world, the overall lapse rate or temperature profile would adjust due to the imbalances you describe, resulting in cooler temperatures near the surface and a steeper temperature profile.

I'm basing my comments on what I have read about radiative-convective models.

Are you suggesting my graphs in the new page are poorly labelled?
eg. my labelling of the Troposphere on the simulator graph and the supporting text is misleading??

or the whole concept is wrong??? 


Some info about RCMs:

Book: A climate modelling primer (Kendal McGuffie and Ann Henderson Sellers)

Apart from my possible poor descriptive terminology I don't think the sim is far removed from the basic concept of an RCM (other than being heuristic and very simplified).

 Regarding surface temperature, that's something I need to convey visually. The simulator doesn't currently indicate surface temp.

2010-12-17 20:08:24

Radiative equilibrium alone would indeed result in a lapse rate; it turns out to be unstable when convection is added, i.e. it's too steep. Convection is more efficient in moving heat upward and will reduce the temperature gradient. So convection dominates the temperature profile.
To first order, the lapse rate (determined by convection) does not change by adding GHG while the radiative equilibrium does. The latter changes the energy balance on top of a constant lapse rate, thus warming the earth.
This is as far as atmospheric physcs is concerned. As for your simulator, it correctly shows how the radiative balance works and how the temperature of the earth changes by adding GHG. It also shows the build up of the lapse rate, although one steeper than real.
I think that if you want to show the effect of adding GHG to the atmosphere as you say ("find out about the processes that cause greenhouse gases to trap energy in an atmosphere and warm the planet."), you can leave convection out and the simulator is good enough. If you want to show how the lapse rate works, you need convection.

2010-12-18 00:32:17
Paul D


Thanks Riccardo at first I thought we were at cross purposes.

I was thinking of steep, in terms of plotting temperature along the x axis and height along y.

The simulator originally was just 'radiative', now it does have a simple emulation of 'convection' as well.
I was aware of the obvious fact that the radiative lapse rate would be 'steep' and convection would kick in at some point.

I think the main aim is to bridge the gap between existing simple explanations using 'blankets' and big diagrams of CO2 molecules, and the complexity of hard core science.

2010-12-28 10:07:21
Paul D


Hi Riccardo

I have been thinking about what you have said. I think in order to simulate convection as the main transport mechanism, I would have to increase the amount of energy (photons) bombarding the surface in the simulator.

I think that would make it harder to show the mechanism that is working. Although if I need to speed the whole thing up as suggested by Albatross, maybe this is the direction to go.

A prototype version 2.0 maybe germinating!??!

Thanks for those that have commented.