search

atmtools / konrad

Implementation of a radiative-convective equilibrium model.
MIT License
19 stars 18 forks source link

Add draft version of Romps (2014) humidity model. #51

Closed lkluft closed 5 years ago

lkluft commented 5 years ago

This PR adds a draft version of a relative humidity model described in Romps 2014.

As is, the model is unstable for the default choices of solar constant (runaway greenhouse). A reduction of the solar constant by about 10 Wm^-2 allows stable simulations.

lkluft commented 5 years ago

Then we need to sort out a "better" relative humidity as a function of temperature relationship. I think using ERA5 would be nice, but as we discussed we would need to extrapolate it somehow. If you send me the data, I can have a go at this, or you can keep working on it if you'd prefer! We don't want a runaway as our reference climate!

The problem with the ERA5 data is, that the change in RH(T) is quite large at the surface. This is translated into a massive feedback in our warming model.

I tried to play around with the surface albedo a bit. One could argue that our current value of 0.2 is rather low given the fact that we do not have clouds.

SallyDa commented 5 years ago

How about we take just the upper part from ERA5, so we don't have such a strong spike in RH in the upper troposphere. We could get rid of the lower tropospheric peak and have RH following the idealised profile above ~300 K. In Romps Fig 7, we see that RH is not conserved with T close to the surface in the "realistic" model, so probably it also isn't in the real world, and maybe it's okay to take a simplified approach here.

I'm not so sure about your albedo arguments, as we're supposed to be modeling clear sky, which in the tropics is mainly over ocean (albedo ~0.07), so I already think our albedo is too high! Especially later if we want to look at the effect of clouds, as I guess low clouds basically only affect things through their albedo.

lkluft commented 5 years ago

How about we take just the upper part from ERA5, so we don't have such a strong spike in RH in the upper troposphere. We could get rid of the lower tropospheric peak and have RH following the idealised profile above ~300 K. In Romps Fig 7, we see that RH is not conserved with T close to the surface in the "realistic" model, so probably it also isn't in the real world, and maybe it's okay to take a simplified approach here.

I am running some simulations to test a fixed relative humidity above 295 K. The RH(T) mapping for increasing temperatures above the tropopause is still very messy, but let's see how that works.

I'm not so sure about your albedo arguments, as we're supposed to be modeling clear sky, which in the tropics is mainly over ocean (albedo ~0.07), so I already think our albedo is too high! Especially later if we want to look at the effect of clouds, as I guess low clouds basically only affect things through their albedo.

I guess it is fine to have a higher surface albedo than the 0.07 for the ocean. Even though we are looking at clear-sky, a cloud-free tropical aqua-planet is a bit too far off. So using a more "global albedo" to account for missing clouds should be fine. You are right, though, that we have to adjust that when including clouds!

SallyDa commented 5 years ago

I am running some simulations to test a fixed relative humidity above 295 K. The RH(T) mapping for increasing temperatures above the tropopause is still very messy, but let's see how that works.

Ok, hopefully that works well. I don't think the region above the cold point should be a problem if we apply the stratospheric humidity adjustment afterwards. So I guess you don't need to care too much about RH(T) there.