how to get the sea level pressure

[jie_wong]

Hello,
I want to change the sea level pressure in CAM, and in the phy_state there is only a variable named pmid which is the midpoint pressure (Pa). So, I am asking how to get the sea level pressure and modify it. Can I treat pmid at the bottom layer as the sea level pressure?

Thanks
Jie

 

[peverley]

Hi Jie,

To clarify, do you want the sea level pressure OR the surface pressure?

Sea-level pressure is calculated only for model output/history. If you want to see it, you can add PSL to your fincl list in user_nl_cam. You can also calculate sea-level pressure using the cpslec subroutine in src/physics/cam.

If you want the surface pressure, that can be found in phys_state%ps.

Courtney

 

[jie_wong]

Hi Courtney,

Thank you for supporting the calculation of sea-level pressure. And my goal is to modify the 3-dimensional pressure, but when I change (just replace the values with my modified values) phys_state%pmid, there is always an error in the model with a slight modification. So, I wonder if the variable phys_state%pmid is not a good place to modify. Anyway, my final purpose is to add a disturbance like the tropical cyclone in the model.

Looking forward to your reply!

Best wishes!

Jie

 

[jie_wong]

Hi,

I mentioned that when I modify phys_state%pmid, there will be an error, but when I modify phys_state%pmiddry, the error disappeared. What's the difference between phys_state%pmid and phys_state%pmiddry, I know phys_state%pmiddry is dry pressure.

Looking forward to your reply!

Best wishes!

Jie

 

[zarzycki]

Hi Jie,

pmid should be total pressure (including moisture) and pdry should be the dry mass which excludes such tracers. They should be relatively similar to one another, although they diverge in regions of high moisture content such as the tropics where atmospheric q can contribute to the integrated mass.

Really, you should be only be adjusting PS given the model's hybrid coordinate. Adjusting PS will then adjust the pmid values based on p(k) = a(k)*p0 + b(k)*ps. Courtney is correct that PS is the variable needed here, not PSL which is just a diagnostic and not used predictively by the model itself. However, over ocean (where tropical cyclones exist) PS = PSL, which makes life easier (one doesn't need to consider PHIS).

Another thing to note is that if you are modifying the state (ex: to add a tropical cyclone) you need to adjust *all* the state fields to be in gradient wind balance. I.e., changing PS changes geostrophic U,V winds which change T via thermal wind, etc.) -- there are a few possible ways to do this, but one starting point is Reed and Jablonowski (2011, An Analytic Vortex Initialization Technique for Idealized Tropical Cyclone Studies in AGCMs) who derive analytic state field perturbations that can be used to initialize a cyclonic vortex.

hope that helps.

 

[jie_wong]

Hi Jie,

pmid should be total pressure (including moisture) and pdry should be the dry mass which excludes such tracers. They should be relatively similar to one another, although they diverge in regions of high moisture content such as the tropics where atmospheric q can contribute to the integrated mass.

Really, you should be only be adjusting PS given the model's hybrid coordinate. Adjusting PS will then adjust the pmid values based on p(k) = a(k)*p0 + b(k)*ps. Courtney is correct that PS is the variable needed here, not PSL which is just a diagnostic and not used predictively by the model itself. However, over ocean (where tropical cyclones exist) PS = PSL, which makes life easier (one doesn't need to consider PHIS).

Another thing to note is that if you are modifying the state (ex: to add a tropical cyclone) you need to adjust *all* the state fields to be in gradient wind balance. I.e., changing PS changes geostrophic U,V winds which change T via thermal wind, etc.) -- there are a few possible ways to do this, but one starting point is Reed and Jablonowski (2011, An Analytic Vortex Initialization Technique for Idealized Tropical Cyclone Studies in AGCMs) who derive analytic state field perturbations that can be used to initialize a cyclonic vortex.

hope that helps.

Hi zarzycki
I am still confused how to add a tropical cyclone to the model. And the first problem is when I modify phys_state%ps or phys_state%pmid there is no difference with the results which means the modification did not work, so I am wondering if you have some method how to modify the pressure. The second problem puzzled me is from the paper Reed and Jablonowski (2011) you mentiond, does that mean the packed variables T,U,V,p,P,q must be derived by the formulation in the paper? How does it different if I modify T,U... with the averaged mean state of observation data of tropical cyclones happen?

Looking forward to hearing from you.

Best Wishes
Jie

 

[zarzycki]

Hi Jie,

Sorry for the late reply.

I am unsure why your modification is an issue. I almost exclusively modify initial condition (ncdata) or restart files. Modifying PS/PSDRY should work in those files. It is possible you are modifying "in-line" and you are only modifying the physics state but the model is not actually updating the full state that is passed back to the dynamics based on that information. If you absolutely need to go this route (I advise against it unless there is a specific reason since it is easier to work with initial condition files) I would suggest finding a "template" subroutine in physpkg.F90 and mimicking that.

In RJ2011, the equations refer to "perturbations" or "deltas" which are added to the existing state of the atmosphere (e.g., T' such that T = T_orig + T'). The perturbations themselves do not necessarily depend on the state of the atmosphere but they then need to be added to the background state to "seed" a vortex into the field.