convert/interpolate from B1850C5CN to B1850CN

[catilev@umich_edu]

I was wondering if anyone has tried starting a case using CAM5 and then doing a branch/hybrid run with data from the original case but using CAM4--what are things to consider other than the difference in the number of levels in the atmosphere (and would interpolation or use of data from the closest comparable level be preferable)? thank you!

 

[catilev@umich_edu]

...alternately, a startup run with my semi-equilibrated ocean and a mean atmosphere probably makes the most sense

 

[hannay]

The atmosphere initial condition won't matter too much. You could start from an interpolated initial condition as you described or just take the initial from anotehr run at the same vertical resolution.For the ocean and land, the spinup is much longer. 

 

[ckuo@lbl_gov]

Hello,This post is very pertinent to me and thank you for your advice on the interpolating the atmospheric model when trying to use initial conditions from B1850C5CN to B1850CN.I have gone in the opposite direction, taking a B1850CN run as an initial condition, and interpolating atmospheric data to 30 layers, to run in B1850C5CN.The vertical interpolation has been checked for mass conservation.  However, while the initial condition was in radiative balance usiing CAM4, for CAM5, it was +5 W/m^2 in the first year (CESM1.2.2).  The file which was interpolated was an initial restart file, b40.t31x3.037c.cam2.i.0507-01-01-00000.nc.   Is +5 W/m^2 too large to start, considering the differences in cloud physics modeling between CAM4 and CAM5?Cecile, what is a reasonable RESTOM at start, and would it be better to use another initial condition from a CAM5 run?What could be an expected time period for the atmospheric spinup, using this b40.t31x3.037c initial condition, on B1850C5CN, CESM 1.2.2 out of the box?Thank you in advance !

 

[hannay]

A RESTOM of 5 W/m2 seems on teh high side. It might be ok if it sharply drops after the first year. There is no strict rule about it, but I would say the average RESTOM for yr5-10 should be less 1 w/m2 (and 0.5 W/m2 would even be better). 
If the RESTOM is larger than this, all the extra hear will go into the ocean (as teh storage capacity of the atm is small). 

 

[hannay]

Also, to answer to your question about time to spin up. The atmosphere initial condition doesn't really matter. It will spend in a few weeks. WHat will matter for the spinup is the ocean state and the land state.The land takes decades to spinup and the ocean takes decades to century dependingor far you are from the initial condition

 

[ckuo@lbl_gov]

Hello Cecile,Thank you so much for your immediate attention to my question thsi morning.   The run is continuing so far to 4 years, and so far there is a 0.7 W/m^2/year reduction in RESTOM.   Keeping my fingers crossed that it will reach and stabilize  to RESTOM < 1 W/m^2 by 10 years.Best regards

 

[ckuo@lbl_gov]

Unfortunately, radiative balance was not achieved by a 13-year run.Here is an explanation from Christine Shields:I have not, nor has anyone here at NCAR (to my knowledge) used the T31 configuration for CAM5... unfortunately, this resolution/dynamical core (eulerian) is NOT supported (scientifically) for CAM5. We apply either the finite volume or spectral element to all our CAM5 runs.   I suspect there is some significant tuning that needs to happen for the T31 dycore to be used for CAM5, (which is likely why you are way out of balance) among any other issues that we haven't bumped into yet.  My recommendations for you: 1. Switch to FV2deg_gx1 resolution... yes, I know this is significantly more expensive, but it is tuned, validated, and ready to go.2. If this is not an option for you, and you would like to use the T31_gx3 for CAM5, you will need to test and tune (which could take significant time. To see how far out you really are, run the AMWG diagnostic package and compare to observations).  Alternately, you could put a request out on the bulletin board to see if anyone else has done this...  I think the folks at Oakridge still use the Eulerian dynamical core, but they focus on super-high resolution and not low. Sorry I couldn't be more helpful!   Christine

 

[ckuo@lbl_gov]

Using f19_g16 grid, B1850C5CN, and hybrid with b40_1850_2d_r07c5cn_160jp,  RESTOM = -0.281 W/m^2 in our modified code.