Different h0 outputs of base case and validation case in PORT

[CallanGwd]

I ran a base case (compset is 2000_CAM50_CLM50%SP_CICE%PRES_DOCN%DOM_MOSART_SGLC_SWAV) and a validation case (compset is PC5(2000_CAM50%PORT_SLND_SICE_SOCN_SROF_SGLC_SWAV)) of PORT. According to the tutorial in https://wiki.ucar.edu/display/port/PORT, "The difference between the fluxes, FLNT,FLNR,FLNS,FSNT,FSNR,FSNS should be zero". The h1 outputs (nhtfrq = 73, avgflag_pertape = 'I', referred from the tutorial website above) of these two cases are zero indeed. However, the h0 outputs (monthly average, nhtfrq = 0, avgflag_pertape = 'A') of them are totally different!

I tried to modify the settings in the namelist, but the results did change at all. It seems that the original data are calculated correctly because at least the h1 outputs are the same. So I suspect that the differences of monthly-averaged h0 outputs are related to the calculation process of these two cases. However, it is hard to distinguish which one is correct? And how to correct the other wrong one so as to ensure the differences between these two cases are zero?

I attached the namelists of these two cases, and two figures of the comparative example results of h0-FSNR (they are different) and h1-FLNR (they are the same) between these two cases opened by ncview.

 

[nusbaume]

Hi Wendong Ge

Apologies for the slow response! Can you send us the model description info using the methods written here:

Information to include in help requests

When you are posting to the forums because you want help with an issue, please include as much of the following information as possible. Before submitting a help request, please check to see if your question is already answered: Search the forums for similar issues Check the CIME...

bb.cgd.ucar.edu

Also, can you send us the env_build.xml file for both the base and validate cases? That way we can make sure everything is configured properly.

Finally, I am moving this post over to the CAM-Chem forum, as that is the forum that is more closely watched by our PORT experts.

Thanks, and have a great day!

Jesse

 

[andrew]

I'm not sure, but I'm guessing that the h1 data are correct: PORT is probably doing something different with averaging than the regular model (PORT may not be running every time step, or it may not run the initial time step the same way). Unfortunately we have lost detailed support for PORT so I cannot go further than that. But if the h1 files are correct, I would say it is working correctly.

 

[CallanGwd]

I'm not sure, but I'm guessing that the h1 data are correct: PORT is probably doing something different with averaging than the regular model (PORT may not be running every time step, or it may not run the initial time step the same way). Unfortunately we have lost detailed support for PORT so I cannot go further than that. But if the h1 files are correct, I would say it is working correctly.

Thanks a lot! I totally agree with you! I also think that it is working correctly and h1 files are correct! I also speculate that the averaging process of PORT is somewhat different from that of other compsets precisely because it doesn't run every timestep due to the original input files of base case are not continuous.

These days my colleague made a comparative test. She ran the base case again but just change the nhtfrq from 73 to 18, and then ran the corresponding PORT validate case. The results showed that all the h1 data of 4 cases (Base & validate cases with nhtfrq = 73 & 18) are totally the same. And h0 data between the base case and validate case (nhtfrq = 18) are still different, but the differences are much smaller than those between those two cases with nhtfrq = 73!! (E.g., FSNR(base-73) = FSNR(base-18) = 391.26 W/m^2, FSNR(validate-73) = 428.447 W/m^2, FSNR(validate-18) =394.182 W/m^2).

So I guessed that the h0-data of PORT cases are directly related to the nhtfrq of base cases. The more frequent we output the result of base cases, the closer to the h0-data of base cases the PORT cases will be. Therefore I speculated that the h0-data of base cases are correct and those of PORT cases are calculated directly with the h1-data. And if we change nhtfrq to 1 directly, which means we output data of every timestep, maybe the h0-data of PORT cases will finally reach the goal.

However, if we really change the nhtfrq to 1, the output files of both cases will be too much and too latge, and we don't have enough storage to save them! What our study need now is just the valid monthly-average result.

 

[CallanGwd]

Hi Wendong Ge

Apologies for the slow response! Can you send us the model description info using the methods written here:

Information to include in help requests

When you are posting to the forums because you want help with an issue, please include as much of the following information as possible. Before submitting a help request, please check to see if your question is already answered: Search the forums for similar issues Check the CIME...

bb.cgd.ucar.edu

Also, can you send us the env_build.xml file for both the base and validate cases? That way we can make sure everything is configured properly.

Finally, I am moving this post over to the CAM-Chem forum, as that is the forum that is more closely watched by our PORT experts.

Thanks, and have a great day!

Jesse

Hi Jesse,

Thank you a lot! Here are my description of version info log and env_build.xml of these two cases. I also post the scripts of building processes of these two cases for reference if you need.

Besides, my colleague and I had some new findings and speculations these days and I wrote them in my reply to andrew.

Wendong