O3 output from CAM/CCSM

[klaus_wyser@smhi_se]

I have a problem with the O3 output in an IPCC scenario run with CCSM. I read time-varying O3 data from the file B1.ozone.128x64_L18_1991-2100_c040521.nc and set ozncyc=.false. in the atm namelist. The logfile tells that the file is opened and a new value is read once a month. The ozone data in the input file varies seasonally and there is a long-term trend. However, the O3 output from CCSM only shows the annual cycle but no trend with time (see fig).

I redid the same type of experiment with CAM stand-alone and now I get the variation of O3 in time. Why does O3 vary in CAM standalone but not in CCSM, given the same inputfile? Where is the difference between CAM and CCSM when it comes to ozone? Is it possible that the ozone data within CCSM are correct (at least the logfile tells that new data are read in every month) and it's only the output that's wrong? Anybody knows?

Note that CCSM can save both O3 and O3VMR, both show the same problem. CAM only saves O3VMR.

Any help is greatly appreciated.
Regards,
Klaus



Explanation to attached figure:
Global average of vertical average O3 mixing ratio for pressure below and above 300 hPa. Input (blue) denotes data from B1.ozone.128x64_L18_1991-2100_c040521.nc and output (red) is from the CCSM run. The 2 datasets have a bias because of different vertical resolution and the crude method of averaging. Despite of this, the output data obviously show no long-term trend which is puzzling

 

[klaus_wyser@smhi_se]

Sorry, the figure got lost in my previous post, here it is.

Explanation to attached figure:
Global average of vertical average O3 mixing ratio for pressure below and above 300 hPa. Input (blue) denotes data from B1.ozone.128x64_L18_1991-2100_c040521.nc and output (red) is from the CCSM run. The 2 datasets have a bias because of different vertical resolution and the crude method of averaging. Despite of this, the output data obviously show no long-term trend which is puzzling

 

[pjr]

Hi Klaus,

CAM and CCSM are outfitted with a set of namelist variables that let them either recycle input datasets (eg repeat the annual cycle) or progress through a series of years. I think you can also tell the models terminate if you run off the end of the series or repeat the last year. I would review the namelist variable options for the model, and make sure that you are using the same namelist settings for the CAM and CCSM runs.

 

[klaus_wyser@smhi_se]

Hej Phil---

Thanks for the reply. Whether the ozone input file is read successively or only the first year (that is recycled) is determined by the namelist variable ozoncyc. I have triple-checked that we have ozncyc=false in our CCSM runs, exactly as it is in the CAM run. The logfiles show that the ozone input file is read for the year that I specify, I can't see any difference between CAM and CCSM logfiles. But still the O3VMR output behaves differently: I find a clear difference in O3VMR between a run for year 2000 and a run for year 2049 with CAM, but O3VMR is exactly the same for the 2 years with CCSM. It doesn't help if I run the full period from 2000 to 2049 with CCSM (see figure), ozone doesn't seem to vary with time (except for an annual cycle).

I don't know whether this really is a problem or not. Since the logfile states that ozone is read for the year and month given, it is possible that the correct O3 data is used in the calculations. The error could be simply in the output routines that don't write the actual concentration but are locked to some climatology or default setting. I have noted that the ozone output differs slightly for CAM and CCSM, CAM gives only O3VMR while I can have both O3 and O3VMR in the output from CCSM. This means that somebody must have added or modified the routine that prints out the ozone (or accumulates the monthly means) in CCSM. This is just a hypothesis and I would like somebody who knows the model better to have a look at it.

Klaus