Prescribed aerosol in CAM5

[liurj04@gmail_com]

 Hi ! I would like to set up new files to prescribe the aerosols in CAM5.Now, I can run F_2000_CAM5_PM compset using CESM1.2.0, and I look at the input data file ( mam3_1.9x2.5_L30_2000clim_c130319.nc )of the prescribed aerosols. There are some variables in the file, such as XX_a1, XX_a1_logm and XX_a1_logv. I want replace the  aerosol concentrations by my own data calculated from atmosperic chemical transport model (GEOS-chem). If I changed the aerosolconcentrations in the input data file, how should I set other variables in the file, such as XX_a1_logm and XX_a1_logv  ? I can not find the method to set the  input data file. Could someone please help me with this? Thanks a lot!
Best Wishes ! 

 

[eaton]

The prescribed modal aerosol functionality uses the *_logm and
*_logv variables in the actual calculation of the mixing ratio
values.  These variables represent the monthly log mean and log
variance for each prescribed specie.  The calculation involves a
random sampling technique from the log mean distribution.  The
monthly mean mixing ratios that are in the prescribed MAM
datasets are not actually used.

This functionality was developed by folks at PNNL.  It was hoped
that it would provide a less expensive option for running with
the cam5 physics package.  However there does not seem to be any
active development work to push it forward, and the code that
produces the *_logm and *_logv output is not in the CAM
development repository.  I suspect that this functionality will
not be a supported CAM feature in future releases.  I suggest
looking into using the prescribed bulk aerosols with the cam5
physics package.

 

[eaton]

The prescribed modal aerosol functionality uses the *_logm and
*_logv variables in the actual calculation of the mixing ratio
values.  These variables represent the monthly log mean and log
variance for each prescribed specie.  The calculation involves a
random sampling technique from the log mean distribution.  The
monthly mean mixing ratios that are in the prescribed MAM
datasets are not actually used.

This functionality was developed by folks at PNNL.  It was hoped
that it would provide a less expensive option for running with
the cam5 physics package.  However there does not seem to be any
active development work to push it forward, and the code that
produces the *_logm and *_logv output is not in the CAM
development repository.  I suspect that this functionality will
not be a supported CAM feature in future releases.  I suggest
looking into using the prescribed bulk aerosols with the cam5
physics package.

 

[liurj04@gmail_com]

Hi, eatonThanks for your answer.As I understand, you said the *_logm and *_logv variables are used to calculate the mixing ratio values. But the mixing ratios in the prescribed MAM datasets are not used . So I think if I run  the prescribed MAM CAM5 , shoud I just replace the aerosol concentrations by my own data and keep the *_logm and *_logv not changed ?I am also trying to use the prescribed bulk aerosols with the cam5 physics package.
Thank you a lot ! Best Wishes !

 

[liurj04@gmail_com]

Hi, eatonThanks for your answer.As I understand, you said the *_logm and *_logv variables are used to calculate the mixing ratio values. But the mixing ratios in the prescribed MAM datasets are not used . So I think if I run  the prescribed MAM CAM5 , shoud I just replace the aerosol concentrations by my own data and keep the *_logm and *_logv not changed ?I am also trying to use the prescribed bulk aerosols with the cam5 physics package.
Thank you a lot ! Best Wishes !

 

[eaton]

If you replace the mixing ratios with your own data, but leave the *_logm and *_logv variables unchanged, then the simulation will be identical to one run with the original datasets.Note that to use prescribed bulk aerosols with the cam5 physics package you need to set the namelist variable prescribed_aero_model='bulk' since the default for cam5 is to use prescribed modal aerosols. 

 

[eaton]

If you replace the mixing ratios with your own data, but leave the *_logm and *_logv variables unchanged, then the simulation will be identical to one run with the original datasets.Note that to use prescribed bulk aerosols with the cam5 physics package you need to set the namelist variable prescribed_aero_model='bulk' since the default for cam5 is to use prescribed modal aerosols. 

 

[liurj04@gmail_com]

Now, I can run CAM5 using  the  prescribed bulk aerosols.I run CESM1_2_0 with FC5 compset and set the model run as follows: set CAM_CONFIG_OPTS to "-phys cam5 -chem none" in  env_build.xmladd  prescribed_aero_model='bulk'  in  user_nl_cam

 

[liurj04@gmail_com]

Now, I can run CAM5 using  the  prescribed bulk aerosols.I run CESM1_2_0 with FC5 compset and set the model run as follows: set CAM_CONFIG_OPTS to "-phys cam5 -chem none" in  env_build.xmladd  prescribed_aero_model='bulk'  in  user_nl_cam

 

[liurj04@gmail_com]

Hi, eatonI looked the introduction about the prescribed bulk aerosol input data file : aero_1.9x2.5_L26_1850-2005_c091112.nc.It says " Processing to create 1850-2005 dataset for 20th century transient runs from the decadal avg data provided by Jean-Francois Lamarque.12 November 2009, eaton"Can you give me some suggestions about  how this input aerosol dataset created ? and where can I get the input aerosol data simulated from NCAR CAM-Chem (Lamarque et al., 2012: CAM-chem: description and evaluation of interactive atmospheric chemistry in CESM)

 

[liurj04@gmail_com]

Hi, eatonI looked the introduction about the prescribed bulk aerosol input data file : aero_1.9x2.5_L26_1850-2005_c091112.nc.It says " Processing to create 1850-2005 dataset for 20th century transient runs from the decadal avg data provided by Jean-Francois Lamarque.12 November 2009, eaton"Can you give me some suggestions about  how this input aerosol dataset created ? and where can I get the input aerosol data simulated from NCAR CAM-Chem (Lamarque et al., 2012: CAM-chem: description and evaluation of interactive atmospheric chemistry in CESM)

 

[eaton]

The data I started from was monthly, 10-year averages.  The
monthly averages labeled 1855 each represent an average of the
corresponding month for the years 1850-1859.  Months labeled 1865
are averages of the period 1860-1869, and so on through
1995 (years 1990-1999).  The final averaging period is years
2000-2009 and this part of the simulation was done holding
emissions constant at year 2000 values.  So this final 10-year
average represents a year 2000 climatology.

The 1855 climatology was also used for year the first year in the
dataset, 1849.  So the annual cycle will be the same for years
1849 through 1855.

The years labeled 2000 and 2006 in the dataset are both from the
2000 climatology.  So the annual cycle will be the same for years
2000 through 2006.

All the data I had access to is in the file
aero_1.9x2.5_L26_1850-2005_c091112.nc.  I essentially just added
a time coordinate to this data as described above.  Further
information about the simulations that produced the decadal
averages would need to come from the authors of the paper you
referenced.

 

[eaton]

The data I started from was monthly, 10-year averages.  The
monthly averages labeled 1855 each represent an average of the
corresponding month for the years 1850-1859.  Months labeled 1865
are averages of the period 1860-1869, and so on through
1995 (years 1990-1999).  The final averaging period is years
2000-2009 and this part of the simulation was done holding
emissions constant at year 2000 values.  So this final 10-year
average represents a year 2000 climatology.

The 1855 climatology was also used for year the first year in the
dataset, 1849.  So the annual cycle will be the same for years
1849 through 1855.

The years labeled 2000 and 2006 in the dataset are both from the
2000 climatology.  So the annual cycle will be the same for years
2000 through 2006.

All the data I had access to is in the file
aero_1.9x2.5_L26_1850-2005_c091112.nc.  I essentially just added
a time coordinate to this data as described above.  Further
information about the simulations that produced the decadal
averages would need to come from the authors of the paper you
referenced.

 

[liurj04@gmail_com]

 Thanks a lot for your detailed answer, I really appreciate it.I am still a little not sure , is the input aerosol data simulated from NCAR CAM-Chem (Lamarque et al., 2012) ?

 

[liurj04@gmail_com]

 Thanks a lot for your detailed answer, I really appreciate it.I am still a little not sure , is the input aerosol data simulated from NCAR CAM-Chem (Lamarque et al., 2012) ?

 

[eaton]

I got the data from Jean-Francois, and know that it came from simulations he did using prognostic bulk aerosols.  I don't know the details of those simulations.  Perhaps Simone knows whether the reference you point out is the correct one.

 

[eaton]

I got the data from Jean-Francois, and know that it came from simulations he did using prognostic bulk aerosols.  I don't know the details of those simulations.  Perhaps Simone knows whether the reference you point out is the correct one.

 

[liurj04@gmail_com]

Hi, eatonBy comparing the two dataset , I found that the prescribed bulk aerosol input data is equal to the dataset  simulated from NCAR CAM-Chem (Lamarque et al., 2012) Best Wishes !

 

[liurj04@gmail_com]

Hi, eatonBy comparing the two dataset , I found that the prescribed bulk aerosol input data is equal to the dataset  simulated from NCAR CAM-Chem (Lamarque et al., 2012) Best Wishes !