Prescribed O3 at sub-monthly timescales

[sean_m_davis@noaa_gov]

Is it possible to prescribe O3 in CAM or WACCM-SC at sub-monthly timescales?It's easy to do this using monthly-mean o3 data, but i'd like to know if it could be done at a finer time resolution, perhaps twice per month. 

 

[mmills]

Sean,My understanding is that, as with other input files, values are interpolated at times between specified dates. So you can pick a time cadence arbitrarily. In the case of monthly specified chemistry, values are specified on the 15th day of each month, and interpolated at times in between. For twice monthly, you could pick values on days 1 and 15 of each month, or whatever you like.Mike

 

[sean_m_davis@noaa_gov]

I'm confused about the "time" variable, and whether or not it is really read in, or whether it is actually the "date" variable that is read in by CAM/WACCM

The 3 time variables have values and metadata like this:

date -- 115, 214, 315, 415, ...
datesec -- 0, 0, 0, 0, ...
time -- 1, 2, 3, 4, ...

    int date(time) ;
        date:long_name = "current date (YYYYMMDD)" ;
    int datesec(time) ;
        datesec:long_name = "current seconds of current date" ;
    double time(time) ;
        time:long_name = "time" ;
        time:units = "month number" ;
        time:calendar = "noleap" ;
        time:bounds = "time_bnds" ;

I am trying to specify monthly-mean values, with the exception that I want two values for september -- one for the first and  one for the second half of the month

So I think my variables should look like this:

time=[1.0d,2,3,4,5,6,7,8,9,9,10,11,12]
date=[115,214,315,415,515,615,715,815,907,922,1015,1115,1215]
datesec =[0,0,0,0,0,0,0,0,0,0,0,0,0]

However, I'm concerned that something could get messed up if the "time" variable is actually used, since there are two entries with the value 9.

Should I make it something like 9, 9.5 instead?

 

[eaton]

I assume we're talking about the prescribed ozone dataset.  The code which handles interpolating that dataset is src/chemistry/utils/tracer_data.F90.  That code uses the date and datesec variables to determine the timestamps for the samples in the file.  The use of date/datesec is a local convention, so we always add the time coordinate to allow utilities that understand the CF metadata conventions to correctly interpret the file.  See the CF convention for a description of the time coordinate, or just follow the example of existing files.