Hi Jonathan,
I wasn't very clear in the earlier mail. Actually, the quality flags
are different, but the attributes flag_values and flag_meanings are same
for all the flag variables. These attributes are being repeated within
each quality flag variable instead of perhaps having a single
"definition" of the flag_values and flag_meanings.
One solution we looked at is to have a "container variable" for flag
definitions which will be referred to by each of the flag variables,
something like this:
float temperature(z);
....................
temperature:ancillary_variables = "temperature_WODflag";
int temperature_WODflag(z);
....................
temperature:flag_definitions = "WOD_flag_definitions";
float salinity(z);
....................
salinity:ancillary_variables = "salinity_WODflag";
int salinity_WODflag(z);
....................
salinity:flag_definitions = "WOD_flag_definitions";
<similarly for other other variables>
int WOD_flag_definitions;
WOD_flag_definitions:flag_values = "0, 1, 2, 3, 4, 5, 6, 7, 8, 9" ;
WOD_flag_definitions:flag_meanings = "accepted_value range_outlier
failed_inversion_check failed_gradient_check
observed_level_bullseye_flag_and_zero_gradient_check combined_gradient
_and_inversion_checks failed_range_and_inversion_checks
failed_range_and_gradient_checks
failed_range_and_questionable_data_checks
failed_range_and_combined_gradient_and_inversion_checks" ;
Of course, flag_definitions is not a CF attribute and the above solution
wouldn't work with CF based software. Is there a way around within the
ambit of CF other than to repeat the flag definitions?
Upendra
On 11/1/2011 11:38 AM, Jonathan Gregory wrote:
> Dear Upendra
>
>> On 11/1/2011 10:47 AM, Upendra Dadi wrote:
>>> The same issue occurs with World Ocean Database which consists of
>>> mainly profile data. Each profile typically consists of several
>>> variables measured along the depth. The quality flags used for all
>>> the variable are same.
>>> On 10/31/2011 12:12 PM, Randy Horne wrote:
>>>> The current CF conventions dictate that quality flags are
>>>> attached to specific variables. The implication is that
>>>> comforming with CF conventions would require the same quality
>>>> flags to be stored multiple times in our NetCDF product files.
> Quality flags are attached to variables using the ancillary_variables att of
> the data variable. If several data variables had the same quality flags and
> dimensions, they could all point to the same quality variable. Perhaps the
> problem is that the different variables have different standard names, and
> this means the quality variables would also have different standard names
> (and therefore could not be the same variable)? If that is the problem, perhaps
> we could find a way round it. Or have I missed the point?
>
> Best wishes
>
> Jonathan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <
http://mailman.cgd.ucar.edu/pipermail/cf-metadata/attachments/20111101/e7b92288/attachment-0001.html>
Received on Tue Nov 01 2011 - 10:44:49 GMT
