Steve,

I like this suggestion.  It's more forward-looking, just as easy to
implement, and still solves the problem that we were trying to solve
with "ocean_binary_mask",  which was to make it easy for providers to
make their data CF-compliant.

-Rich

On Mon, Jan 3, 2011 at 3:34 PM, Steve Hankin <[email protected]> wrote:
> Trial balloon:
>
> This conversation circles around the idea of masks that serve a
> discipline-specific purpose:  a land mask for terrestrial types; or a sea
> mask for ocean types.  Each discipline finds it natural to have "1" indicate
> valid points for his particular outlook.  It will always be an effort for
> the data providers in one discipline to adopt the conventions of another.
>
> One could imagine masks for other (less common) purposes as well in which
> the 1's signify other things.  For example there are valid uses for time
> masking, in which the 1's would indicate valid time indices.  Perhaps the
> fact that this conversation is occurring illustrates that we should be
> approaching masking in a discipline-neutral way -- defining a new attribute,
> and a more generic new standard_name.  Something like:
>
> netcdf mask_eg {
> dimensions:
>     AX003 = 10 ;
>     AX002 = 20 ;
> variables:
>     float LON_U(AX002, AX003) ;
>         LON_U:long_name = "curvilinear longitudes" ;
>         LON_U:units = "degrees_north" ;
>     float LAT_U(AX002, AX003) ;
>         LAT_U:long_name = "curvilinear latitudes" ;
>         LAT_U:units = "degrees_east" ;
>     float U(AX002, AX003) ;
>         U:coordinates = "LAT_U LON_U" ;
>         U:_FillValue = -1.e+34f ;
>         U:long_name = "Zonal Velocity" ;
>         U:units = "meters/sec" ;
>         U:binary_mask = "U_MASK";
>     float U_MASK(AX002, AX003) ;
>         U_MASK:coordinates = "LAT_U LON_U" ;
>         U_MASK:standard_name = "binary_mask" ; // "1" indicates valid
>         U_MASK:long_name = "Ocean mask" ;
>
> // global attributes:
>         :Conventions = "CF-1.5" ;
>
> }
>
> Is this a preferable approach?
>
>     - Steve
>
> ===========================
>
> On 1/3/2011 11:35 AM, Rich Signell wrote:
>
> Chris,
>
> On Mon, Jan 3, 2011 at 12:57 PM, Christopher Barker
> <[email protected]> wrote:
>
> On 1/2/11 6:11 PM, Rich Signell wrote:
>
> But they are not the same thing.  They are the inverse.
>
> yes, of course, but they carry exactly the same information, do they not.
>
> Yes, one could be inferred from the other.
>
> Why have two ways to express the same information?
>
>  Yes, it would
> be possible to have data sets providers create NcML for every ROMS
> dataset that has ever been written and serve the data with a
> land_binary_mask instead of a sea_binary_mask.
>
> well, I suppose it may be a question of whether there are more data
> providers or data consumers...
>
> Since most consumers use some kind of tool, I would says it's more a
> question of whether there are more data providers or more CF-compliant
> tool developers.   And since many tool developers use NetCDF-Java or
> some other package to enable CF compliance, perhaps there are really
> not so many software changes to be made.
>
> That also implies that there are a bunch of ROMS-output netcdf files that
> already have a sea_binary_mask variable, and are therefor not currently
> CF-compliant. Is that the case? Do we want to add things to the standard to
> make common, but not compliant, use cases compliant? Perhaps so.
>
> I think "Perhaps so" is exactly right.   The advantage of making it
> easier for providers to standardize their datasets vs the additional
> burden to CF-compliant tool developers.
>
> -Rich
>
>
> On Sun, Jan 2, 2011 at 12:31 PM, Chris Barker<[email protected]>
>  wrote:
>
> On 12/30/2010 2:40 PM, Rich Signell wrote:
>
> CF Standard Name Team:
>
> I would like to request a new standard_name="sea_binary_mask" defined as
>
> sea_binary_mask       X_binary_mask has 1 where condition X is met, 0
> elsewhere. 1 = sea, 0 = land.
>
> This is used by the popular ROMS ocean model, and perhaps others.
>
> The new "sea_binary_mask" would join the existing "land_binary_mask",
> which has 1 = land, 0 = sea.
>
> which makes it completely redundant. How hard it is to translate a
> sea_binary_mask into a land_binary mask?
>
> as an end user, now all my code has to look for both, despite them being
> the
> same thing.
>
> Isn't it an ideal to have only one standard way to express a given
> quantity?
>
> -Chris
>
>
>
>
>
> --
> Christopher Barker, Ph.D.
> Oceanographer
>
> Emergency Response Division
> NOAA/NOS/OR&R            (206) 526-6959   voice
> 7600 Sand Point Way NE   (206) 526-6329   fax
> Seattle, WA  98115       (206) 526-6317   main reception
>
> [email protected]
>
>
>
>



-- 
Dr. Richard P. Signell   (508) 457-2229
USGS, 384 Woods Hole Rd.
Woods Hole, MA 02543-1598
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata

Reply via email to