Its clear once you understand it. Im just reporting that others continue to misread it.
Its an interesting question about "units", I was thinking that one could say that the "unpacked units" = scale_factor * "packed units", eg .001 * m/s is a perfectly good udunit. The I realized that we dont say anything about units, and I dont know if its clear which units attribute should be put into the file ??!! Common usage is, im pretty sure, "unpacked units", but we should probably make that explicit.
Brian Eaton wrote:
> The sentence in question is:
>
>
>>"The missing values of a variable with scale_factor and/or
>>add_offset attributes (see section 8.1) are interpreted relative to
>>the variable's external values, i.e., the values stored in the
>>netCDF file. "
>
>
> I agree that the term "external values" would be ambiguous by itself. But
> the sentance doesn't stop there. The phrase is immediately clarified by
> stating that the external values are the ones in the netCDF file. I'm
> having a hard time seeing what's ambiguous about that.
>
> The use of "packed encoding" is not as precise as what's intended here.
> Packing is only one application of scale_factor/add_offset. There are
> others that don't involve reducing the data size.
>
> Brian
>
>
> On Fri, Oct 20, 2006 at 10:02:12AM -0700, Steve Hankin wrote:
>
>>Jon,
>>
>>It certainly does look like the wording as-is (as you have quoted) is
>>self-contradictory. But the phrase "units of the packed values", while
>>intuitively clear, is actually a mis-statement, since "units" do not
>>really apply to the packed data. The right wording is tough to find ....
>>
>>How about
>>
>> "When a variable uses a packed encoding (i.e. the variable has
>> scale_factor and/or add_offset attributes -- see section 8.1)
>> missing values are interpreted in the packed encoding. In other
>> words, grid cells with missing values are detected before the data
>> are unpacked."
>>
>>Not entirely happy with this, either ...
>>
>> - Steve
>>
>>=========================================
>>
>>John Caron wrote:
>>
>>>A user has misread section 2.5.1 on how missing values interact with
>>>scale/offset, resulting in incorrectly written files. We might want to
>>>consider some rewording on this in the future.
>>>
>>>Here is the relevent conversation:
>>>
>>>
>>>>>"The missing values of a variable with scale_factor and/or
>>>>>add_offset attributes (see section 8.1) are interpreted relative to
>>>>>the variable's external values, i.e., the values stored in the
>>>>>netCDF file. "
>>>>>
>>>>>If I change this in my head to
>>>>>
>>>>>"The missing values of a packed variable are in the units of the
>>>>>packed values, i.e., the values stored in the netCDF file."
>>>
>>>>I was confused by the "relative to the variable's external values".
>>>>In my mind the "external values" are the ones that an external app
>>>>sees, i.e. the unpacked values. Your interpretation makes perfect
>>>>sense though.
>>>
>>>_______________________________________________
>>>CF-metadata mailing list
>>>CF-metadata at cgd.ucar.edu
>>>http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
>>
>>--
>>--
>>
>>Steve Hankin, NOAA/PMEL -- Steven.C.Hankin at noaa.gov
>>7600 Sand Point Way NE, Seattle, WA 98115-0070
>>ph. (206) 526-6080, FAX (206) 526-6744
>>
>
>
>>_______________________________________________
>>CF-metadata mailing list
>>CF-metadata at cgd.ucar.edu
>>http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Fri Oct 20 2006 - 12:30:41 BST
