-------- Original Message --------
Subject: RE: interpretation of missing_value by ncWMS
Date: Wed, 25 Apr 2012 11:05:04 +0100
From: Gaffney, Sean P. <sgaf@xxxxxxxxxx>
To: John Caron <caron@xxxxxxxxxxxxxxxx>, Jon Blower
<j.d.blower@xxxxxxxxxxxxx>
Hi John and Jon,
Thank you both for your replies. With your information, I was able to
re-examine the files in a bit more detail and I discovered that the problem was
that the values for missing_value in the header and the actual missing data
values in the file weren't the same, so they were being treated as real data.
This is therefore not a CF issue but a file generation issue and I'll get back
to the originator and let them know. It does further my resolve to somehow come
up with a checking mechanism that can automatically assess these sorts of
issues though.
Cheers
Sean
-----Original Message-----
From: John Caron [mailto:caron@xxxxxxxxxxxxxxxx]
Sent: 24 April 2012 19:40
To: Jon Blower
Cc: Gaffney, Sean P.; Unidata netCDF Java Support
Subject: Re: interpretation of missing_value by ncWMS
Hi all:
missing_value, _FillValue are both mapped to NaNs
details are here:
http://www.unidata.ucar.edu/software/netcdf-java/v4.2/javadoc/ucar/nc2/dataset/EnhanceScaleMissing.html
send me an example file if you think somethings not working.
John
On 4/24/2012 11:00 AM, Jon Blower wrote:
Hi Sean,
I must admit I hadn't appreciated the semantic distinction between
missing_value and _FillValue. However, I assumed that ncWMS would treat both
of these essentially the same and recognize missing_value as data outside the
dataset. The Java-NetCDF libs automatically recognise these and convert data
to NaN (I thought).
I've copied to John Caron, who can hopefully comment on whether Java-NetCDF
treats the attributes differently.
Cheers, Jon
-----Original Message-----
From: Gaffney, Sean P. [mailto:sgaf@xxxxxxxxxx]
Sent: 24 April 2012 10:20
To: Jon Blower
Subject: interpretation of missing_value by ncWMS
Hi Jon,
I've just found out from John Caron that the attribute missing_value is not
being deprecated in the CF conventions so is an acceptable CF attribute. A lot
of the feedback I've had from the community has been that _FillValue should
only be used to define the actual default value used to generate the file
structure before it was populated, and that if there are any actual absent data
values, these should be indicated using missing_value. I'd puzzled over this
because I thought the missing_value attribute was being lost, but this
obviously is no longer the case.
My understanding of how the ncWMS works at the moment is that it doesn't
recognise missing_value as data outside the dataset - I've had this problem
with the data that Helen sent me, where she had left out _FillValue but
supplied missing_value and the points covering land surface weren't being made
transparent.
Therefore, my question to you is, can the ncWMS be made to treat missing_value
in the same way it treats _FillValue, so that if it encounters either one, it
will regard them as a NaN and make the cell of the model transparent for
visualisation purposes?
Cheers
Sean
--------------------------------------------------------------------------------
Sean Gaffney
Data Scientist
British Oceanographic Data Centre
Joseph Proudman Building
6 Brownlow Street
Liverpool
L3 5DA
UK
+44 (0)151 795 4950
--
This message (and any attachments) is for the recipient only. NERC is subject
to the Freedom of Information Act 2000 and the contents of this email and any
reply you make may be disclosed by NERC unless it is exempt from release under
the Act. Any material supplied to NERC may be stored in an electronic records
management system.
