Re: GeoX and GeoY units

• To: Michael Toews <mwtoews@xxxxxx>
• Subject: Re: GeoX and GeoY units
• From: John Caron <caron@xxxxxxxxxxxxxxxx>
• Date: Mon, 27 Feb 2006 16:21:01 -0700

Hi Michael:

It is sort of a bug. If you look at the x,y data variables they stay in meters. 
When the georeferencing code (GeoCoordSys) gets called, it needs to convert to 
km, because the projection routines use km. The CoordSysTable does that 
conversion as a side-effect that really shouldnt be there, i will look at 
removing that.

Michael Toews wrote:

Hi John,
I have uploaded a similar, but fresher file to:

http://www.sfu.ca/~mwtoews/annual_oliver_0.2_obs.nc.gz

I have changed my 'proj4string' (from the metadata in my last email) to 'WTK' for Well Known Text, which is used by several open and closed source GIS packages, including the PROJ.4 library, to describe projections of gridded spatial elements. The WTK was generated by GDAL within Python by mocking the properties of a GIS raster. I have only been creating NC files for the past month, so if there is anything that is incorrect, please let me know.

WKT I think is a reasonable way to go. Im not much of an expert in it, though, 
I think I just followed an examle also generated by GDAL.

We (Rich Signell and others) have been pushing CF to specify Datum etc, so far 
without much luck. The best thing to do is for you to post your thoughts and 
needs to the CF group, eventually we will probably form a subgroup to bless 
some proposal.

You file looks correct.

Thanks.

+mt

John Caron wrote:

Hi Michael: can you make put your file on ftp, so i can have a look? thanks

Michael Toews wrote:

Hi,

I'm wondering why NetCDFTools 2.2 in the CoordSys tab, the units for my projected GeoX and GeoY dimensions are displayed as "km" and not as "m" as I have specified? Shouldn't this be a bug? I realize some people use km for spatial dimensions, however a vast number of GIS users use meters for Eastings and Northings. The current CF-1.0 conventions do not specify that the spatial units need to be in "km" or any particular unit (or at least what I read).

Also, do you know if there are there any plans for CF conventions to recommend a horizontal datum? I've included the PROJ.4 string in my file to include this information, since I need meter accuracy. I'm curious how ESRI will handle the NetCDF format in the pending ArcGIS 9.2 release, specifically if they will make new convention, or adopt an existing one.

I'm also slightly intrigued at the "regular" fields in NetCDFTools. I have 1 that was automatically generated for my xdim, which is a strange looking number. This should be exactly 100.0 m (and so should the ydim). How is this generated, and why for one var, but no the other?

Thanks for any help.

+mt