[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Fwd: RE: [Ncwms-users] Selection of individual images from anaggregation?]



 Hi Paul:

CDM knows to convert the units to km. Not sure why ncWMS doesnt, but one guess is that we have had problems in trying to implement "deferred enhancement". the latest version of ncWMS (i think) has this problem "fixed" (by not doing deferred enhancement, and fixing some other bugs), so id be interested if you still see it in the latest ncWMS.

correcting the units is a good fix, though!

John

On 9/9/2010 7:03 AM, Paul Hamer wrote:
John,

When ToolsUI reads in this grid information it, somehow, correctly calculates the grid so that a WCS query will give you the correct bounding box. I can only assume that it doesn't use the x0, y0 since it couldn't possibly make that calculation if it did.

If you like you can download a sample file from

http://ngen-ciws.wx.ll.mit.edu/atomfeeds/ciwsDataFeed.atom

Look at Current CONUS Precip (VIL) Dataset - 1-KM Resolution for an example.

The ncWMS package uses this specification as you'd expect so the lat/lon bounding box is not the same and not the intended one.

A ncml wrapping of this file changing the units of x0, y0 to the correct unit (kilometers) solved the problem but not without much head scratching about the raw file not being correctly handled by ncWMS on my part even though it was obvious once I realized what was going on. No one ever called me quick. :-)

Paul.

John Caron wrote:
 Hi Paul:

Sorry I dont quite understand, can you expand a bit ?

On 9/5/2010 1:13 PM, Paul Hamer wrote:
It turns out that this is an error in the MIT-LL original dataset. For some reason ToolsUI ignores x,y dimensions provided (except in handling variable shape) and uses the grid mapping definition to derive the extent for the WCS. ncWMS on the other hand uses the x,y dimension literally from the mapping definition origin.

i.e.
 dimensions:
  y0 = 3520;
  x0 = 5120;

 variables:
  int grid_mapping0;
     :earth_radius = 6370997.0; // double
     :false_northing = 0.0; // double
     :false_easting = 0.0; // double
     :longitude_of_projection_origin = -98.0; // double
     :latitude_of_projection_origin = 38.0; // double
     :grid_mapping_name = "lambert_azimuthal_equal_area";
     :long_name = "Lambert Azimuthal Equal Area Projection";
     :_CoordinateTransformType = "Projection";
     :_CoordinateAxisTypes = "GeoX GeoY";

   double y0(y0=3520);
     :standard_name = "projection_y_coordinate";
     :long_name = "Distance from projection reference point latitude";
     :units = "meters";
     :_CoordinateAxisType = "GeoY";
   double x0(x0=5120);
     :standard_name = "projection_x_coordinate";
     :long_name = "Distance from projection reference point longitude";
     :units = "meters";
     :_CoordinateAxisType = "GeoX";


It took me a while so sorry if this caused any significant effort at your end.

Paul.


-------- Original Message --------
Subject: RE: [Ncwms-users] Selection of individual images from anaggregation?
Date: Tue, 24 Aug 2010 09:21:02 +0100
From: Jon Blower <address@hidden>
To: Paul Hamer <address@hidden>
CC: John Caron <address@hidden>
References: <address@hidden> <address@hidden> <address@hidden>

Hi Paul,

Thanks for this.  Yes, the Godiva2 client doesn't reflect all of its
changes in the page URL.  It probably should, to allow correct
back-button behaviour, but it's quite difficult to achieve.

Regarding the bounding-box error in ncWMS, this is a bit puzzling
because ncWMS simply calls a method on the Java-NetCDF libraries to get
the bbox, and I assumed that the same method is also used by ToolsUI.

I've copied John Caron in on this message.  John - ncWMS uses
GridCoordSystem.getLatLonBoundingBox() to get this information.  Is
there a better method we should be using?

Cheers, Jon

-
Paul Hamer

NOAA Earth System Research Laboratory (ESRL)
Global Systems Division (GSD)
325 Broadway R/GS5
Boulder, CO 80305-3328

phone : 303.497.6342