Re: EUMETSAT's NetCDF Query

On 2/1/2012 11:46 PM, Peter Miu wrote:
> Hi John,
>
> Thanks for the fast reply.
>
> I thought about this a bit more and its not really the parent classes of the 
> MSG15natIOSP class (that EUMETSAT wrote) that needs to be triggered to create 
> these variables as this IOSP is only responsibility for convert the native 
> format to the netCDF format.
>
> Its actually the netCDF classes for reading the resultant MSG15 netCDF file 
> that are required to be triggered to create these 3 variables based on the 
> contents of the MSG15 netCDF file.  I assume this code is generic (Unidata does 
> not hardcode condition for the MSG15 netCDF format or can you tell me something 
> here ... ?), the 3 variables must be created by the IOSP (in this case 
> MSG15natIOSP) to tell the Unidata generic netCDF code what the projection for 
> the data is in the netCDF file.  Does this make sense to you ?

yes

> If so, then the confusion from us was that we expected the generic netCDF code 
> to know the MSG projection simply by specifying the following:
>
>      short chan1;
>        :grid_mapping = "Space_View_Perspective_or_Orthographic";
>
>      char Space_View_Perspective_or_Orthographic;
>        :grid_mapping_name = "MSGnavigation";
>        :longitude_of_projection_origin = 0.0; // double
>        :latitude_of_projection_origin = 0.0; // double
>        :semi_major_axis = 6356755.5; // double
>        :semi_minor_axis = 6378140.0; // double
>        :height_from_earth_center = 4.2163970098E7; // double
>        :scale_x = 35785.830098; // double
>        :scale_y = -35785.830098; // double
>        :_CoordinateTransformType = "Projection";
>        :_CoordinateAxisTypes = "GeoX GeoY";
>
> We did not define the x and y projection arrays to define the km areas of the 
> projection.
its the combination of the projection and the x,y coordinates thats 
needed. if you look at the 
ucar.unidata.geoloc.projection.sat.MSGnavigation code you see that it 
takes an x,y coordinate and translates to lat, lon. the x,y coordinates 
have to correspond to the grid points.

does that make sense?

> I've spoken to Michael and Jags on this, I have ask Jags to add a generic 
> MSGProjection class to the called by all EUMETSAT netCDF IOSP that will create 
> these 3 variables in the netCDF format.  We can then test whether ToolsUI and 
> IDV can display the resultant netCDF formats with the correct projection.  What 
> do you think?

im not sure if thats the same as the existing

ucar.unidata.geoloc.projection.sat.MSGnavigation

class?

> One final questions, how do we plug-in IOSP code into the ToolsUI so that it 
> can understand the EUMETSAT MSG15 native data sets ?

http://www.unidata.ucar.edu/software/netcdf-java/reference/RuntimeLoading.html