Re: [cf-satellite] Ang.: Making a proposal for the addition of the geos projection

Hi Martin,

On 11/23/11 4:34 AM, Martin Raspaud wrote:

Geostationary
-------------

grid_mapping_name = geos


Map parameters:

         latitude_of_projection_origin

         longitude_of_projection_origin

         perspective_point_height
        
        scale_x
        
        scale_y

        offset_x

        offset_y

         false_easting

         false_northing


Actually, we could use add_offset and scale_factor in the x and y
variables. x and y would then be integer pixel positions for example,
and scale and offset would transform them to viewing angles in radians.
The netcdf header would look like this (hand typed):

dimensions:
    y = 328;
    x = 580;
    reflective_bands = 5;
variables:
    short reflective_band_data(y, x, reflective_bands)
      :long_name = "Radiance of reflective bands of the SEVIRI instrument
(MSG)";
      :units = "milliwatts/m2/steradian/cm-1";
      :grid_mapping = "Projection";
      :coordinates = "y x";
    short x(x);
      :standard_name = "projection_x_coordinate";
      :add_offset = 1856
      :scale_factor = -208.16554260253906
      :units = "rad"
    short y(y);
      :standard_name = "projection_y_coordinate";
      :add_offset = 1856
      :scale_factor = -208.16554260253906
      :units = "rad"
    char Projection;
      :grid_mapping_name = "geos";
      :perspective_point_height = 35785831.;
      :semi_major_axis = 6378169.;
      :semi_minor_axis = 6356583.8;
      :latitude_of_projection_origin = 0.;
      :longitude_of_projection_origin = 0.;

What do you think ?
This is a neat idea, that way different conventions for north-to-south (GOES) 
south-to-north (SEVIRI) could be handled, ie. if I'm understanding your
suggestion?

Tom