WPS Lambert Geogrid LonLat mismatch

[MarcTree]

Hi,

I'm getting an offset converting between long/lat mass grid coordinates and x/y synthesized geo coordinates in the geogrid file. In my example this is around 40-50m for a file with dx/dy of 1000m. Using geogrid version 4.2

Namelist parameters:
map_proj = 'lambert',
ref_lat = 35.085,
ref_lon = 130.95,
truelat1 = 35.085,
stand_lon = 130.95,

Resulting lambert variable in both the geogrid and output netcdf file, which is effectively the same as namelist inputs:
char Lambert;
:grid_mapping_name = "lambert_conformal_conic";
:latitude_of_projection_origin = 35.084991455078125; // double
:longitude_of_central_meridian = 130.9499969482422; // double
:standard_parallel = 35.084999084472656; // double
:earth_radius = 6371229.0; // double
:_CoordinateTransformType = "Projection";
:_CoordinateAxisTypes = "GeoX GeoY";

I get the same conversion to Lon/Lat using MIKE Zero python and MATLAB toolboxes and the following website:

CS2CS - Transform Coordinates On-line - MyGeodata Cloud

This on-line tool allows you to insert value pairs of geographic coordinates and transform them to different coordinate system or cartographic projection.

mygeodata.cloud

Using these parameters error is greater at the grid corners, indicating the earth radius is wrong. If I change the ellipsoid to a sphere of 6370000m the error is consistent at all points. For example, before and after:
+proj=lcc +lat_0=35.084991455078125 +lon_0=130.9499969482422 +lat_1=35.084999084472656 +lat_2=35.084999084472656 +a=6371229 +b=6371229 +no_defs
+proj=lcc +lat_0=35.084991455078125 +lon_0=130.9499969482422 +lat_1=35.084999084472656 +lat_2=35.084999084472656 +a=6370000 +b=6370000 +no_defs

After some iteration I get a nice conversion with the following projection string:
+proj=lcc +lat_0=35.08505847 +lon_0=130.95043037 +lat_1=35.08505847 +lat_2=35.08505847 +a=6370000 +b=6370000 +no_defs

So the real question is which coordinates do I trust? XLONG_M, XLAT_M and synthesized x/y using the iterated projection string above? Or x/y using the namelist parameters?
+proj=lcc +lat_0=35.085 +lon_0=130.95 +lat_1=35.085 +lat_2=35.085 +a=6370000 +b=6370000 +no_defs

My worry is XLONG/XLAT are synthesized incorrectly in the first place. Which coordinates are forcings interpolated onto for the WRF simulation?

 

[Ming Chen]

WPS assumes a spherical earth with a radius of 6370km.

I would suggest that you use the coordinate created by WPS for WRF simulation.