[netcdf-java] Reduced Gaussian to Regular Lat/Lon Interpolation has odd spacing

Hi all,

we’re using the netcdf lib to open Grib Files with Reduced Gaussian grids. 
Those are automatically interpolated to Regular Lat Lon by QuasiRegular.java as 
far as I got it. Also as far as I understood that code, I would expect equal 
spacings between the gridpoints. 

However: what we get out looks like this…

Latitude Dimension starting at Greenwich:
0.000000000°
0.140624851°
0.281249702°
0.421874553°
0.562499404°
0.703124285°
0.843749106°
0.984373927°
1.124998808°
1.265623689°
1.406248569°
1.546873331°
1.687498212°
1.828123093°
1.968747854°
2.109372854°
2.249997616°
2.390622377°
2.531247377°
2.671872139°
2.812497139°
2.953121901°
3.093746662°
3.234371662°
3.374996424°
3.515621185°
3.656246185°
3.796870947°
3.937495708°
4.078120708°
4.218745708°
4.359370232°
4.499995232°
4.640620232°
[…]

…which results to a spacings as follows:
0.140624851°
0.140624851°
0.140624851°
0.140624851°
0.140624881°
0.140624821°
0.140624821°
0.140624881°
0.140624881°
0.140624881°
0.140624762°
0.140624881°
0.140624881°
0.140624762°
0.140625°
0.140624762°
0.140624762°
0.140625°
0.140624762°
0.140625°
0.140624762°
0.140624762°
0.140625°
0.140624762°
0.140624762°
0.140625°
0.140624762°
0.140624762°
0.140625°
0.140625°
0.140624523°
0.140625°
0.140625°
0.140624523°
[…]

…similar for longitude, starting far north:
89.89239502°
89.75300598°
89.61279297°
89.47238922°
89.33191681°
89.19141388°
89.05088806°
88.91035461°
88.76980591°
88.6292572°
88.4887085°
88.34815216°
88.2075882°
88.06703186°
87.9264679°
87.78591156°
87.6453476°
87.50478363°
87.36421967°
87.22364807°
87.08308411°
86.94252014°
86.80194855°
86.66138458°
86.52082062°
86.38024902°
86.23968506°
86.09911346°
85.9585495°
85.81797791°
85.67741394°
85.53684235°
85.39627838°
85.25570679°
[…]

…spacings:
-0.139389038°
-0.140213013°
-0.140403748°
-0.140472412°
-0.14050293°
-0.140525818°
-0.140533447°
-0.140548706°
-0.140548706°
-0.140548706°
-0.140556335°
-0.140563965°
-0.140556335°
-0.140563965°
-0.140556335°
-0.140563965°
-0.140563965°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140571594°
-0.140563965°
-0.140571594°
-0.140571594°

…a bit too much for being just rounding issues.

Does anyone know why the spacing isn’t equal?

BTW: used latest netcdfAll.jar V 4.3.21 and the code is as follows:
NetcdfFile grib = NetcdfFile.open("dsd_201201011200_120.grb");
NetcdfDataset nds = new NetcdfDataset(grib);
GridDataset gds = new GridDataset(nds);

for (GridDatatype var : gds.getGrids()) {
        if ("2_metre_temperature_surface".equals(var.getFullName())) {
                GridCoordSystem gcs = var.getCoordinateSystem();
                CoordinateAxis lon = gcs.getXHorizAxis();
                for (double value : (double[]) 
lon.read().get1DJavaArray(double.class)) {
                        System.out.println(value);
                }
                System.out.println("...........");
                CoordinateAxis lat = gcs.getYHorizAxis();
                for (double value : (double[]) 
lat.read().get1DJavaArray(double.class)) {
                        System.out.println(value);
                }
        }
}               

Thanks a lot in advance & happy easter! ;-)

Jochen

Attachment: smime.p7s
Description: S/MIME cryptographic signature