Skipping mapping of lon to [0:360] [?]

[Ian Culverwell]

https://trac.romsaf.org/ropp/changeset/5427/ropp_src/branches/dev/Share/dmi_trunk_9.0/ropp_io/tools/grib2bgrasc.f90

Original changesets:

Skipping mapping of lon to [0:360]: https://trac.romsaf.org/ropp/changeset/5102/ropp_src/branches/dev/Share/dmi_trunk_8.1/ropp_io/tools/grib2bgrasc.f90

[Ian Culverwell]

As Stig says, in ROPP longitudes lie in the range [-180, 180] deg. Neither of the two longitude 'wrappings' in grib2bgrasc.f90 does the job, as this quick test shows:

lon_in1, MOD(lon_in1+180.0_wp,360.0_wp)-180.0_wp, MOD(lon_in1+360.0_wp,360.0_wp), ATAN2(SIN(dtor*lon_in1),COS(dtor*lon_in1))/dtor = 
   -3.60000E+02   -3.60000E+02    0.00000E+00    1.40334E-14
   -2.70000E+02   -2.70000E+02    9.00000E+01    9.00000E+01
   -1.80000E+02   -1.80000E+02    1.80000E+02   -1.80000E+02
   -9.00000E+01   -9.00000E+01    2.70000E+02   -9.00000E+01
    0.00000E+00    0.00000E+00    0.00000E+00    0.00000E+00
    9.00000E+01    9.00000E+01    9.00000E+01    9.00000E+01
    1.80000E+02   -1.80000E+02    1.80000E+02    1.80000E+02
    2.70000E+02   -9.00000E+01    2.70000E+02   -9.00000E+01
    3.60000E+02    0.00000E+00    0.00000E+00   -1.40334E-14

Both formulas generate longitudes outside the range [-180, 180] deg. What we need is the last column, which is ATAN2(SIN(lon),COS(lon)), aka the principal value of the argument of e^{i lon}.

[Ian Culverwell]

(Actually the line

    grib1%georef%lon   = mod(grib1%georef%lon+180.0_wp,360.0_wp) - 180.0_wp 

, just before Sec 5.3.2, is not in the official ROPP development branch.)

[Ian Culverwell]

Using the ATAN2 formula in the code doesn't affect the longitudes in the netCDF file, which are (usually) range-checked into [-180, 180] anyway, and therefore equal -3.53 deg in the core test. (Note, however, that #562 makes range-checking optional.) But the output namelist is changed from

 LON     =   356.470000000000     ,

to

 LON     =  -3.53000000000000     ,

, so the reference file needs to be changed.

Changes committed at r6234.

[Ian Culverwell]

While we're here, replace the risky

  gribfile_lun = 12

by the safer

  gribfile_lun = Get_IO_Unit()

in grib2bgrasc.f90 at r6235.

Closing ticket.