Multifile vs single file processing differences in ropp_fm_bg2ro_1d

[Ian Culverwell]

There are some differences in the the output, esp of WARNING messages, when ropp_fm_bg2ro_1d is run on single files and multifiles. See attached GRAS_xxx_1B_M02_20141115142313Z_20141115142727Z_N_O_20141115152139Z.bg.nc (singlefile) and GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.bg.nc (multifile), and their associated outputs and run script.

In addition, undulations seem to be handled differently in the different profiles of a multifile, so that this input

 /data/nwp1/idculv/ROPP/Work/ROPP8.0/ropp_fm
idculv@eld037:> ncks -H -vundulation GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.bg.nc
dim_unlim[0] undulation[0]=-9999.9 
dim_unlim[1] undulation[1]=-9999.9 

gives this output

/data/nwp1/idculv/ROPP/Work/ROPP8.0/ropp_fm
idculv@eld037:> ncks -H -vundulation GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.fw.nc
dim_unlim[0] undulation[0]=0 
dim_unlim[1] undulation[1]=-9.9999e+07 

This is wrong.

[Ian Culverwell]

GRAS_xxx_1B_M02_20141115142313Z_20141115142727Z_N_O_20141115152139Z.bg.nc

[Ian Culverwell]

GRAS_xxx_1B_M02_20141115142313Z_20141115142727Z_N_O_20141115152139Z.fw.nc

[Ian Culverwell]

GRAS_xxx_1B_M02_20141115142313Z_20141115142727Z_N_O_20141115152139Z.bg.out

[Ian Culverwell]

GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.bg.nc

[Ian Culverwell]

GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.fw.nc

[Ian Culverwell]

GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.bg.out

[Ian Culverwell]

GRAS_xxx_1B_M02.sh

[Ian Culverwell]

The different warnings are fine: the undulations are, after all, missing from the multifile, and the code correctly raises some warnings in this case. But the fact that the undulations are changed in the FM output from the multifile input needs fixing.

[Ian Culverwell]

For some reason, obs1dbangle%undulation equals zero for the first profile in a multifile, but ropp_MDFV thereafter. Don't know why. The value is unspecified in ropp_fm/common/ropp_fm_types.f90, so it may be compiler-dependent. The safest solution would appear to be setting it equal to ropp_MDFV in ropp_fm/common/ropp_fm_roprof2obs.f90 if it is found to be out of range in that routine. This results in

ncks -H -Q -vundulation GRAS_xxx_1B_M02_20141031001713Z_20141031002055Z_N_O_20141031014235Z.fw_test.nc
undulation[0]=-9.9999e+07 
undulation[1]=-9.9999e+07

Doesn't break the single file output, so commit this change (also made to the 2d bending angle subroutine in ropp_fm_roprof2obs.f90) at r4773.

[Ian Culverwell]

Closing ticket.