Changes to the ncdf4 data reader

[Ian Culverwell]

Axel von Engeln (EUM) has provided the attached patch file which lists they apply to the ROPP7.0 version of ropp_io/ropp/ropp_io_read_ncdf_get.f90 to read the latest version of their ncdf4 files. Try to include these in ROPP8.0.

[Ian Culverwell]

3 test datasets now attached.

[Ian Culverwell]

In fact these are just the changes I made to ROPP7.1 at r3958.

Fixed 3 small errors (tidied up calculation of offset, allowance of time units like 'seconds since YYYY-MM-DD HH-MM-SS', and correct placement of bangle_qual assignment) at r4218.

Closing ticket.

[Ian Culverwell]

Some further corrections to the preferred EUM data format are included in the attached files.

[Ian Culverwell]

GRAS_1B_M02_20120909000057Z_20120909000118Z_N_T_20141111112929Z_G15_NN.nc

[Ian Culverwell]

GRAS_1B_M02_20120909000225Z_20120909000357Z_N_T_20141111112955Z_G17_NN.nc

[Ian Culverwell]

GRAS_1B_M02_20120909000330Z_20120909000549Z_N_T_20141111113043Z_G23_NN.nc

[Ian Culverwell]

GRAS_1B_M02_20120909000603Z_20120909000655Z_N_T_20141111113116Z_G30_NN.nc

[Ian Culverwell]

GRAS_1B_M02_20120909000626Z_20120909000855Z_N_T_20141111113137Z_G08_NN.nc

[Ian Culverwell]

GRAS_1B_M02_20120909000857Z_20120909001011Z_N_T_20141111113206Z_G16_NN.nc

[Ian Culverwell]

Email reply to Axel on the subject:

1) The change of the lev1b group name from "high_res" to "high_resolution" requires
a change to

 IF (TRIM(resolution) .eq. 'high_res') THEN

     data%thin_method = 'Unthinned data'

  ELSE 

     data%thin_method  = ' '

     CALL ncdf_getatt(TRIM(ddir)//'thinner_method', data%thin_method)

  ENDIF

Otherwise it looks for the non-existent attribute /data/level_1b/high_resolution
/thinner_method.  

This new method of getting the thinner method works fine when this problem has been
fixed.

This change also requires an update to eum2ropp.f90, because I currently default any 
resolution not equal to "high_res" to "thinned"; obviously this needs to be changed 
to "high_resolution".


2) The variable '/science_data/'quality_control/rs_external_navbits_applied' is now
called '/quality/rs_external_navbits_applied', so you need to change

        CALL ncdf_getvar(TRIM(sdir)//'quality_control/rs_external_navbits_applied', have_nb)

to

        CALL ncdf_getvar('/quality/rs_external_navbits_applied', have_nb)

This only applies if you have RS data, which you don't in your small standard
dataset GRAS_1B_M02_20120909000057Z_20120909000118Z_N_T_20141103103146Z_G15_NN_0001.nc.  But 
RS data are present in all the others. 


3) I tidied up the reading of the processing time a little bit more. 


I attach my revised codes, or you can just update your copy of ROPP80_prototype
(r4368). 
   
eum2ropp and eum2bufr work OK with and without -r "high_resolution" for all 6 
datasets which you put on dropbox for me.

(These tests were, in fact, made on an earlier set of test datasets than the ones uploaded today. But these later datasets required no further change to the data reader.)

ROPP changes to support the new datasets have been made at r4361, r4362, r4364 and r4368.

[Ian Culverwell]

All works OK, so closing ticket.