Problem using FM output as input

[Ian Culverwell]

ROPP user Axel von Engeln (EUMETSAT) writes:

Hi Ian,

I was testing our internal ROPP 9.1 (with netCDF-4 string handling) 
for our usual monitoring task (ropp_fm_bg2ro_1d). I found no 
differences to ROPP 9.0 when starting from a dedicated background 
file, down to the last digit, except for the start_time entry that 
has the leap seconds now included. 

But I then tried to use the output file of the ropp_fm_bg2ro_1d 
(where all background info is also included, thus it can be used as 
input to ropp_fm_bg2ro_1d), and there I find tiny differences in 
bangle, shum, and a lot of differences in the missing fields (fields 
like refrac_sigma, etc. Some are set to -9999.9, some are set to 100 
or other values). Nothing major, I was just wondering. I guess that 
when running the actual bg2ro tool, an array is initialized for all 
those missing output data. But when running from a forward file, 
these arrays are not re-initialized, nor are they just transferred 
from the input to the new output file, so they end up with some odd 
values. I am actually doing this fw2ro since we started to delete 
all intermediate background files and only keep the output forward 
files. And I don’t see why this should be related to any updates we 
did with the strings for ROPP 9.1.

I guess you can test that with files you have, but here are also 
attached some examples where I ran:

# usual bg2ro
ropp_fm_bg2ro_1d -comp -new_op --no-ranchk --no-zapem -f GRAS_DAY_1B_M02_20171201000158Z_20171202000110Z_R_O_20181113203954Z.bg.nc -o GRAS_DAY_1B_M02_20171201000158Z_20171202000110Z_R_O_20181113203954Z_ROPP9.1_bg2fw.fw.nc

# using output of command above and do a fw2ro 
ropp_fm_bg2ro_1d -comp -new_op --no-ranchk --no-zapem -f GRAS_DAY_1B_M02_20171201000158Z_20171202000110Z_R_O_20181113203954Z_ROPP9.1_bg2fw.fw.nc -o GRAS_DAY_1B_M02_20171201000158Z_20171202000110Z_R_O_20181113203954Z_ROPP9.1_fw2fw.fw.nc

Available here: ftp://eumftp:eum4ftp@ftp.eumetsat.int/out/RSP/engeln/  

Cheers and sorry for this minor stuff, put it on priority -9999 in 
your list,

                            Axel

Assign ticket, probably temporarily, to ROPP10.0.

[Ian Culverwell]

ROPP_9.1_Testing.tgz

[Ian Culverwell]

Should probably look at this for ROPP-11.0.

[Ian Culverwell]

This might be connected with a problem Chris M saw with ROPP-11.0. Don't defer yet.

[Ian Culverwell]

Actually move it to ROPP-11.1 at a high priority, so we can close all the ROPP-11 tickets.