Add (FE)DIACOV to DA Tools

[ewhelan]

Add DIACOV to AccordDaTools.

Source code in Harmonie:src/utilities/bcov_lam but we can create scripts to produce data and plot data.

Added here @magnuslindskog and so it won't be forgotten!

[ewhelan]

@uandrae : as discussed, let's try to use GitHub issues to discuss - you can also gently nudge Magnus in this direction!

[magnuslindskog]

that will not be an easy task for Ulf... have a nice weekend ..

[ewhelan]

that will not be an easy task for Ulf... have a nice weekend ..

Welcome @magnuslindskog! I am all set to nudge from Ireland too!

[ewhelan]

Hej @magnuslindskog ,

Testing the version in pre-CY46h1 (same as develop) with fort.4 containing (for a 100x100xL65 domain):

&NAMDIACOV
NFLEV=65,
NSMAX=49,
NDGL=100,
PPDELTAX=15000.,
CJBTYPE='STABAL96',
/

I get an ABORT as follows:

ewhelan@realin23:gcc-8.4.1:run_diacov> ./DIACOV 
 NFLEV           65
 NSMAX           49
 NDGL           100
 PPDELTAX    15000.000000000000     
 CJBTYPE STABAL96
 Subroutine DIAGCOV started, Jb conf=STABAL96
 Reading covariance files, set=STABAL96
    searching parameter            4  in GSA file stabal.cv                                                                       ...
    clid =  ALADIN98
    clcom =   Balanced statistcs for a LAM, after L. Berre 1998                    
     ID=ALADIN98
     comment= Balanced statistcs for a LAM, after L. Berre 1998                    
               85    20110801           0           5
     scanning set no.           1
               1         168           0           1           0
               1          13          50           0           0           0
     scanning set no.           2
             300          45           1           2           1
              65          65           4           4           1           1
    ...found in set no.           2 . Reading the data.
 ABOR1 CALLED
 ERROR - BAD MATRIX NUMBER
 ABORT!    1 ERROR - BAD MATRIX NUMBER

Not convinced of some of the checks in src/utilities/bcov_lam/others/rdgsacov.F90

Let me know how you are getting on.

[magnuslindskog]

I did this like 15 years ago when Gergely thought me .. I think on issue
here is that you need to have teh cvt file that we usually through away,
not the cv file. I have not tried yet.

[ewhelan]

Commenting out the "BAD MATRIX" check on L60 helps. I get a selection of *.y and *.xy output. Until next week ...

[magnuslindskog]

From my old script below.
it will generate a lot of. .x.xy. files

#!/bin/ksh

namelist variable CJBTYPE :

NONSEP93 for diagnosing stabal.cvt (covs of full (=total) variables)

STABAL96 for diagnosing stabal.cv (covs of control variables)

echo ".... catting the namelist ...."
cat > nam_stat << EOF
65
323
648
2500
NONSEP93
EOF

cat nam_stat

ls -lrt

./diacov.x

[magnuslindskog]

Hi,
I have renewed the Gergo's package for computation of B matrix (cy43t2_bf10). There is also new visualization which is mixture of perl, ksh and gnuplot.
I hope that it will help you with computation of B matrix on cycle 43.
Cheers, Tonda.
i did not manage to upload the tar file to this cjhat but i do have it, magnus

[ewhelan]

Hej @magnuslindskog ,

Glad you had some success (your results attached). I did too - with different Jb files. I think my problem may have been an issue with files produced by jbconv. With "uncooked" Jb files it works fine for me too. I will work on a script to call DIACOV.

Eoin

[magnuslindskog]

/home/sm_mlind/hm_home/cy43mc_modarl/src/utilities/bcov_lam/programs
I see
-rw-r--r-- 1 sm_mlind sm_foup 856 Sep 28 12:51 diacov.F90
-rw-r--r-- 1 sm_mlind sm_foup 1066 Sep 28 12:51 festat.F90
But not yet nmcstat.F90, Is it in develop so that I can fetch?
Thanks,
Magnus

[magnuslindskog]

https://doi.org/10.1175/1520-0493(2000)128<0644:EOSAMF>2.0.CO;2
plot like in figs 4,5, and 9 are very informative

[ewhelan]

/home/sm_mlind/hm_home/cy43mc_modarl/src/utilities/bcov_lam/programs I see -rw-r--r-- 1 sm_mlind sm_foup 856 Sep 28 12:51 diacov.F90 -rw-r--r-- 1 sm_mlind sm_foup 1066 Sep 28 12:51 festat.F90 But not yet nmcstat.F90, Is it in develop so that I can fetch? Thanks, Magnus

NMCSTAT is called from festat - I should have said festat can produce the "cov" files. Add FESTAT not NMCSTAT to your defaults.mk

[magnuslindskog]

check if i did a pull correctly

[ewhelan]

check if i did a pull correctly

Hej @magnuslindskog, I can see you have created a fork.
The next steps are to

1. Create a feature branch
2. Commit you script to this branch
3. Create a "pull request"

Or if you think you have done this in a clone on your PC then you need to push these changes with git push origin feature/branch_name

Happy to give a quick tutorial if you have time today.

[swapanmallicksmhi]

Hej @ewhelan,

Please suggest the best way where I can find the program 'diacov.F90'.
Actually, I need to plot all the statistics from only three available files stat.bal , stat.cv and stat.cvt file.
And for this I need to run 'rundiacov' , but no exe file is available.

[ewhelan]

Hi @swapanmallicksmhi You need to use a DIADOV exe compiled by Harmonie or similar.