4DVAR observation impact driver problem WC13

[lmp4]

Hi all,

I am following the ROMS 4DVAR tutorials with the WC13 model and have come across an issue with the I4DVAR observation impact driver (Exercise 6).

Firstly the model compiled fine with the obs impact driver enabled but would not proceed to run successfully.

The error message was full of unknowns ;
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
oceanM 0000000000D24614 Unknown Unknown Unknown

so I then proceeded compiling with USE_DEBUG=on & export MY_CPP_FLAGS="${MY_CPP_FLAGS} -DDEBUGGING".

The error message then read as follows;

forrtl: severe (408): fort: (2): Subscript #1 of the array CLIMA has value 1 which is greater than the upper bound of -1

Image PC Routine Line Source
oceanG 00000000040B4A1D Unknown Unknown Unknown
oceanG 00000000040B3525 Unknown Unknown Unknown
oceanG 000000000405A160 Unknown Unknown Unknown
oceanG 000000000401529A Unknown Unknown Unknown
oceanG 0000000004015692 Unknown Unknown Unknown
oceanG 0000000001B191B2 ad_get_data_ 521 ad_get_data.f90
oceanG 0000000000D815BD ad_initial_ 234 ad_initial.f90
oceanG 00000000004096F7 ocean_control_mod 422 ocean_control.f90
oceanG 0000000000404E8D MAIN__ 108 master.f90

Looking at line 521 in ad_get_data.f90;

IF (SCALARS(ng)%Lstate(isFsur)) THEN
CALL get_2dfldr (ng, iADM, idZads, ADS(ng)%ncid, &
& 1, ADS(ng), update(1), &
& LBi, UBi, LBj, UBj, 2, 1, &
& GRID(ng) % rmask, &
& CLIMA(ng) % zeta_adsG)

Then at get_2dfldr and finally into mod_clima.f90;

! Adjoint-based algorithms arrays.
!
allocate ( CLIMA(ng) % zeta_ads(LBi:UBi,LBj:UBj) )
allocate ( CLIMA(ng) % zeta_adsG(LBi:UBi,LBj:UBj,2) )

There must be something going wrong with this CLIMA array although I can't figure out where why CLIMA would be causing the issue.

ng is simply the nested grid number?

Here almost everything bar the the distributed-memory partition (2,2) is set up as recommended.

My distributed-memory partition set up is as follows;

NtileI == 3 ! I-direction partition
NtileJ == 4 ! J-direction partition

These setting have been run with all the other 4DVAR drivers in the tutorials i.e. weak, strong, PSAS, Errors, EOFs etc, so why it would stop now running for the obs impact is a mystery to me.

Could it be a compiling issue with ifort? i.e. need to update intel-suite compiler? (Currently using 11.1 & /2011)
Or the distributed-memory partition? change to 2,2

Any suggestions would be greatly appreciated!

[arango]

I don't know what it is the problem here. The CLIMA(:) structure has a -1 index. That's is kind of weird. I will need to reproduce the problem. I added it to my to-do list.

[lmp4]

The latest bug fix Ticket #658 https://www.myroms.org/projects/src/ticket/658 appears to target this issue with the CLIMA(ng) structure. I haven't tried to run this again but I shall assume this is sorted now

[arango]

Yes, that fixed the problem.