Possible bugs in tl_uv3dmix2_geo.h

Bug reports, work arounds and fixes

Moderators: arango, robertson

Post Reply
Message
Author
xinyou_lin
Posts: 2
Joined: Tue Sep 22, 2009 3:54 pm
Location: Xiamen University

Possible bugs in tl_uv3dmix2_geo.h

#1 Unread post by xinyou_lin »

I have been working on a strong constraint data assimilation test with version 424. If # define MIX_GEO_UV and # define UV_VIS2
are selected, the model may blow up suddenly with NaN error. The first NaN error came from the variable dVdz in tl_uv3dmix2_geo.h.
Furtherly, I found that the Loop index to get dVdz is inconsistent with that to use it. Lines 442~443 and lines 541~542 are shown as follows:

442 : DO j=-1+JV_RANGE+1
443 : DO i=IV_RANGE

541 : DO j=JU_RANGE+1
542 : DO i=IV_RANGE+1

We can see clearly that 542 range is out of 443.
The similar mistake can be found between lines 426~427 and lines 541~542

426: DO j=JU_RANGE
427: DO i=-1+IU_RANGE+1

In order to solve the bug, lines 541~542 were changed as follows:
541 : DO j=1+JU_RANGE
542 : DO i=1+IV_RANGE

Now the model can run right. I am not sure that my correction is consistent with
the original design. I need your help. Thank you!

User avatar
arango
Site Admin
Posts: 1367
Joined: Wed Feb 26, 2003 4:41 pm
Location: DMCS, Rutgers University
Contact:

Re: Possible bugs in tl_uv3dmix2_geo.h

#2 Unread post by arango »

I finally have time to look at this problem. The issues here are very tricky and I decided to rework all the logic. It took me several days to re-work all the routines associated with MIX_GEO_UV. See repository :arrow: ticket for more details. Please update.

Thank you for reporting this problem :oops:

Post Reply