late blow up with WET_DRY

Report or discuss software problems and other woes

Moderators: arango, robertson

Post Reply
Message
Author
nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

late blow up with WET_DRY

#1 Unread post by nacholibre »

I am using ROMS 3.2 svn 351M. I am modeling a real case and I only have the tidal forcing on. My model blows up after 12 days after 20 tidal cycles. I have checked the depth averaged (ubar, vbar) and depth varying (u, v) values, and also the water surface elevation (zeta), but there are no extreme values that indicates a blow up related to those. It is not like the early blow up issues that I had in the past, which were easier to debug.
I wonder if anybody else who had a similar problem and can tell me what else should I check and help me diagnose the problem.
Thank you
Zafer

jcwarner
Posts: 1172
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

Re: late blow up with WET_DRY

#2 Unread post by jcwarner »

did u look in the restart file? The last time step of that file contains the final fields.

nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

Re: late blow up with WET_DRY

#3 Unread post by nacholibre »

jcwarner wrote:did u look in the restart file? The last time step of that file contains the final fields.
Yes sure. Nothing unusual: the velocity and the water surface elevations are physically acceptable. No large values.

jcwarner
Posts: 1172
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

Re: late blow up with WET_DRY

#4 Unread post by jcwarner »

not sure. look at all the fields. look at all the forcing data. Can u restart it?

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: late blow up with WET_DRY

#5 Unread post by kate »

Did you check the density field? ROMS looks for outrageous values of both velocity and density.

Can you run with a shorter timestep? Sometimes ROMS just needs a shorter timestep to run stably, being happy for months before blowing up, but needing more steps to get through the occasional rough spot.

nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

Re: late blow up with WET_DRY

#6 Unread post by nacholibre »

jcwarner wrote:not sure. look at all the fields. look at all the forcing data. Can u restart it?
Unfortunately I am having a problem restarting my runs. I have been trying to restart from the netCDF restart file, but no luck so far. I have created a post about this very recently...
viewtopic.php?f=14&t=1403
Thank you
Zafer

nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

Re: late blow up with WET_DRY

#7 Unread post by nacholibre »

kate wrote:Did you check the density field? ROMS looks for outrageous values of both velocity and density.

Can you run with a shorter timestep? Sometimes ROMS just needs a shorter timestep to run stably, being happy for months before blowing up, but needing more steps to get through the occasional rough spot.
No outrageous values in velocity and density. All Courant numbers are small enough. It still blows up even if I reduce the time step.

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

Re: late blow up with WET_DRY

#8 Unread post by kate »

You need to figure out the restart problem then figure out the ROMS problem by restarting and seeing what is what. The restart is failing with that posixio issue - which field from which file is it failing to read?

nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

Re: late blow up with WET_DRY - possible bug with masking?

#9 Unread post by nacholibre »

kate wrote:You need to figure out the restart problem then figure out the ROMS problem by restarting and seeing what is what. The restart is failing with that posixio issue - which field from which file is it failing to read?
The model runs fine when I turn the wet_dry off and set the minimum depth larger than the tidal range. That makes me think that it is probably the wet_dry that is causing the blow-up. I have been reading the bulletin board for some time and I have seen some posts that might be related such as
viewtopic.php?f=19&t=1087&p=3847&hilit=dcrit#p3847
I wonder what the final situation on this is.
Also after a successful restart I was able to identify the troubled spot this time. I have a grid cell where v is blowing up, v(80,618). The rho and v masks for that grid cell at the step of blow up are as follows. Considering the Arakawa C grid at this step wetdry_mask_v(80,617) should be masked, but it is not. This seems to me that there might be a glitch in this part of the routine. I am willing to look into this in more detail. Any comment from Dr.Warner would be very helpful as well.
Attachments
wdm_rho.png
wdm_v.png

jcwarner
Posts: 1172
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

Re: late blow up with WET_DRY

#10 Unread post by jcwarner »

it is difficult for me to really see what is going on here. Can i get the Projects folder for this app and try to run it here, and step thru what is happening?

nacholibre
Posts: 81
Joined: Thu Dec 07, 2006 3:14 pm
Location: USGS
Contact:

Re: late blow up with WET_DRY

#11 Unread post by nacholibre »

jcwarner wrote:it is difficult for me to really see what is going on here. Can i get the Projects folder for this app and try to run it here, and step thru what is happening?
Sure, I will share it with you. I was looking at the results today and it is interesting to note that there are very high elevations above the MWL that are not masked at all (For example around (625,80). Water level never rises that high at any time in my model). Below are h and h with wetdry rho masking.
Attachments
h with mask.png
h.png

Post Reply