Bug in wet_dry- the lake_jersey case has the same issue

[simion1232006]

Dear ROMS users,

I had a problem with the information exchange between donor and receiver grid months ago, while after the update at the end of 2014, the problem disappear. But now it comes back. I hope someone can give me a hint.

Simulation with one_way nesting.

The donor grid runs fine ,since there is no feedback from receiver grid. while the receiver grid went crazy, and the boundary values pass down from donor grid were definitely interpreted in a wrong way. I did not plot the case of two_way nesting - the donor grid was ruined by the incorrect feedback and blow up soon.

I did run the example of lake-jersey and it works all right, even I regenerated the contact file by the contact code.

So, I rechecked the differences between my case and lake_jersey case. And the only difference I can find is I used the spherical coordinate. I assume this part of code in contact.m might not be fully tested. I hope the ROMS group or other experienced user can have an look at these codes in contact.m.
Any comments or suggestions are appreciated here. Many thanks.

[rock]

Below are what I found recently,

Line 609 'R.x_rho' should be 'R.x_psi';
line 610 'G(dg).y_rho(:)' should be 'G(dg).y_psi(:)', and 'R.y_rho' should be 'R.y_psi';
line 753~755 'FCr' should be 'FSr'.

Yan

[simion1232006]

Thank you. That is helpful. I did a couple of tests, and the run without wet_dry looks fine, the wired results only appear with wet_dry on. I will keep updating my findings.

Below is another nesting domain - twoway nesting, wet_dry. Notably there is no dry land in the domain.

I assume the wet_dry option changed the interpolation algorithm in contact region.

[simion1232006]

I retested the lake_jersey case. With wet_dry on, the test case have the similar problem. I guess now I can have a rest (too tired). It confirms that the wet_dry option is still buggy in the newest ROMS. See below:

Anyone can try this out by switch on the wet_dry in lake_jersey case. You should be able to notice the same problems.

[jcwarner]

i am running an application right now for Hurricane Sandy using 3 roms grids and 3 swan grids, coupled, with grid refinement. I have wet_dry on. I am not seeing the issues that you have. We did just recently issue an update in Rutgers ROMS/Utility/set_contact.m . This was for svn #659. Do you have that most recent update?

-john

[simion1232006]

Hi Warner,

Thanks for your reply. Mine is svn #756, and I could not get the #659 you mentioned.I am quite curious of the ROMS version you use. I know you have developed your own branch of ROMS (forgot the name, sorry). Is that all right if I can use svn to download your version?
I just looked into the nest.f90 code, and I will update my findings. Oh, by the way , if I switch on the QUADRATIC_WEIGHTS, the model could not be compiled.

FOR bug, I mean it could be just a typos. As I mentioned, the previous version （as far as I can remember, it is #737 ） actually works with wet_dry (not perfect), at least showing no drama on the donor receiver interface.

[jcwarner]

sorry, i did not mean roms svn version number, I was referring to the ticket number. But I am using the latest version of roms, svn #756, although I do not use rutgers directly, we have it implemented as part of COAWST. But it should be the same as what you have. Did you also update and rerun your Tools to make the contact grid file?

[arango]

You need to start paying more attention to the updates to the code, specially if you are using nesting. You need to read carefully the following post. It contains information about the quadratic interpolation. It is not ready yet Also, the information about the wet/dry update was provided in the following trac ticket.

Again, you are jumping into conclusions about contact.m...

[simion1232006]

Thanks a lot for your reply,Arango. I apologize for directly jumping into contact.m(Although there are couple of none essential typos mentioned by rock, the code works fine). The spherical grid also works great with the contact.m. I did not know the 659 ticket, but I do constantly update my ROMS at monthly base. Anyway, I will check the 659 ticket.

Very good news: the model runs like fly...Thanks everyone. I updated my code 15/1/2015, while the recent update was around 3 weeks ago. SO......
For anyone who has interests in nesting, it is better to update to the newest version always.. For me it is a good opportunity to learn ROMS codes.