Problem and question about SWAN in COAWST

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Message
Author
jordiiglesias
Posts: 2
Joined: Sat Nov 09, 2019 2:48 am
Location: Univeristy Rovira i Virgili

Problem and question about SWAN in COAWST

#1 Unread post by jordiiglesias »

Dear All,

I'm runing ROMS+SWAN+WRF model and I have problems with SWAN

I started the simulation with initial default, after 3h (5 seconds timestep 3 models) the waves still have 1mm Significant height, and a terrible resolution, it's a 600x600 grid and looks like 10x10.
Then I switched to only SWAN and tried to run in stationary and when got 75 % convergence i let it run for 1h in non stationary, the result is attached here, doesn't have sense, looks like numerical unestability and waves have 45m height in Mediterranean Sea. Do you know what could happen? wrong initial convergence (75%) ?

I also have 2 questions about SWAN,
1- Is it possible to create stationary hotfile with more than 1 node ( In coawst manual says to use 1), I guess is because doesn't merge all hotfiles at the end, but it takes days to converge with a single node.
2-What is the maximum number of nodes I can use? because if I run with 12 is ok, but at 20 or 32 the simulation crashes before start, I guess domain is wrongly splitted, but I don't know how to specify domain partition.

I attach 3 images, one is a subplot of 6 elements of hs, correspond to the timestep when i run coawst (at 3h of simulation ) another is skin temeprature in WRF, where you can see the resolution over sea is not good, and 3rd image is Hs when I run swan only

If someone could help me I would be really grateful
Thanks in advance
Jordi
Attachments
skintemp.jpg
hwave_3_models.jpg
hs_only_swan.jpg

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

Re: Problem and question about SWAN in COAWST

#2 Unread post by jcwarner »

sorry for the late reply on this.
Looks like the wind data to drive swan is not constructed correctly, or the swan grid file was not constructed correctly.
That is why the lines look like that.
Can you try to run swan by itself with just one grid? You could even do
WIND 20. 45.
(wind speed dir) in the INPUT file instead of reading any wind files. no boundary conditions. that would be a simple way to start.

You can run swan with multiple processors to get init files. suggest you do this with the same number of cores taht you will use later on.

-j

Post Reply