ROMS with river discharge

[luohaogit]

Hi All,

I have experienced a problem regarding the river implementation in ROMS that I cannot explain.

I have set up a run without-river and got some results. Then I turned on the flag "TS_PSOURCE" in cppdefs.h to apply river forcing as point sources. After generating the river forcing files, I gave all the river discharge rate zeros and keep the S/T constant values as initial condition state. Then I would like to run this "river input" ready model with "ZERO" river input discharge to see whether I would like to get the same answer as without-river one. Here are my results after 132000s run (my timestep is 100s).

Clearly, there are something (noise or so)? I have also tried to give a very small discharge rate, say 2m^3/s, the results are almost identical. But if I changed the Salinity with time in the river forcing file (with river_flag as 2) to simulate freshwater input, even still with ZERO discharge rate, I can still clearly see the big difference at all the river input locations. Not sure what is wrong. Can anyone explain to me what I should do to correct this?

Any suggestions are highly appreciated.

Sincerely,

Hao

[kate]

Where did you put your river sources? Are you sure they are along the walls at the ends of your fjords? If they are on walls, I don't see how they can do anything because the river tracers are multiplied by Huon or Hvom, which ought to be zero. On the other hand, if the sources are in the middle of the fjord, of course they will change the tracer distributions.

[luohaogit]

So, the river point source should be put along the wall at the end of fjords? I put them in the middle of the fjords,is that the reason?
Sorry i am kind of new for the river run. The rookie ☺

Is there a test case or something i can check and learn from it.
By searching all myroms.org, i cannot find details of instruction to put rivers in.

[kate]

River sources are meant to provide water coming into the domain from the land, so should be at the land-water boundary. We have a domain now in which the rivers are coming from a hydrology model of flow into the Gulf of Alaska all along the edge of the land, from rivers and from flow into fjords all along their length - like when we saw Yosemite in the rain and there were waterfalls *everywhere*.

[justina0622]

I suggest you use map_rivers.m to plot the rivers' locations. It will return you the starting and ending points, which is very useful to check whether you have set the rivers as you expected. I downloaded the code here: viewtopic.php?f=14&t=2200, where you can also find a useful discussion about the setting-up of rivers.
Cheers,
Qianqian

[luohaogit]

Thank you so much for providing me suggestions.

I have downloaded the map_river.m, and changed the river locations against the wall (u/v mask point). Here is the figure generated with map_river.m, I am not sure it is right?

For the river flow east: river 36
For the river flow north: river 40 and river 10
For the river flow south: river 22
For the river flow west: river 24

I am thinking something weird here, especially when river discharge rate is negative to show the direction. Which ones are right? (10, 40, 36) or (22 and 24)

Please let me know then I will make corrections for the river locations.

Thanks

Hao

[kate]

That matlab script didn't work on my rivers file for some reason, so I have to guess what it's doing. Are you getting no influence from 36 and 40, influence from 22 and 24? It looks like 36 and 40 are in the land mask, not on the land-sea boundary.

For flow into cell i,j, you need to set:

From west, cell i,j, dir=0
From south, cell i,j, dir=1
From east, cell i+1,j, dir=0
From north, cell i,j+1, dir=1

[justina0622]

Kate is right. It seems you have 36, 40, and also 10 in the land mask. In this code, river is pointing from circle to star, and for a right location, its circle should be on the land-sea boundary and star in the water point, as points 22 and 26 do.

[luohaogit]

Update on this:

Thank you again for the advise. It is weird for me that for the locations shown in the map, all 36, 40, 10 and 22 and 26 location rivers disappeared on my test case. The figure I posted is plotted with map_rivers.m, I have checked to move the river input points around, and I found out that:

If the river goes east like 36, I have to put river at point (408,292)
If the river goes north like 40, I have to put river at point (408, 288)
If the river goes west like 24, I have to put river at point (429,303) moving left one
If the river goes south like 22, I have to put river at point (421, 300)

I have followed the suggestion of kate, 36, 40 and 10 are working as shown above. But all negative rate (west, south) rivers all disappeared if I applied the location in the map I posted.

I am using the ROMS3.3 version, does ROMS version matter?

Thanks

Hao

[kate]

The rivers are defined on the u-points and v-points of the grid into a rho point at i,j. See the diagrams here.

Note that other models use a different numbering scheme where rho point i,j is surrounded by u at i-1,j and i,j. Viewing fields in ncview might make you think that's how ROMS numbers things, but you'd be wrong.

[nacholibre]

Hao,
Probably a bit too late to be of any help to you, but I have provided some explanation on ROMS and MATLAB indices here. Hope this helps as a future reference to anybody who has issues with the river / point source locations.
Zafer