River locations when there is wetting/drying

General scientific issues regarding ROMS

Moderators: arango, robertson

Post Reply
Posts: 157
Joined: Mon Apr 28, 2003 5:12 pm
Location: NOAA

River locations when there is wetting/drying

#1 Post by lanerolle » Fri Dec 07, 2012 4:52 pm

In ROMS, in general, we place rivers/point sources adjacent to land mask cells - as otherwise (e.g. placing them in the middle of the ocean) a dipole-type behavior results in T, S and possibly also blow-ups.

When we have wetting/drying, this is not easy to do because (i) the permanent dry/land mask could be very far from sufficiently deep bathymetry (e.g. 5-10m) and (ii) the "wet" cells immediately adjacent to the dry/land mask are usually shoals which wet and dry (due to tidal action) and they may indeed have negative h values (as in the grid NetCDF file) implying that they are not bathymetry but topography. I am experiencing these for an application I am doing in Alaska where the tidal range is ~10m and the wetting/drying shoals are extensive.

In applications such as these, what is the best way to place rivers/point sources? Can we place a river/point source next to the permanent land mask even if h < 0 implying that these cells are topography?

Posts: 855
Joined: Wed Dec 31, 2003 6:16 pm
Location: USGS, USA

Re: River locations when there is wetting/drying

#2 Post by jcwarner » Fri Dec 07, 2012 6:31 pm

we have had some success in placing rivers in tidal wet/dry locations. if there is a constant flow out of the river, then a series of cells should stay wet and allow the river to maintain a constant conection to the moving water line. It is fine to have the river come out of a location that has a negative depth (ie a location of topography). in the model, when wet/dry is activated, the free surface is draped over all the topography. So in the model, there is really no "dry", it is just really shallow (ie = to Dcrit).
- You have to put the point source at the edge of a land masked area. You can not put a point source in a location where the land sea mask is water. (this is different than wet/dry mask).
- the wet/dry mask is computed internally, and should take care of itself. but a location that has a land mask of 0 will always be dry.
I suggest you locate a cell that has a land mask of 0, have the river come out of that cell edge. You may need to channelize a direction for the flow, which is mimicing real life, so that at low tide the flow has a sense of direction to know where to go.

Posts: 87
Joined: Wed Mar 31, 2010 3:29 pm
Location: SKLEC,ECNU,Shanghai,China

Re: River locations when there is wetting/drying

#3 Post by tony1230 » Sat Dec 08, 2012 2:49 am

I'v had the same problem long time ago. I have a extensive shoal as lanerolle said that can be wetting/drying along with tidal flood/ebb. As commonly think the point source/river would be connected to the sea with a permanent water line/flow even when reaching a low tide. But in roms we alway get the point source isolated(some cells would be wetted adjacent) during ebb for the point sources were usually located far from the water-land interface(swing with tidal behavior). Sometimes, i may want to excavate a pathway for the river to connect with the open water, by changing the depth values(along the flow line) from ~ 0.5 m to a bit more large ones, but never to be done.

In my application , i have ~ 10 point sources flow into a bay. I would like to show you a figure that can explicitly point this problem out. In the figure the river flow is connceted with open water on and off, but the small drainages would be totally isolated when tides run offshore and reconnected when towards land. If this the case, the points flow into the bay at just some timesteps. Is that can make sense? I dont think so and dont find a better way to treat it.

- Shou
the dashed rectangle is a river and small ellipse are outflows
the dashed rectangle is a river and small ellipse are outflows
River flow.png (11.14 KiB) Viewed 2104 times

Posts: 157
Joined: Mon Apr 28, 2003 5:12 pm
Location: NOAA

Re: River locations when there is wetting/drying

#4 Post by lanerolle » Thu Dec 13, 2012 3:44 pm

Thanks very much John - I followed your advice and generated a layer of wet cells across my mudflats which never dry and it all works well.

In reality, if you do go to AK and look at the mudflats (e.g. in Turnagain Arm on the way from Anchorage to Seward) you will see that even in low tide, rivers continue to flow into the bay/estuary/ocean. The flow occurs via crevices carved into the mud in the mudflats. Now ideally, we can emulate this with our models if and only if we are able to resolve these crevices on our model grid and this alone is tough enough. To further complicate matters, the regions in AK with mudflats also have heavy sediment movement and these crevices can change their locations, depth and geometry from day to day!!! I wonder whether scenarios such as this can ever be modeled via ocean models and what we can do for the moment I guess is to have rivers whose geometry is stationary in time.

User avatar
Posts: 35
Joined: Tue Jun 19, 2007 3:32 pm
Location: National Center for Atmospheric Research

Re: River locations when there is wetting/drying

#5 Post by drews » Fri Dec 14, 2012 4:10 pm

I'm working in the Nile delta, which is really flat. I have strong winds but no tide, and consequently there is a lot of wetting and drying going on. I have set my permanent land-sea mask at a height of 30 meters above sea level (absurdly high). I have a river point source at 5 meters above sea level, and this source is 70 kilometers from the Mediterranean. This point source is nowhere near permanent land or permanent sea.

To handle the "edge" requirement for a point source, I set the land-sea mask for one single cell to 0. That single cell is adjacent to the point source and directly upstream of it. For what it's worth, here is the Python code and my comments from 2 years ago:

# Mask out the grid cell upstream of the point source.
# If you don't do this, you'll get huge currents there
# if the grid cells near the point source ever dry out!
mask[92, 3] = 0

The river source itself is at [92, 4], but keep in mind that these are staggered grids. Now I'm not sure what I meant by the huge currents. Amazingly, the downhill flow in the river is fairly realistic. I have excavated a river channel in the topography like you said (actually made sure the existing one connects from cell to cell). I think you could simply pour water on the topography and it would run downhill like it should.

The upshot is that I don't think you have to carefully place your river point sources on the very edge of your calculated land-sea mask. It may be easier to place them anywhere you want in the wetting and drying zone, above sea level, and mask out the single cell directly upstream. That's what worked for me.

Here is the paper for more info: http://www.plosone.org/article/info%3Ad ... ne.0012481

Post Reply