yanglong09

interactive channel/overland flow: how is top of bank determined

12 posts in this topic

I was trying to simulate interactive flow between the channel and the corresponding grid cell. I turned on the OVERBANK_FLOW on so that flow from the channel will spill onto the grid cell when surface elevation exceed the top of bank for the channel. My question is: how top of bank (top) is determined in the model. I'm using channel cross sections extracted from DEM. Does the model use the highest point in the elevation, or there are more sophisticated ways of doing that? Thanks!

Share this post


Link to post
Share on other sites

Yanglong,

I think the top of the bank is determined from the elevation of the cell through which the stream travels, but I'm not completely sure about this.  There is a channel Depth parameter that's defined in GSSHA, but I don't know if this is used to determine the top of the bank.  I'll need to check into this more and get back with you.  If the OVERBANK_FLOW option is on, flow from the channel to the overland flow cells occurs if the water surface elevation in the channel is higher than the water surface elevation on the overland flow cells.  I'll see what I can figure out and get back with you about this.

Chris

Share this post


Link to post
Share on other sites

Yanglong,

After a little bit of looking around, I found the following description on this page:

"The top of bank (TOB) is defined as the thalweg elevation plus the stream depth. For trapezoidal channels the user specifies the stream depth and the thalweg elevations [this is the "depth" parameter in WMS]. For natural cross sections TOB is taken from the XY point series inputs."

I hope this helps,

Chris

Share this post


Link to post
Share on other sites

Hi Chris,

Thanks for your swift response. I've checked the page before. According to your understanding, the model just picks up the largest Y values in the XY point series as the TOB (I'm using natural cross sections)? Is that what it is? Another follow-up question, if we assume no interactive flows from channel to grid cell, what would happen in the model when the water depth in the channel exceeds TOB? 

Thanks,

Long Yang

 

Share this post


Link to post
Share on other sites

Long Yang,

Yes, the model picks the largest Y values in the XY series as the TOB for natural channels.

I'm not sure what you're asking in the second part of your question, but unless you have the OVERBANK_FLOW option defined, any water in your channel will stay in your channel when the depth exceeds the TOB.  In this case, the model uses linear extrapolation to determine the channel elevations beyond the given channel geometry data.  You will get a warning in the GSSHA summary file if this happens.

Hope this helps,

Chris

Share this post


Link to post
Share on other sites

Hi Chris,

This is exactly what I'm asking. Thanks!

I'm currently trying to simulate extreme flood events using GSSHA. I tried not to turn on OVERBANK_FLOW, cause the simulation just cost too much time to finish. However, if I did not turn on OVERBANK_FLOW option, the model crashed at certain nodes, and I got the error message as "The channel routing time step is getting really small less than 1/1000 of a second. Reduce the model time step by a factor of 2 and try again. Program stopped." I know there might be some abrupt transitions of flow volume around the nodes, due to which the model stability could not be satisfied by reducing time steps. The time step is 1s for my case. I then tried to reduce the time step to 0.5s. The new simulation passed the crashing nodes of the 1-second simulation, but crashed over other nodes anyway.

I'm using the natural channels (cross sections extracted from DEM). I know there might be sharp longitudinal transitions of cross sections over the channels, but not quite sure if this would crash the model. Any suggestions?

Thanks,

Long

Share this post


Link to post
Share on other sites

Long,

Yes, I'd say if there are sharp transitions between adjacent cross sections of your channel you could get some model stability problems.  I don't know if it is feasible, but could you try changing all your channels to just use trapezoidal cross sections?  Then you could work on adding the natural cross section information once you get the trapezoidal cross sections running with the OVERBANK_FLOW option.  Maybe you could save your modified model to a different folder.

Another thing you should check is your stream thalweg elevations in relation to your grid cell elevations that intersect your stream.  The grid cell elevations should be approximately the same as your TOB elevations.  You can check and edit these elevations in the WMS Map Module under the GSSHA | Smooth Stream/Pipe Arcs menu item.

Check your GSSHA summary file (.sum) to see if there are any warnings you're overlooking when you run your GSSHA project.

If you check these things and make sure your cross sections have fairly smooth transitions and you're still having trouble getting a stable GSSHA model, you could contact Chuck Downer.  He is the main GSSHA programmer and he should be able to help you if you send him your model and the specific issue you're having.  His email address is on the main gsshawiki.com page.

I hope this helps,

Chris

Share this post


Link to post
Share on other sites

Thanks, Chris.

I tried to use the same cross section for all the channels (ensuring no transition), and the problem was perfectly solved. The model ran with no warnings or any error messages.

I was just wondering if there are alternative ways to solve the 'abrupt transition' issue, cause sometimes transitions in the channels are physically present. The way I can think of would be to add a structure link in between the channels with sharp transitions. However, I'm not sure if it is entirely feasible, cause I do not really know what kind of transition does the model regard as too 'sharp' to solve. It seems that I could not locate the sharp transitions by simply looking up warning messages in the summary file, cause the nodes that broke down are not always the ones with sharp transitions. Any suggestions?

Thanks,

Long

 

Share this post


Link to post
Share on other sites

Long,

I don't have any suggestions other than trying to fix the sharp transitions in channel geometry by making sure your channel cross sections represent the actual channel bathymetry.  What is the resolution of the DEM you're using to extract the cross sections?  Are the channel bottom elevations represented in the DEM?  If not, maybe you can modify the cross sections extracted from the DEM to include an approximation of the actual channel cross section as well as the channel thalweg.  If your model runs with the same cross sections assigned to the nodes of your links, this tells me that the instability is introduced from the different cross sections at each node of your stream network.  You could try decreasing your number of nodes/cross sections in your model by merging the arcs in your GSSHA coverage.  Alternatively, you could try making some of the cross sections closer together where there are sharp transitions, changes in slope, or abrupt changes in Manning's roughness.  I think it's just a matter of trial and error from here.  You could also try sending your model to Chuck Downer and see if he replies and has any input.  He would be interested in getting the model working better for your case.

Hopefully this helps,

Chris

Share this post


Link to post
Share on other sites

Thanks, Chris. The resolution of DEM is 30m, I guess the channel bottoms are not represented in the DEM. I'll do some trial and errors. 

Long

Share this post


Link to post
Share on other sites

Long,

Let me know how things go.  Thanks,

Chris

Share this post


Link to post
Share on other sites

Hi Chris,

Thanks for your follow up. I have frozen this work for a while, and pick it up again this week. As I mentioned earlier, I'm using GSSHA for the simulation of an extreme flood event, but somehow the model breaks down at certain point. It might be due to the abrupt transitions across the cross sections of channels within the model, which is at least what we concluded weeks ago. After many trials and errors, I'm now 'unifying' the cross sections, meaning that all the channels are sharing the same cross-section profile (also same Manning). But, the model still breaks down. I guess there might be some other tricks that we did not pay attention.

Cheers,

Long

 

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!


Register a new account

Sign in

Already have an account? Sign in here.


Sign In Now