Storms and Boundaries

[Kyle Mandli]

Hi everyone,

I wanted to let users know who may be interested that I have added an example to the surge-examples that explores boundary conditions for storm surge. The example looks at what happens when a storm completely crosses the boundary of the domain and proposes some more complex boundary conditions. Unfortunately, it does not entirely work, so I would strongly encourage users of GeoClaw to continue to keep storms away from the boundary. The example also demonstrates how to construct a storm object directly.  If you have questions or would like to discuss the boundary conditions please reply to this thread.

Additionally, new capabilities were added to the Storm object in GeoClaw to help to plot the track, intensity, and area around a storm. This can be helpful when looking at whether a storm is getting too close to an area or boundary. A Jupyter notebook was added to show a couple of examples on how to do this along with ways to construct a storm object directly rather than from a file.

Kyle

[Kyle Mandli]

I forgot to mention that the example is at 

https://github.com/mandli/surge-examples/tree/main/synthetic/boundary

and depends on the PR

https://github.com/clawpack/geoclaw/pull/622

[Robina]

Hello Kyle,

Is it possible to keep storms away from the boundary while applying the ARM with more levels of refinement only to a specific region within the whole area? Would this approach ensure the simulation works well for that specific region?

Thank you.

[Kyle Mandli]

That is exactly what we suggest.  Generally expanding the domain in space so that the eye of the storm is a storm width away from the boundary works.  Also, if you do not need the entire time frame that you have data for, you can constrain the times so that the storm does not approach a boundary.  This is often relevant for storms in the Atlantic that curve N towards Europe for instance.

If you do expand the domain, constraining the AMR regions is the way to go.  Often times I will place a region across the entire domain that restricts the refinement to to the first few levels and then place a region around the area that I am interested, allowing full refinement in these regions.  The following code does this.

regions.append([1, 3, clawdata.t0, clawdata.tfinal, clawdata.lower[0], clawdata.upper[0], clawdata.lower[1], clawdata.upper[1]])

regions.append([3, 4, clawdata.t0, clawdata.tfinal, -98, -96, 25.5, 28.0])

Here the first region lets refinement occur at levels 1-3 throughout the entire domain and the second allows up to 4 and forces refinement to 3 in the region specified.  You could also just change 3 -> 1 and allow the code to decide how to refine in that region.

Kyle

--
You received this message because you are subscribed to the Google Groups "claw-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to claw-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/claw-users/ebf36d1b-4b49-438d-b514-64db86152009n%40googlegroups.com.

[Robina]

Thank you so much, Kyle.

Best,