Create tsunami module#324
Conversation
stephankramer
left a comment
stephankramer
left a comment
There was a problem hiding this comment.
@jwallwork23 : still planning to add more tsunami cases? In any case this seems to move a module from an example to thetis/ - which looks fine to me.
Thanks. No plans to do any tsunami modelling in the near future myself. |
|
[Merged |
|
@joewallwork : same here. If you still want to proceed could you resolve conflicts, or otherwise close it? On the surface looks to be a useful change, but understand if you don't have the time for it now. |
Seemed like a straightforward merge - let's see what the CI says. |
| R = self.rotation_matrix() | ||
| x, y = ufl.dot(R, X) | ||
| L, W = self.fault_length, self.fault_width | ||
| cond = ufl.And(ufl.And(x > -L / 2, x < L / 2), ufl.And(y > -W / 2, y > W / 2)) |
There was a problem hiding this comment.
| cond = ufl.And(ufl.And(x > -L / 2, x < L / 2), ufl.And(y > -W / 2, y < W / 2)) |
Just playing around with AI, and it points me at what looks like a serious bug that was already there... I'm a little confused: has it been applying the source in completely the wrong place? Previously it would have been only applied where y > W/2, so above the rectangle?
| def strike_slip(y1, y2): | ||
| d_bar = y2 * sn - q * cs | ||
| r = sqrt(y1**2 + y2**2 + q**2) | ||
| a4 = 2 * poisson * (ln(r + d_bar) - sn * ln(r + y2)) / cs |
There was a problem hiding this comment.
Another AI observation: we're dividing here by cs, cosine of the dip, but in the OkadaArrayTsunamiSource, we bound 0<=dip<=90 so in an optimisation it's allowed to have dip=90 which would blow things up. Might well be that this is fine, if that in fact means the optimisation will steer things away from dip=90.
2 participants
Minor update that's been on my to do list for a while: create a Thetis module for the tsunami utils. Would allow us to consider tsunamis other than Tohoku and will also allow me to write a Tohoku demo (also on the to do list).