Difference between revisions of "Tsunami simulation"
| Line 2: | Line 2: | ||
Typically you need to follow these 2 steps in tsunami simulations: | Typically you need to follow these 2 steps in tsunami simulations: | ||
| − | + | <OL> | |
| − | + | <LI> Deformation run (EX15/Def/ in the sample run): this simulates the earthquake and the set-up of the initial surface | |
waves. For this you need bdef.gr3 (refer to the user manual) which | waves. For this you need bdef.gr3 (refer to the user manual) which | ||
specifies the total seafloor deformation. We typically run this stage for 10 sec duration and | specifies the total seafloor deformation. We typically run this stage for 10 sec duration and | ||
| Line 11: | Line 11: | ||
is used in the next stage. For completeness you need to attach the boundary condition | is used in the next stage. For completeness you need to attach the boundary condition | ||
part (b.tmp) of hgrid.gr3 to the end of hgrid.new; | part (b.tmp) of hgrid.gr3 to the end of hgrid.new; | ||
| − | + | <LI> Propagation and inundation stage (EX15/ in the sample run): this run continues from the deformation | |
run above with no further seafloor movement. You'll notice that many input files are identical | run above with no further seafloor movement. You'll notice that many input files are identical | ||
to the Def/ run, but be careful of differences in hgrid.gr3 (linked to Def/hgrid.new) and param.in | to the Def/ run, but be careful of differences in hgrid.gr3 (linked to Def/hgrid.new) and param.in | ||
| Line 22: | Line 22: | ||
If you have to use non-zero friction you'll need to use more layers and make sure the bottom boundary | If you have to use non-zero friction you'll need to use more layers and make sure the bottom boundary | ||
layer is resolved. If you use the recent parallel version, you can simply use a 2D model. | layer is resolved. If you use the recent parallel version, you can simply use a 2D model. | ||
| − | + | <LI> After the run is done you can look at global outputs (elevation, depth-averaged velocity etc). | |
The maximum elevation (maxelev.gr3) and depth-averaged velocity (maxdahv.gr3) are also part of the | The maximum elevation (maxelev.gr3) and depth-averaged velocity (maxdahv.gr3) are also part of the | ||
outputs (for parallel versions, use trunk/src/Utility/Combining_Scripts/combine_gr3.f90). The maximum inundation can be computed from maxelev.gr3. | outputs (for parallel versions, use trunk/src/Utility/Combining_Scripts/combine_gr3.f90). The maximum inundation can be computed from maxelev.gr3. | ||
| + | </OL> | ||
Revision as of 13:06, 23 July 2013
A sample tsunami run (impact of Alaska tsunami waves on Cannon Beach, OR) can be downloaded here. Note that the files are compatible with the serial version of SELFE, but the idea for parallel version is similar.
Typically you need to follow these 2 steps in tsunami simulations:
- Deformation run (EX15/Def/ in the sample run): this simulates the earthquake and the set-up of the initial surface waves. For this you need bdef.gr3 (refer to the user manual) which specifies the total seafloor deformation. We typically run this stage for 10 sec duration and at the end of the run you'll find an output files called 10_hotstart which is then used as hotstart.in for the next (propagation and inundation) stage. In addition, use mod_depth.f (inside the bundle), which takes info in bdef.gr3 and hgrid.gr3 (pre-earthquake depths) to generate hgrid.new (post-earthquake depths). The latter is used in the next stage. For completeness you need to attach the boundary condition part (b.tmp) of hgrid.gr3 to the end of hgrid.new;
- Propagation and inundation stage (EX15/ in the sample run): this run continues from the deformation run above with no further seafloor movement. You'll notice that many input files are identical to the Def/ run, but be careful of differences in hgrid.gr3 (linked to Def/hgrid.new) and param.in (imm=0 etc). For this run we also turn off the advection in the open ocean (up to say 20m isobath) because it's negligible there (adv.gr3) - see for info on CFL number. Also note that hotstart.in is from the previous stage. A relatively small time step of 1 sec (by SELFE standard) is used because of the constraint from inundation stage. The operational time step for tsunami applications is generally in the range of a few seconds. The inundation results may also be sensitive to the min. depth used in the run (1cm in this example). Also since only 2 S layers are used in the vertical (quasi-2D mode) we set the bottom friction to 0. If you have to use non-zero friction you'll need to use more layers and make sure the bottom boundary layer is resolved. If you use the recent parallel version, you can simply use a 2D model.
- After the run is done you can look at global outputs (elevation, depth-averaged velocity etc). The maximum elevation (maxelev.gr3) and depth-averaged velocity (maxdahv.gr3) are also part of the outputs (for parallel versions, use trunk/src/Utility/Combining_Scripts/combine_gr3.f90). The maximum inundation can be computed from maxelev.gr3.
