Setting up and running GeoClaw#
This is a rough outline with some pointers to existing documentation and examples. More will be added.
See also the GeoClaw documentation and Other learning resources.
Things to set/adjust in setrun.py#
See also
Copalis Beach setrun
description
for an annotated setrun script explaining the commonly changed parameters.
Computational domain#
How large should it be? (Outflow boundaries)
Should edges be adjusted so that cells on some fine level are aligned to match some desired grid?
Choosing grid resolutions and refinement factors#
How coarse should the initial (Level 1) grid be?
How many AMR levels are needed, what refinement factors are best?
Topography DEMs (topofiles)#
Sources of data (needs updating)
Adjusting background sea level#
Relation to vertical datum of the topography files.
Choice of tide level for simulation(s), e.g. MHW is often used for modeling onshore inundation, MLW for modeling drawdown in marinas.
Incorporating future sea level projections.
How
sea_levelis adjusted by ground deformation for initializing water level on fine grids, see Set Eta Init documentation
Specifying ground motion (dtopofiles)#
Coastal subsidence / uplift
Choice of
dtopo_data.dt_max_dtopofor static vs. kinematic sources
Refinement criteria and flagregions#
Finite volume methods, cell-averaged values, and consistency between levels
Specifying flagregions for adaptive refinement documentation
Using Ruled Rectangles for refinement regions, e.g. MakeFlagregionsCoast notebook
Guiding AMR with adjoint flagging See also the example in
$CLAW/geoclaw/examples/tsunami/chile2010_adjointand its README in the Clawpack Gallery. Additional examples in the paper [DL16].
Duration of simulation and time between outputs#
Size of output files,
asciivsbinaryoutputTime frame output (full AMR solution) vs.
fgoutoutput (on fixed grids)Choice of Courant number,
cfl_desired
Synthetic gauges#
Centering gauges in grid cells
Capturing pre-seismic and/or final topography at gauge points
When to turn gauges on
fgmax grids to capture maxima#
Centering fgmax grid points in computational grid cells
What levels to monitor, when to turn on
Capturing pre-seismic and/or final topography at grid points
fgout grids for frequent output on fixed grids#
Bottom friction and the Manning Coefficient#
Other topics#
Initializing some land below sea level as dry#
Land that is behind a dike or levee should be forced to be dry initially.
This can be done with a force_dry array.
Initializing some land above sea level as wet#
A lake that is near the coast may have an initial surface elevation that is
different from sea_level.
Modifying
set_eta_init.f90to changesea_levelover specified region.See the sample code commented out in the default set_eta_init.f90
Modifying
Makefileto used your version of this Fortran code, see Makefile documentation
Adding river flow, precipitation, overland flooding#
Landslide generated tsunamis#
Need for dispersive modeling, see Boussinesq solver documentation
Running GeoClaw#
Basic workflow and output#
Using OpenMP, how many threads?#
Checkpoint files and restarts#
Running on cluster / supercomputer#
Sending output to a scratch directory
SLURM scripts
Running dispersive (Boussinesq) code with MPI, PETSc on TACC