ocean/hurricane test group defines meshes,
initial conditions, forward simulations, and validation for global,
realistic ocean domains with regional refinement. These simulations
are forced with time-varying atmospheric reanalysis data for tropical
cyclone events and tides. The meshes contain refined regions in order
to resolve coastal estuaries, making it possible to simulate the
storm surge that results from a given hurricane event.
Currently, one mesh resolution and one storm, Hurricane Sandy, are supported.
More mesh resolutions will be supported in the future.
These test are configured to use the barotropic, single layer configuration of MPAS-Ocean. Each mesh can optionally be created to contain the floodplain which can be used to simulate coastal inundation using MPAS-Ocean’s wetting and drying scheme.
The time stepping options to run the simulations include the fourth
order Runge-Kutta scheme (RK4), which is the default, and a local
time-stepping (LTS) scheme, see
Lilly et al. (2023)
for details. Each test case in the
ocean/hurricane test group has an
LTS counterpart which is identified by the
_lts appendix in the test case name.
Note that the executable to be used with LTS has to be compiled with
the following options:
mesh test case¶
The mesh test case uses the mesh step from the global ocean test group.
First, it generates the global mesh based on a specified mesh resolution
function. Next, bathymetry/topography data is interpolated on the mesh from the
STRM15+ data product. This interpolation step is necessary, because the
topography in the floodplain is used to set a mask for the cell culling
process. The land cells above the
floodplain_elevation are then culled
from the mesh. Finally, the bathymetry is re-interpolated onto the mesh
since this data is not carried over from the cell culling process.
If the LTS option is selected for the mesh test case, an additional step is carried out after the mesh culling. This step appropriately flags the cells of the mesh according to a user defined criterion in order to use time-steps of different sizes on different regions of the mesh. The parallel partitioning is modified accordingly to achieve proper load balancing.
init test case¶
The init test performs steps to set up the vertical mesh, initial conditions, atmospheric forcing, and prepares the station locations for timeseries output.
initial state step¶
The initial state step runs MPAS-Ocean in init mode to create the initial condition file for the forward run. The vertical mesh is setup for a single layer case and the ssh with a thin layer on land for wetting and drying cases.
interpolate atmosphere forcing step¶
The CFSv2 reanalysis wind vector components and atmospheric pressure fields for the storm event are interpolated onto the horizontal mesh at hourly intervals. These are read in and used to update the atmospheric forcing in the forward run.
create pointstats file step¶
In order to perform validation of the forward simulation, timeseries data is recored at mesh cell centers which are closest to observation stations. This set reads in the observation station locations and finds the cells closest to them. A file is created that is the input to the pointWiseStats analysis member for the forward run.
compute topographic wave drag step¶
This step is carried out only if the LTS option is selected for the init test case.
The reciprocal of the e-folding time,
r_inv, from the HyCOM model,
is computed in this step. See
Buijsman et al. (2016)
for details on the computation. This coefficient is needed to account
for the topographic wave drag tendency in the model.
sandy test case¶
The sandy test case is responsible for the forward model simulation and analysis.
The forward step runs the model simulation of the storm. The simulation begins with a spinup period, where the tides and atmospheric forcing are ramped to their full value to avoid shocking the system.
The analysis step plots the timeseries data at each observation station to compare the modeled and observed data. Both NOAA and USGS station data is used for the validation.
If the LTS option is selected for the sandy test case, the LTS scheme is used to advance the solution in time rather than the default RK4 scheme.