janahaddad / adcircpy

Python library for managing input and output files for the ADCIRC model

ADCIRCPy

Python library for automating ADCIRC model runs.

Documentation can be found at https://adcircpy.readthedocs.io

Organization / Responsibility

ADCIRCpy is currently maintained by the Coastal Marine Modeling Branch (CMMB) of the Office of Coast Survey (OCS), a part of the National Oceanic and Atmospheric Administration (NOAA), an agency of the United States federal government. It was majorly developed by @jreniel.

• Zachary Burnett (lead) - [email protected]
• Jaime Calzada - [email protected]

Installation

Please use a virtual environment with Python>=3.6. You may use conda or the OS's Python to provide a virtual environment for the application.

You may install the application though pip. This will install the latest tagged version.

pip install adcircpy

Alternatively, you many manually install the repo by cloning it and then running

pip install .

Usage

Command Line Interface (CLI)

This program exposes a few commands available from the command line interface. You may pass the -h flag to any of this commands to explore their functionality.

• tide_gen
• plot_mesh
• tidal_run
• best_track_run
• best_track_file
• plot_maxele
• plot_fort61
• fort63

examples

generate tidal constituent template

You can quickly create a tidal component table for your your mesh by executing the tide_gen command and by passing a mesh, a start date and number of run days as arguments. This functions sources data from the HAMTIDE database by default.

tide_gen \
    /path/to/your/fort.14 \
    2021-02-26T00:00:00 \
    15 \
    --mesh-crs='epsg:4326'

run best-track run for Hurricane Sandy (AL182012)

To create the ADCIRC input files includes both tides and storm data for Hurricane Sandy:

best_track_run \
    /path/to/your/fort.14 \
    Sandy2012 \
    --fort13=/path/to/your/fort.13 \
    --crs=EPSG:4326 \
    --output-directory=/path/where/you/want/the/files \
    --constituents=all \
    --spinup-days=15.0 \
    --elev=30. \
    --mete=30. \
    --velo=30. \
    --skip-run

Note that the --crs flag is required due to the fort.14 not containing Coordinate Reference System information which is required for correct operation. EPSG:4326 means that the mesh is in WGS84 ( lat/lon). Note that the backlash represents "continue on next line" for the shell. You may write the command above on a single line after excluding the backslashes.

plot results

These are two examples for doing quick plots with the package. These are given here as illustrative examples only. There is support for more file types than this examples, but the program does not yet support every output input/output file type. As a user, you are encouraged to explore what's available and suggest and contribute your improvements.

plot_fort61 /path/to/fort.61.nc MSL --show --coops-only

plot_mesh /path/to/fort.14 --show-elements

Python API

See the examples directory for usage examples.

example_1.py

The following code builds a simple ADCIRC run configuration by doing the following:

1. reads a fort.14 mesh file (specifically a test mesh for Shinnecock Inlet)
2. adds tidal forcing to the mesh
3. creates an AdcircRun driver object with the mesh, including start and end dates
4. overrides default model options in the resulting fort.15
5. runs ADCIRC if present, otherwise writes configuration to disk

from datetime import datetime, timedelta
from pathlib import Path
import shutil

from adcircpy import AdcircMesh, AdcircRun, Tides
from adcircpy.utilities import download_mesh, get_logger

LOGGER = get_logger(__name__)

DATA_DIRECTORY = Path(__file__).parent.absolute() / 'data'
INPUT_DIRECTORY = DATA_DIRECTORY / 'input' / 'shinnecock'
OUTPUT_DIRECTORY = DATA_DIRECTORY / 'output' / 'example_1'

MESH_DIRECTORY = INPUT_DIRECTORY / 'shinnecock'

download_mesh(
    url='https://www.dropbox.com/s/1wk91r67cacf132/NetCDF_shinnecock_inlet.tar.bz2?dl=1',
    directory=MESH_DIRECTORY,
    known_hash='99d764541983bfee60d4176af48ed803d427dea61243fa22d3f4003ebcec98f4',
)

# open mesh file
mesh = AdcircMesh.open(MESH_DIRECTORY / 'fort.14', crs=4326)

# initialize tidal forcing and constituents
tidal_forcing = Tides()
tidal_forcing.use_constituent('M2')
tidal_forcing.use_constituent('N2')
tidal_forcing.use_constituent('S2')
tidal_forcing.use_constituent('K1')
tidal_forcing.use_constituent('O1')
mesh.add_forcing(tidal_forcing)

# set simulation dates
duration = timedelta(days=5)
start_date = datetime(2015, 12, 14)
end_date = start_date + duration

# instantiate driver object
driver = AdcircRun(mesh, start_date, end_date)

# request outputs
driver.set_elevation_surface_output(sampling_rate=timedelta(minutes=30))
driver.set_velocity_surface_output(sampling_rate=timedelta(minutes=30))

# override default options so the resulting `fort.15` matches the original Shinnecock test case options
driver.timestep = 6.0
driver.DRAMP = 2.0
driver.TOUTGE = 3.8
driver.TOUTGV = 3.8
driver.smagorinsky = False
driver.horizontal_mixing_coefficient = 5.0
driver.gwce_solution_scheme = 'semi-implicit-legacy'

if shutil.which('padcirc') is not None:
    driver.run(OUTPUT_DIRECTORY, overwrite=True)
elif shutil.which('adcirc') is not None:
    driver.run(OUTPUT_DIRECTORY, overwrite=True, nproc=1)
else:
    LOGGER.warning(
        'ADCIRC binaries were not found in PATH. '
        'ADCIRC will not run. Writing files to disk...'
    )
    driver.write(OUTPUT_DIRECTORY, overwrite=True)

example_2.py

The following code is similar to example_1.py, above, except it adds a static Manning's N coefficient to the mesh.

from datetime import datetime, timedelta
from pathlib import Path
import shutil

import numpy

from adcircpy import AdcircMesh, AdcircRun, Tides
from adcircpy.utilities import download_mesh, get_logger

LOGGER = get_logger(__name__)

DATA_DIRECTORY = Path(__file__).parent.absolute() / 'data'
INPUT_DIRECTORY = DATA_DIRECTORY / 'input'
OUTPUT_DIRECTORY = DATA_DIRECTORY / 'output' / 'example_2'

MESH_DIRECTORY = INPUT_DIRECTORY / 'shinnecock'

download_mesh(
    url='https://www.dropbox.com/s/1wk91r67cacf132/NetCDF_shinnecock_inlet.tar.bz2?dl=1',
    directory=MESH_DIRECTORY,
    known_hash='99d764541983bfee60d4176af48ed803d427dea61243fa22d3f4003ebcec98f4',
)

# open mesh file
mesh = AdcircMesh.open(MESH_DIRECTORY / 'fort.14', crs=4326)

# generate tau0 factor
mesh.generate_tau0()

# also add Manning's N to the domain (constant for this example)
mesh.mannings_n_at_sea_floor = numpy.full(mesh.values.shape, 0.025)

# initialize tidal forcing and constituents
tidal_forcing = Tides()
tidal_forcing.use_constituent('M2')
tidal_forcing.use_constituent('N2')
tidal_forcing.use_constituent('S2')
tidal_forcing.use_constituent('K1')
tidal_forcing.use_constituent('O1')
mesh.add_forcing(tidal_forcing)

# set simulation dates
spinup_time = timedelta(days=2)
duration = timedelta(days=3)
start_date = datetime(2015, 12, 14) + spinup_time
end_date = start_date + duration

# instantiate driver object
driver = AdcircRun(mesh, start_date, end_date, spinup_time)

# request outputs
driver.set_elevation_surface_output(sampling_rate=timedelta(minutes=30))
driver.set_velocity_surface_output(sampling_rate=timedelta(minutes=30))

# override default options
driver.timestep = 4.0

if shutil.which('padcirc') is not None:
    driver.run(OUTPUT_DIRECTORY, overwrite=True)
elif shutil.which('adcirc') is not None:
    driver.run(OUTPUT_DIRECTORY, overwrite=True, nproc=1)
else:
    LOGGER.warning(
        'ADCIRC binaries were not found in PATH. '
        'ADCIRC will not run. Writing files to disk...'
    )
    driver.write(OUTPUT_DIRECTORY, overwrite=True)

example_3.py

The following code is similar to example_1.py, above, except it adds HURDAT BestTrack wind forcing and also builds a Slurm job script for submission to a job manager.

from datetime import datetime, timedelta
from pathlib import Path

from adcircpy import AdcircMesh, AdcircRun, Tides
from adcircpy.forcing.winds import BestTrackForcing
from adcircpy.server import SlurmConfig
from adcircpy.utilities import download_mesh

DATA_DIRECTORY = Path(__file__).parent.absolute() / 'data'
INPUT_DIRECTORY = DATA_DIRECTORY / 'input'
OUTPUT_DIRECTORY = DATA_DIRECTORY / 'output' / 'example_3'

MESH_DIRECTORY = INPUT_DIRECTORY / 'shinnecock'

download_mesh(
    url='https://www.dropbox.com/s/1wk91r67cacf132/NetCDF_shinnecock_inlet.tar.bz2?dl=1',
    directory=MESH_DIRECTORY,
    known_hash='99d764541983bfee60d4176af48ed803d427dea61243fa22d3f4003ebcec98f4',
)

# open mesh file
mesh = AdcircMesh.open(MESH_DIRECTORY / 'fort.14', crs=4326)

# initialize tidal forcing and constituents
tidal_forcing = Tides()
tidal_forcing.use_all()
mesh.add_forcing(tidal_forcing)

# initialize wind forcing
wind_forcing = BestTrackForcing('Sandy2012')
mesh.add_forcing(wind_forcing)

# initialize Slurm configuration
slurm = SlurmConfig(
    account='account',
    ntasks=1000,
    run_name='adcircpy/examples/example_3.py',
    partition='partition',
    walltime=timedelta(hours=8),
    mail_type='all',
    mail_user='[email protected]',
    log_filename='example_3.log',
    modules=['intel/2020', 'impi/2020', 'netcdf/4.7.2-parallel'],
    path_prefix='$HOME/adcirc/build',
)

# set simulation dates
spinup_time = timedelta(days=15)
duration = timedelta(days=3)
start_date = datetime(2012, 10, 21, 18)
end_date = start_date + duration

# instantiate driver object
driver = AdcircRun(mesh, start_date, end_date, spinup_time, server_config=slurm)

# write driver state to disk
driver.write(OUTPUT_DIRECTORY, overwrite=True)

example_4.py

The following code is similar to example_3.py, above, except it uses ATMESH wind forcing and WW3DATA wave forcing.

from datetime import datetime, timedelta
from pathlib import Path

from adcircpy import AdcircMesh, AdcircRun, Tides
from adcircpy.forcing.waves.ww3 import WaveWatch3DataForcing
from adcircpy.forcing.winds.atmesh import AtmosphericMeshForcing
from adcircpy.server import SlurmConfig
from adcircpy.utilities import download_mesh

DATA_DIRECTORY = Path(__file__).parent.absolute() / 'data'
INPUT_DIRECTORY = DATA_DIRECTORY / 'input'
OUTPUT_DIRECTORY = DATA_DIRECTORY / 'output' / 'example_4'

MESH_DIRECTORY = INPUT_DIRECTORY / 'shinnecock'

download_mesh(
    url='https://www.dropbox.com/s/1wk91r67cacf132/NetCDF_shinnecock_inlet.tar.bz2?dl=1',
    directory=MESH_DIRECTORY,
    known_hash='99d764541983bfee60d4176af48ed803d427dea61243fa22d3f4003ebcec98f4',
)

# open mesh file
mesh = AdcircMesh.open(MESH_DIRECTORY / 'fort.14', crs=4326)

# initialize tidal forcing and constituents
tidal_forcing = Tides()
tidal_forcing.use_all()
mesh.add_forcing(tidal_forcing)

# initialize atmospheric mesh forcings (for NUOPC coupling)
wind_forcing = AtmosphericMeshForcing(
    filename='Wind_HWRF_SANDY_Nov2018_ExtendedSmoothT.nc', nws=17, interval_seconds=3600,
)
mesh.add_forcing(wind_forcing)

# initialize wave mesh forcings (for NUOPC coupling)
wave_forcing = WaveWatch3DataForcing(
    filename='ww3.HWRF.NOV2018.2012_sxy.nc', nrs=5, interval_seconds=3600,
)
mesh.add_forcing(wave_forcing)

# initialize Slurm configuration
slurm = SlurmConfig(
    account='account',
    ntasks=1000,
    run_name='adcircpy/examples/example_4.py',
    partition='partition',
    walltime=timedelta(hours=8),
    mail_type='all',
    mail_user='[email protected]',
    log_filename='example_4.log',
    modules=['intel/2020', 'impi/2020', 'netcdf/4.7.2-parallel'],
    path_prefix='$HOME/adcirc/build',
)

# instantiate driver object
driver = AdcircRun(
    mesh=mesh,
    start_date=datetime.now(),
    end_date=timedelta(days=7),
    spinup_time=timedelta(days=5),
    server_config=slurm,
)

# write driver state to disk
driver.write(OUTPUT_DIRECTORY, overwrite=True)

Citation

Calzada, J., Burnett, Z., Moghimi, S., Myers, E., & Pe’eri, S. (2021). ADCIRCpy: A Python API to generate ADCIRC model input files (Technical Memorandum No. 41; NOAA NOS OCS). National Oceanic and Atmospheric Administation.

@techreport{calzadaADCIRCpyPythonAPI2021,
    type = {Technical {{Memorandum}}},
    title = {{{ADCIRCpy}}: A {{Python API}} to Generate {{ADCIRC}} Model Input Files},
    author = {Calzada, Jaime and Burnett, Zachary and Moghimi, Saeed and Myers, Edward and Pe'eri, Shachak},
    year = {2021},
    month = dec,
    number = {41},
    institution = {{National Oceanic and Atmospheric Administation}},
    abstract = {The Advanced Circulation Model (ADCIRC) is a Fortran program used for modeling ocean circulation due to tides, surface waves and atmospheric forcings. However, the input formats and configuration are inflexible and not straight forward for operational implementation, making rapid iteration of model testing, ensemble configuration, and model coupling complicated. Here, we introduce a flexible abstraction of model inputs and outputs written in Python, called ADCIRCpy, that provides a simpler user interface for automatically generating ADCIRC configuration to a variety of inputs and model scenarios. This documentation outlines 1. the needs for such an abstraction, 2. the peculiarities and challenges with the ADCIRC model that necessitate custom logic, and 3. methodologies for generalizing user input in such a way as to make generating model configurations consistent, fast, and efficient.}
}

Acknowledgements

The majority of ADCIRCpy was written by Jaime Calzada @jreniel.