SWAN WAXA 500m wave hindcast

Brief description

This collection contains a regional SWAN wave model configured over the southwest WA region at 500m resolution, run from 1980 through to November 2021, produced in 2022.

The SWAN wave model was configured on an extended “WAXA” domain from around Guilderton, WA to Gnarabup, WA and extending offshore to 114.4°E.
The model was forced on the boundary with CAWCR Wave Hindcast spectral wave data. Wind forcing in the domain was provided from a CCAM downscaling of ERA5 data at 12km resolution. Bathymetry was an internally generated product comprising blended low-resolution GEBCO bathymetry, GA250 bathymetry/topography and high-resolution LiDAR-survey derived bathymetry/topography.
The hindcast was run using an Airflow workflow tool to concurrently run all months of the 40 years on the high performance computer, by adding spin up time to the start of each month.

The model was run from 1980 to 2021 inclusive, however due to spin up time, January 1980 had to be run separately, and December 2021 could not be run due to needing data from January 1, 2022 which was not available.

The data is arranged in monthly folders (of the form YYYYMM01.000000_0), each month contains 3 netCDF files:
- swan_out.nc, the gridded output,
- spec_out.nc and schism_bnd_spec_out.nc, which are spectral output across the domain, and along the boundary of a nested SCHISM-WWM3 hydrodynamic-wave coupled littoral model to be nested inside this domain.



Lineage: Wave boundary conditions are taken from the CAWCR Wave Hindcast hourly spectral output points on the grid boundary.
Wind boundary conditions are 10m near-surface winds at 30-minute intervals downscaled from the ERA5 hindcast using CCAM (input data courtesy Marcus Thatcher and Chloe Mackallah) at approximately 12.5km resolution.

We have developed an open-source python library called rompy (“Relocatable Ocean Modelling in PYthon”) to allow us to easily import data and other code required for this activity. The code is shared under a BSD 3-Clause licence. Documentation for this library is
available at https://rom-py.github.io/rompy/.

The SWAN model is configured in an Airflow workflow development environment using the rompy library, with the model jobs submitted to run on CSIRO’s Petrichor High Performance Computer (HPC). The directed acyclic graph (DAG) executed by Airflow monitors for model results and submits jobs as data becomes available.

Output data is stored on CSIRO Bowen cloud infrastructure, and published here and via THREDDS.