WHACS: Wave Hindcast for the Australian Climate Service

Brief description

WHACS, the Wave Hindcast for the Australian Climate Service, is an ocean wave hindcast running from 1979 to the present* (2023 at the time of publication).
WHACS is designed to replace the CAWCR Wave Hindcast (Durrant, Hemer, Smith, Trenham & Greenslade (2019) http://hdl.handle.net/102.100.100/137152).
A manuscript describing this dataset is in preparation.

WHACS data was generated using the WAVEWATCH III v6.07 wave model using ST6 physics, forced with ERA5 hourly winds and daily sea ice. The model was run on a spherical multi-cell (SMC) grid with a base resolution of 0.25 degrees in the polar regions and 0.125 degrees the non-polar global ocean, and a higher resolution of ~7km in the coastal zone globally.

The dataset comprises 4 components: the full SMC grid data; regridded data at 1/8deg globally; regridded data at 1/16deg in the Australian region, and spectral outputs.
The model produced spectral wave output at 11224 points: 286 buoys; 370 points on a global 10-degree grid; 10055 points from the coastal SMC grid in the Australian, New Zealand, Indian and Pacific Islands and the Maritime Continent regions; an additional 80 points around NZ; and 443 points that define the SCHISM-WWMIII based Coupled Coastal Hazard Prediction System (CCHaPS) hindcast.

Examples of how to work with the data in Python are provided in the accompanying notebook "whacs_examples.ipynb" (see Supporting files).

Further information will be provided in an accompanying paper which will be linked to this metadata record in due course.

N.B. December 1978 was run as a "model spin-up" month, data should be reliable from January 1979.

SMC parameters: longitude; latitude; time; uwnd; vwnd (Eastward and Northward wind); hs (significant wave height); t01(mean period T01); t02 (mean wave period T02); t0m1 (mean period T0m1); fp (peak wave frequency); dir (mean wave direction); spr (directional spread); dp (peak direction); phs0; phs1; phs2; phs3 (significant wave height partitions); ptp0; ptp1; ptp2; ptp3 (peak period partitions); pdir0; pdir1; pdir2; pdir3 (mean wave direction partitions); pspr0; pspr1; pspr2; pspr3 (wave directional spreading partitions); pdp0; pdp1; pdp2; pdp3 (peak wave direction partitions); pqp0; pqp1; pqp2; pqp3 (wave peakedness partitions); ppe0; ppe1; ppe2; ppe3 (wave peak enhancement factor partitions); CgE (wave energy flux).

Regridded parameters: longitude; latitude; time; hs (significant wave height); uwnd; vwnd (Eastward and Northward wind); t01(mean period T01); t02 (mean wave period T02); t0m1 (mean period T0m1); fp (peak wave frequency); dir (mean wave direction); dp (peak direction); phs0; phs1; phs2 (significant wave height partitions); ptp0; ptp1; ptp2 (peak period partitions); pdir0; pdir1; pdir2 (mean wave direction partitions); CgE (wave energy flux).

Spectral wave parameters: efth (sea surface wave directional variance spectral density) including time; station; longitude; latitude; frequency; frequency1; frequency2 (centre, upper and lower bands); direction; dth (depth); wnd; wnddir (wind speed and direction 10m above surface).

Some corrections were made to the data during post-processing:
- CgE took negative values in sea ice regions and due to numerical overflow in some cases. These values were reset to NaN or brought within range by changing the datatype from int16 to uint16.
- fp took negative values sometimes due to a known bug in this version of WW3 (https://github.com/NOAA-EMC/WW3/pull/741)
- Negative values of pqp* variables were reset to the maximum allowed, however such values would be considered nonphysical (https://github.com/NOAA-EMC/WW3/issues/210), If using the partitioned Goda's peakedness parameter variables, please do so with caution.
Lineage: The WHACS hindcast was performed on NCI's Gadi HPC using the WAVEWATCH-III model v6.07 using the ST6 physics formulation. The model was run on a global SMC grid at ~12.5km globally and ~7km in most coastal zones. Wave spectra were discretised over 28 frequencies exponentially spaced from 0.04118 Hz to 0.54 Hz and 30 directions with a constant 12° directional resolution. The model was forced with ERA5 surface winds, surface pressures, and sea ice concentrations at hourly resolution. Data output in NetCDF4 format. This configuration was based on the operational AUSWAVE forecast model operated by the Bureau of Meteorology, but the model was tuned by varying the WCOR parameter to boost winds to better capture extreme conditions.

The SMC data was post-processed to re-chunk for usability: changing the chunking from (1, 1270610) to (372, 11551) for enhanced performance on time-series analysis. Regridded data at 1/8 degree (globally) and 1/16 degree (Australian region) was also rechunked to improve performance. Metadata was rewritten to enhance compliance with CF and ACDD metadata standards.

The spectral data was post-processed to re-chunk for usability: changing the chunking (time, station, frequency, direction) from (1, 443, 28, 30) to (372, 1, 28, 30) for enhanced performance. Metadata was rewritten to ensure compliance with CF and ACDD metadata standards. Please note that `direction` is now encoded as an unsigned integer and may need to be sorted to work with some tools like python's wavespectra. Our thanks to Emilio Echevarria for assistance with this dataset.