WW3 Standup Meeting Minutes

COSIMA Waves
1

Date: 29 January 2026
Attendees: @ezhilsabareesh8, @NoahDay, @sofarrell, @alberto
Chair & Minutes: @ezhilsabareesh8

1. ERA5 Forcing & Standalone WW3

  • ERA5 used as primary wind input for standalone WW3.
  • PR1 propagation scheme may be contributing to wave behaviour issues.
  • Use ERA5 neutral wind speed, which corrects for atmospheric instabilities.
  • Corrected ECMWF ERA5 dataset expected Q2 2026.

2. Wave–Current Interaction

  • Currents influence wave energy balance and redistribute energy across frequency bands.
  • Mean wave period biases may be caused by Doppler shifting from currents.
  • MOM6 current representation quality requires assessment.
  • Standalone WW3 should be tested with and without current forcing (e.g., ACCESS‑OM2 currents).

3. Excessive Low‑Frequency Dissipation

  • Too much dissipation observed in low‑frequency bands, likely from strong refraction in swell regions.
  • ST6 parameters, tuned for uncoupled WW3, may require retuning for coupled model.

4. Unresolved Obstacles Impacting Wave Biases

  • Significant wave height (Hs) biases appear near small islands and unresolved bathymetric features.
  • Dissipation parameterisation for unresolved obstacles should be explored.

5. Directional Spectra Evaluation

Directional spectra comparisons to be conducted with:

  • WHACS dataset
    CSIRO Data Access Portal
  • Southern Ocean Time Series (SOTS) directional spectra
  • NDBC deep‑water buoys (offshore stations)

6. Wave–Ice Considerations

  • Consider inclusion of ERA5 sea‑ice concentration in standalone WW3.
  • Compare attenuation behaviour for solid vs broken ice regimes.

7. Mixing & MOM6 Diagnostics

Ezhil’s tests indicate that WW3 EFACTOR improves MLD performance in MOM6 over:

  • KPP (no waves)
  • ePBL

Add the following vertical mixing diagnostics to MCW KPP runs (with and without EFACTOR):

diff_cbt_t      # vertical heat diffusivity
diff_cbt_s      # vertical salt diffusivity
diff_cbt_back   # background tracer diffusivity
diff_cbt_tides  # tidal tracer diffusivity

8. Action Items

8.1 Wave Model Evaluation (Ezhil)

  • Run standalone WW3 (ERA5 winds) and compare with:
    • WHACS
    • SOTS directional spectra
    • NDBC offshore buoys
  • Analyse PR1 propagation scheme.
  • Test WW3 with/without currents.
  • Investigate low‑frequency dissipation & refraction.
  • Evaluate WW3 wind‑correction switches.
  • Develop parameterisation for unresolved obstacles.

8.2 Data & Forcing (Ezhil)

  • Use ERA5 neutral wind speeds.
  • Assess feasibility of ERA5 sea‑ice concentration.

8.3 Wave–Ice Coupling (Noah)

  • Compare broken vs solid ice attenuation regimes.
  • Validate model against ice‑affected directional spectra.
  • Compare attenuation with CICE wave forcing
  • Evaluate fast‑ice vs broken‑ice behaviour
  • Optimise breakup scheme
2

Date: 18 March 2026
Attendees: @NoahDay, @ezhilsabareesh8, @lgbennetts, @sofarrell, @cbull, @AlbertoMeucci
Chair & Mintues: @NoahDay

To-do:

  • [ ] Investigate using 20-year KPP spinup as initial conditions for EPBL runs to avoid World Ocean Atlas heat issues

  • Determine if WaveWatch standalone can be run with sea ice concentration forcing via CDEPS

  • Check CICE code for numerical issues causing uptick in largest floe size category

Model Validation: WaveWatch3 Standalone vs Coupled

  • Comparing standalone WaveWatch3 with ERA5 and JRA55 forcing against coupled MCW (MOM6-CICE6-WW3) with validation from WHACS
  • First-order propagation scheme is known to leak energy behind islands
  • Can continue with first-order propagation scheme for this grid

Key Findings on Model Biases

  • ERA5-forced standalone WW3 performs best, as expected since it uses similar physics to WHACS (apart from the propagation scheme)

  • Polar ocean biases (both Arctic and Antarctic) should be ignored in standalone runs due to lack of sea ice representation

  • WHACS uses simple sea ice attenuation: 0-25% concentration allows waves, 25-75% has linear decay, 75%+ blocks waves

MCW Wave Results

  • Coupled model shows increased scatter but data clusters around 1:1 line

  • Should not necessarily expect one-to-one agreement when coupling due to ocean currents and other feedbacks

  • Negative bias appears when coupling, magnitude needs further investigation

Langmuir mixing

  • Compared KPP mixing scheme vs EPBL mixing scheme with and without wave-induced mixing

  • Blue line (KPP) and dotted red line (EPBL without waves) show very similar mixed layer depths

  • Some deviation appears in Antarctic region (60°S) during July-September

  • Summer mixing (January-March) shows disappointing lack of difference—Siobhan was hoping for more mixing

  • MC 25km ePBL performs best against obs

Initial Conditions Issue

  • ePBL additive scheme appears to be convecting sub-surface heat too quickly

  • World Ocean Atlas initial conditions may contain excess subsurface heat in Weddell and Ross Seas

  • Use 20-year MCW KPP spinup as initial conditions for ePBL runs

Wave Propagation Through Sea Ice

  • Comparing model against satellite transect data

Sea Ice Concentration Validation

  • MCW ice edge location is similar to satellite data but ice concentrations differ in marginal ice zone

  • There may be discrepancies in sea ice concentrations from different remote sensing product

  • West of Ross Sea shows different ice edge in model compared to observations

  • Model shows too little summer ice—standard issue in ACCESS-OM2

WaveWatch Standalone with Sea Ice

  • Current coupler may not support ice concentration forcing, would need CDEPS setup

Modifying wave-induced fracture in CICE

  • New approach reduces ice in first category, increases ice in categories 2–4

  • Differences most apparent in marginal ice zone, minimal in consolidated ice

  • Still seeing uptick in largest floe size category—suspected to be numerical artifact in the MIZ

  • FSDs produced are more stable with lower variability between realizations

  • Method is obvious improvement but makes model ~10% slower

3

Date: 27 May 2026
Attendees: @NoahDay, @ezhilsabareesh8, @lgbennetts, @sofarrell, @anton, @AlbertoMeucci
Chair & Minutes: @NoahDay

To-do:

  • Write up a report with the improvements to wave-induced ice fracture (both computational and theoretical)
  • Complete profiling of MCW-100km (particularly tuning DATM and CICE)
  • Compare MCW-100km-ERA5 with ACCESS-OM2 ERA5 and ERA5 sea ice extents
  • Run MCW-100km-ERA5 on the new 100km grid

MCW-100km alpha release model parameters

  • WW3 wave attenuation: IC4M8 (Meylan et al., 2021)
  • MOM6 mixing scheme: KPP (e-factor Langmuir turbulence)
  • DATM: ERA5 (12 streams)
  • CICE (Icepack): wave_spec_type = 'ICDR' (ice-covered dispersion relation solver)

Wave-induced fracture option (wave_spec_type)

  • Using wave_spec_type='random' can increase the model cost by 400%
  • Push new ICDR fracture method into ACCESS-NRI’s Icepack fork, so that the alpha release can be profiled with this my efficient method

C-grid for CICE

  • Possible to use, but requires changing ice strength to Hibler and decreasing the dynamic timestep in CICE (quite considerably)

MCW-100km-ERA5 evaluation

  • KPP is producing reasonable MLDs and SSTs
  • Sig. wave heights biases are relatively low (-6 cm in WW3-dice and -11 cm in MCW), the difference is likely from sea ice covers being in different locations
  • Mean wave period biases are ~0.5 s for both WW3-dice and MCW
  • Arctic sea ice extent agrees well with NSIDC, Antarctic sea ice extent is too large (~20 million km^2 in winter)

Antarctic sea ice extent

  • MCW-100km-ERA5 is producing too much sea ice in winter, but not enough in summer
  • We will compare MCW with ACCESS-OM2 runs using ERA5 forcing as well as the sea ice extents directly calculated from ERA5