Analyses & Numerical Prediction

Operations Bulletin No. 47
Changes to the Operational Sea State Forecast System
August 1999

Introduction

The WAM wave model has been the operational sea state model in NMOC since 1 June 1994 and its configuration was described in Analysis and Prediction Operations Bulletin No. 29 of 30 May 1994. Details of the model formulation are contained in WAMDI Group (1988), while the finite differencing scheme adopted in the Australian implementation of the model is detailed in BMRC Research Report No 43.

While the physics and numerics of the model remain essentially unchanged this bulletin describes changes made to incorporate the assimilation of ERS-2 altimeter significant wave height data during the hindcast period together with an up to date summary of the current operational configuration.

Minor changes have also been made to the model code to enable Y2000 compliance.

Altimeter Data Assimilation

Significant wave height data from the ERS fast delivery product set are received in NMOC in near real time. The development of the scheme to incorporate these data into the model has been carried out in the Bureau of Meteorology Research Centre by and is described in detail in BMRC Research Report No. 73. The assimilation is achieved in two steps - combination of the model first-guess significant wave height field with the altimeter data followed by a modification of the model wave energy spectrum, the prognostic variable, as a consequence.

Incorporation of the altimeter data into the significant wave height field is achieved using univariate 2-D statistical interpolation. The RMS errors for both model prediction and observation have been set at 0.5 metres, based on comparisons with wave-rider buoy data, and the decorrelation length scale has been set at 300 Km. Insertion of the satellite data occurs every three hours over the twelve hour hindcast period

As significant wave height is proportional to the square root of the wave energy integrated over all frequencies and directions no single method of adjusting the wave energy spectrum exists. The strategy adopted has been to constrain the slope of the spectrum to remain constant thereby reducing the rate of dissipation of the assimilated information. This entails shifting energy to different frequencies as the total energy is altered in response to changes in significant wave height. Energy density at the highest and lowest model frequencies is not altered so as to avoid the loss of energy, although in practice these values are usually found to be zero.

Comparison of the model performance with buoy observations during a parallel trial conducted in July 1999 showed the global model forecasts to be improved or equal at all sites and the Australian Region forecasts to be equal or improved at all but three sites. RMS errors obtained during the parallel trial are summarised in table 1.

Table 1. RMS Error in metres for 12 and 36 hour forecast periods.

Similarly, comparison with satellite significant wave height observations showed an improved performance in both domains, although it should be noted that the satellite data themselves have a positive bias when compared to buoy data.

Model Configuration

The model is run twice daily, with base times of 00 and 12 UTC, on three domains: Global, Australian Region and Southeast Coast. Details of these are summarised in table 2 below.

Table 2. Model Configuration Summary.

SE Coast model as the third order upwinding scheme employed in the other models is incompatible with use of the shallow water equations.

Each run consists of a 12 hour hindcast period using analysed winds from the relevant atmospheric model followed by a forecast using prognostic winds. Forecasts are made to base time plus 96 hours for the global system and base time plus 36 hours for the other systems.

Output

Forecast wave energy spectra are post-processed to provide output fields of the following variables:
Significant wave height Wave direction 10m Wind speed 10m Wind directionMean period Peak period Swell wave height Wind wave heightMean swell direction Mean swell period Mean sea directionFriction velocity Drag coefficient Normalised wave stress

Plots of model forecasts compared with buoy observations together with summary tables of forecast verification for the previous 30 days are produced in each run for the following sites:
North Rankin Jurien Bay Rottnest Island Cape Naturaliste Byron Bay Coffs Harbour Crowdy Head Sydney Bondi Port Kembla Batemans Bay Eden Kingfish B Cape Sorell

Access to these fields is as follows:

Global System

Difacs chart numbers are:

1 Southern Hemisphere Significant Wave Height +36 hrs

628 Southern Ocean Significant Wave Height 00 hrs

629 Southern Ocean Significant Wave Height +12 hrs

630 Southern Ocean Significant Wave Height +24 hrs

631 Southern Ocean Significant Wave Height +36 hrs

632 Southern Ocean Significant Wave Height +48 hrs

635 Indian Ocean Swell Wave Height +24 hrs

636 Indian Ocean Swell Wave Height +48 hrs

637 Indian Ocean Wind Wave Height +24 hrs

638 Indian Ocean Wind Wave Height +48 hrs

948 Indian Ocean Significant Wave Height +48 hrs

949 Pacific Ocean Significant Wave Height +48 hrs

Australian Region System

Difacs chart numbers are:

46 Australian Region Significant Wave Height +24 hrs

47 Australian Region Significant Wave Height +36 hrs

50 Australian Region Swell Wave Height +24 hrs

140 Australian Region Swell Wave Height +36 hrs

49 Australian Region Wind Wave Height +24 hrs

141 Australian Region Wind Wave Height +36 hrs

Further Enquiries

Routine enquiries regarding product availability or access should be made to the NMOC Help Desk on (03) 9662 2182. Further information on the implementation of the WAM model can be obtained from Barry Southernon (03) 9669 4502 or B.Southern@BoM.GOV.AU.

References

Bender, L.C and Leslie, L.M. (1994). 'Evaluation of a third generation oceanwave model for the Australian region'. BMRC Research Report No 43. Bur.Met. Australia.

Greenslade, D. (1999). 'The Assimilationof ERS-2 Altimeter Data into the Australian Wave Model'. BMRC Research Report No 73. Bur.Met. Australia

WAMDI Group (1988): Hasselman, S., Hassleman, K., Bauer, E., Janssen,P.A.E.M., Komen. G.J., Bertotti, L., Lionello, P., Guillaume, A., Cardone,V.C., Greenwood, J.A., Reistad, M., Zambresky, L., and Ewing J.A., 'The WAM model - A third generation wave prediction model'. J.Phys. Oceanogr., 18, 1775-1810.