An Ensemble Ocean Wave Forecast System and Preliminary Experiment

The basic methods of ensemble wave prediction at home and abroad and an ensemble wave numerical forecast system are introduced. The system is established by running the operational wave model WAVEWATCH Ⅲ using wind data from the 15 members' ensemble weather forecast system, which is based on the T213L31 model and operational running in China National Meteorological Center. The ensemble model calculates 15 members wave field, including one control forecast and 14 perturbation members forecast. The model computes the waves over all the oceans up to 10 days at 12:00 with 1° by 1° resolution. Based on hot initialization, the model uses the 12 h wave hindcast as its initial field. Through the information extraction and processing from the 15 member's wave fields, the system produces the ensemble wave products such as ensemble mean and spread, spaghetti charts and probability charts. Forecast experiment for the period from 1 Sep to 31 Oct in 2007 is done using the system and buoys data in the Pacific and Atlantic at the same period. The main verifying indexes are bias and root mean square error. For the mean monthly bias of 10 m wind speed and significant wave height, the performance of the ensemble mean is better than the control forecast, however both of them have a relatively low system deviation and one cause is that the spin up time of the system may not be long enough. For the monthly root mean square error, the ensemble mean of 10 m wind speed is lower than the control forecast by 3—12 percentage for the different forecast periods, meanwhile the ensemble mean of significant wave height is lower than the control forecast by 3—17 percentage with consistent variation trend. Comparing 15 ensemble members' forecast to the Buoy_46006 data, it shows that the ensemble members can follow the changes of the observation well and cover the variation range of the observation other than the deterministic forecast. In summary, the verification results show that the ensemble technique can provide a successful method of extending the standard deterministic forecasts to the probabilistic domain and the prediction skill of the ensemble mean is better than that of one single deterministic forecast. The ensemble method could provide additional information that could not be gleaned from a purely deterministic approach and has a good prospect of application in the field of wave forecast.