Multi-week Prediction of Significant Wave Heights over the Western Pacific and Indian Ocean

The performance of multi-week predictions of significant wave height (SWH) in the Western Pacific and Indian Ocean (WP-IO) region are investigated. The Wavewatch-3 (WW3) model is forced with daily 10m winds from re-forecasts made using the National Centers for Environmental Prediction (NCEP) Climate Forecast System version-2 (CFSv2), initialized with four sets of ocean initial conditions. Re-forecasts using January and May initial conditions for the period 1979-2008 are considered. For each OIC, four ensemble members were generated by using atmospheric and land surface initial conditions (AIC and LIC) taken from the instantaneous fields at 0000 UTC of the first 4 days in January and May of each year. The total number of ensemble members was 16 = 4 OICs X 4 AIC/LICs – for both January and May cases. The main features of the climatological mean 10m winds in weekly means 1-4 are well captured by CFSv2, although the magnitude of the bias increases with lead-time over much of the region in both January and May cases. The CFSv2-WW3 system similarly captures the magnitude and spatial structure of SWH in weeks 1-4 well in both cases; however, the magnitude of positive biases increases with lead-time over the Southern Ocean (SO), South China Sea and northwestern Pacific region in the January cases, and over the SO in the May cases. The magnitude of the SWH variability grows weaker with lead-time over SO, which may be related to the weaker inter-annual variability of predicted 10m winds. During the first 2 forecast weeks, the temporal anomaly correlation skill of SWH is significantly higher than it is during weeks 3 and 4 in the WP-IO region. Based on a categorical forecast verification, the CFSv2-WW3 can predict the statistics of rare events, even at these extended lead times.