Abstract
Epistemic uncertainties of wave power estimation based on spectral wave model were assessed utilizing the community wave model WW3 of NOAA. A four tiered nested model was constructed covering the Pacific Ocean, the sea around Japan, the North-Eastern Japan and Kamaishi regional model at 100km, 10km, 1km and 100m resolutions respectively. The sensitivity of the model output to four wind products (NCEP-GFS, NCEP-CFSR, GSM/JMA, ERA-interim/ECWMF) was tested; the wave model outputs as well as wind products were validated against observations of NDBC buoys, JKEO/JAMSTEC buoy, TAO array and NOWPHAS buoy along the north-eastern Japan. The differences of the estimations using four wind products were relatively small whereas model errors were spatially inhomogeneous. The dependence on grid-resolution was relatively small at depth 100m or so. On the other hand, the modeled significant wave period had a large bias from the observation because of the inconsistency in the spectral moment used to estimate the significant wave period. The energy period (T-1,0) was 5% larger than the period from WW3 (T-0.5,0) and 20% larger than the period from NDBC (T0,2). The difference of wave period, dependent on the spectral shape and location, are often overlooked despite their significant impact on the resource estimate.