The Evaluation of Spectral Wave Models in Hawai'I Relative to Observations

Abstract

Waves impact the heavily populated coastal regions in Hawaiʻi. Waves also affect commerce, maritime operations, renewable energy development, and recreational activities. In order to mitigate damages to offshore activities, coastal infrastructure, and life, reliable wave forecasts are needed. The Pacific Islands Ocean Observing System (PacIOOS) operates a nested spectral wave model that predicts waves in nearshore and offshore waters of Hawaiʻi. This thesis analyzes the archived PacIOOS wave forecast from 2011 to 2021 using observations from 13 moored wave buoys and space-borne satellite altimeters from 7 different satellite missions to gain insight into the model errors. Using the observation datasets as reference, several error metrics are used to evaluate the forecast uncertainties, as well as the uncertainties that evolve with the increasing lead time. The wave forecast performs best for wave heights around 2 meters at the start of the forecast period. At the start of the forecast period during days 1 + 2 across the Hawaiʻi regional model domain, NW wave events (281° - 348°) are overestimated by 6.0%, N-NE wave events (348° - 56°) are underestimated by -1.7%, SW wave events (191° - 236°) are underestimated by -2.7%, and S-SE wave events (146° - 191°) are underestimated by - 1.5%. Most large wave events over 3 meters are underestimated at the start of the forecast period, and this occurs for most wave directions with the exception of some outlier wave events. The wave forecast skill decreases with increasing lead time, and the skill decreases at faster rates by days 3-4 of the forecast and occurs for all incoming wave directions. The approaches developed here can be used to evaluate other wave models worldwide and the results establish a baseline for future model improvements.