A 45-year high-resolution unstructured wave hindcast for the Australian coast: Validation and climatological insights

Abstract

This study presents a 45-year (1979–2023) high-resolution wave hindcast for the entire Australian coast, conducted using WAVEWATCH III on an unstructured mesh with a resolution ranging from 1 to 15 km. Incorporating the observation-based source term package (i.e., ST6) and the subgrid-scale reef parameterization, the model performs well in simulating widely-used bulk wave parameters. The simulated significant wave height $H_s$ achieves a correlation greater than 0.96 and exhibits only a centimeter-scale bias when compared to both altimeter and buoy observations. The use of BARRAv2 winds provides a clear advantage under extreme conditions, with negligible underestimation below the 99.9th percentile of $H_s$. The model performs particularly well in the Great Barrier Reef (GBR) region, where the bathymetry is complicated and could not be resolved well by typical km-scale wave models. Unlike previously published hindcasts that excessively overestimated wave energy in the GBR, our results show only a marginal $H_s$ bias of approximately −0.05 m, against the shallow water buoys in this specific region. This better performance can be attributed to the subgrid-scale reef parameterization adopted. Long-term validation results demonstrate the robustness of the model framework, with the hindcast showing good temporal homogeneity and reliability. Building upon these results, this study reveals statistically significant increasing trends in wave heights along most of the Australian coast, with particularly pronounced upward trends in extreme wave heights (90th, 95th, and 99th percentiles) in the GBR. Furthermore, the GBR $H_s$ trends based on our hindcast are much stronger than those estimated from previous hindcasts, and this new finding requires further investigation.