Analysing observation based source terms: validations and spatial analysis of high-resolution wave hindcast in the Arabian Sea, emphasizing low wind conditions

Abstract

The performance of third-generation spectral wave models under low-wind conditions remains insufficiently evaluated despite the prevalence of such conditions across large portions of the global ocean. This study evaluates the WAVEWATCH III (WW3) model with the observation-based ST6 source-term parameterization in the Arabian Sea, where the low-wind pre-monsoon and post-monsoon seasons together span approximately eight months of the annual cycle. ERA-5 and CFSR/CFSv2 reanalysis wind datasets are first validated against the subset of globally calibrated satellite altimeter dataset for the year 2016 and used to force two independent WW3 simulations. Both wind products overestimate wind speed at very low winds and underestimate at low to moderate winds, while at high winds ERA-5 underestimates and CFSR/CFSv2 overestimates. The time series analysis further reveals that, during the low wind non-monsoon seasons, the CFSR/CFSv2-forced simulation produces a relatively higher positive Hs bias than the ERA-5 forced simulation, even though both wind fields carry a negative bias during the low wind periods. To examine this further with a statistically robust sample, a 5-year CFSR/CFSv2-forced hindcast (2014–2018) is conducted, which shows that the Hs overestimation is strongest during the pre-monsoon, with an Hs bias of + 0.10 m coexisting with a CFSR/CFSv2 wind bias of − 0.08 m s⁻¹ at mean winds of 4–7 m s⁻¹, and weaker during the post-monsoon, with an Hs bias of + 0.04 m coexisting with a CFSR/CFSv2 wind bias of − 0.12 m s⁻¹ at mean winds of 5–9 m s⁻¹. This counter-intuitive combination of positive Hs bias with negative wind bias during the low wind seasons is diagnostically significant, as a negative wind bias would normally be expected to produce a negative Hs bias, and cannot be explained by wind forcing errors alone. Spatially, this positive Hs bias is concentrated in the northern Arabian Sea, where Southern Ocean swells, Shamal swells, and Makran swells converge under weak local winds and where the CFSR/CFSv2 wind bias is near-zero to negative. The combined spatial and seasonal pattern is inconsistent with a wind-forcing induced error in Hs. The Hs overestimation is plausibly linked to two established physical low-wind thresholds that are absent from the ST6 physics, namely the viscous dissipation threshold for wave generation and the wave breaking onset threshold at low wind speeds. These findings establish the baseline for the implementation of low-wind thresholds in the ST6 source-term package.