Inferring directional wave spectra via motion sensing on ocean-going autonomous surface vessels

This paper investigates the estimation of directional wave spectra from autonomous surface vessels by leveraging their dynamic response to ambient wave forcing. The approach combines
pre-computed hydrodynamic transfer functions with a two-stage processing chain: (i) estimation of the directional spreading function via the Extended Maximum Entropy Principle using heave acceleration, roll, and pitch cross-spectra; and (ii) transformation from the encounter domain to intrinsic wave frequency, explicitly accounting for Doppler effects due to forward speed. The methodology is applied to a July 2025 deployment of the wind-propelled Oshen C-star in the English Channel, and validated against a nearby scientific waverider buoy (E1). Results show strong agreement for significant wave height (RMSE ≈ 0.16 m, 𝑅2 ≈ 0.94) and improved estimates of the mean zero up-crossing period after Doppler correction (RMSE ≈ 0.46 s, 𝑅2 ≈ 0.80), while peak period is largely unaffected at the observed low vessel speeds. Directional estimates are more sensitive, with accuracy improving under segments exhibiting limited heading variability. The study demonstrates the feasibility of ASV-based sea state monitoring to complement traditional networks.