A hybrid framework for efficient real-time ocean noise simulation on dynamic hydrophone and vector sensor arrays

Accurate and computationally efficient real-time simulation of ocean ambient noise is essential for the design and evaluation of modern underwater acoustic systems, particularly dynamic arrays used in adaptive sampling. This paper presents a hybrid ocean noise modeling architecture that combines high-fidelity wavenumber integral modeling with efficient Bessel-based covariance computation to generate calibrated element-level time series. Slowly varying vertical noise directionality is obtained from the Kuperman-Ingenito surface-generated noise model or from at-sea measurements and is updated infrequently. Hydrophone and vector sensor covariance matrices are then evaluated within a unified pressure-gradient-based framework for directional plane-wave fields, three-dimensional isotropic noise, azimuthally symmetric isotropic noise, and Kuperman-Ingenito surface-generated ocean noise, with the latter represented in both normal mode and wavenumber integral form. Element-level time series are synthesized using Cholesky decomposition, frequency- and time-dependent scaling, and inverse fast Fourier transforms. By separating computationally intensive directionality estimation from efficient covariance updates for evolving array geometry and orientation, the framework enables high-fidelity real-time ambient noise simulation for dynamic arrays.