Data-driven technique for decomposing the relative effects of waves and turbulence

The flow near the ocean surface modulates key climate processes ranging from mixed-layer dynamics to air-sea fluxes of heat, gas, and momentum. These quantities are commonly parameterized by the Reynolds stress or the friction velocity and can be inferred from point velocity measurements. However, these observations are frequently contaminated by surface waves which precludes bandpass filtering from accurately isolating the turbulence- and wave-induced velocity components. Decomposing wave and turbulence signals from ocean flow data is an ongoing challenge. Most previously developed methods are limited to ideal wave conditions and assume that waves and turbulence do not interact. We demonstrate a data-driven approach using dynamic mode decomposition (DMD) to recover and filter out dominant wave dynamics from turbulent flow data without any prior system knowledge or assumptions about the wave-turbulence interaction. We consider flow scenarios with different ratios of wave and shear strain to quantify the wave motion influence on turbulence. DMD shows promising results when applied to single-point measurements from field and laboratory data.