Impact of sea-state dependent Langmuir turbulence on the ocean response to tropical cyclones

Tropical cyclones are driven by the air-sea heat flux, which is reduced when the ocean surface cools due to deepening of the mixed layer and upwelling. Wave-driven Langmuir turbulence (LT, due to interaction between Eulerian current vorticity and the Stokes drift) may significantly modify these processes. This study investigates the impact of LT on the ocean response to tropical cyclones using the Princeton Ocean Model (POM), which is coupled to the WAVEWATCH-III wind-wave model. This version of POM uses the KPP vertical mixing scheme. We examine three KPP configurations. The first, KPP-nw, is tuned to conditions of shear turbulence only. The second, KPP-df, is tuned to typical ocean conditions (with typical Langmuir turbulence), but does not include any sea-state dependence. The third, KPP-LT, includes explicit sea-state dependent Langmuir turbulence effects by parameterizing the vertical turbulent momentum flux using the Lagrangian current (Eulerian current plus Stokes drift) with an enhanced vertical mixing coefficient based on the turbulent Langmuir number. We find that both KPP-df and KPP-LT enhance sea surface cooling due to vertical mixing at all locations compared to KPP-nw because Langmuir turbulence enhances mixed layer deepening. For quasi-stationary storms, the additional cooling due to upwelling (caused by storm-induced horizontal divergence of near surface current) is reduced with KPP-LT compared to KPP-df. This is because KPP-LT reduces the near surface current and its horizontal divergence by increasing the vertical diffusion of momentum. As the storm translation speed increases, the sea-state dependent LT effects become more complex as the upwelling becomes less important but the reduced near surface currents significantly modify the horizontal advection within the cold-wake. The difference between the predicted surface cooling using KPP-LT and KPP-df has potential implications for tropical cyclone prediction models.