Improving intensity simulation and forecast of tropical cyclones has always been a challenge, although in recent years the track forecasts have been remarkably improved. In this study, we explore the sensitivity of typhoon simulation to three physical processes using a fully coupled atmosphere‐ocean‐wave model. Two storms, a strong and a weak one, have been chosen. The effects of wave breaking induced sea spray, ocean vertical mixing associated with nonbreaking surface waves, and sea surface cooling due to intense rainfall are assessed by means of a set of numerical experiments. The results show and confirm that sea spray leads to an increase of typhoon intensity by enhancing the air‐sea heat flux, while nonbreaking wave‐induced vertical mixing and rainfall lead to a decrease. Each process can be relevant, depending on wind and wave conditions. These can vary dramatically when typhoons interact with not sufficiently well‐defined coastal areas, typically an archipelago. Compared with the control runs, when all the three physical processes are considered, the (absolute) difference between the modeled sea level pressure and best track data is reduced from 26.05 to 0.70 hPa for typhoon Haiyan, and from −9.42 to −8.67 hPa for typhoon Jebi. We have found a steady overestimate of the dimensions of the typhoons. We have verified an extreme sensitivity to the initial conditions, especially when small differences in the typhoon track may imply different relevance of the physical processes, like the ones we have considered, governing the evolution of the storm.

No abstract

The deviation of the wind stress vector from the wind direction at the air‐sea interface under low wind conditions was investigated based on direct eddy covariance flux measurements taken at a coastal tower in the northern South China Sea. The wind stress deviates significantly from the mean wind direction under low wind conditions, with the deviation angle sometimes exceeding 90°, indicating upward momentum transfer from the ocean to the atmosphere. Negative downwind drag coefficient values begin to occur at a wind speed of approximately 4 m/s. Our results show that ocean swells and nonstationary airflow play critical roles in wind stress. Prominent peaks at the dominant swell frequency in the vertical velocity spectra are observed at a height of 17 m over the mean sea surface, implying that swell‐induced perturbations can reach a height of at least 17 m, and the wave boundary layer can extend more than 10 m above the sea surface. The results of our analysis indicate that at the observation height, the influence of nonstationarity in the wind field is more significant than that of swell‐induced motions on the deviation of wind stress. After the removal of nonstationary motions, the deviation angles of the wind stress from the wind direction are generally reduced and vary substantially at low wind speeds.