Eddy-covariance Method And Bulk Parameterization On The Air-sea Fluxes

The air–sea interface fluxes are important parameters for the descriptions of a large numberof dynamic processes, such as: the air-sea interactions, global climatechange, large-scaleatmospheric and ocean circulation, the development of hurricane, the generation of waves, thedevelopment of the mixed layer and the seasonal thermocline. Therefore, a simple, fast andaccurate calculation of the air-sea interface flux has always been an important issue.This study is on the basis of the datasets about the atmospheric boundary layer turbulenceand the waves in the air-sea interface, from an air-sea interface flux buoy and anAWAC（Acoustic Wave and Current Profiler）. On the calculation and the bulk parameterization ofthe air-sea interface flux, we perform the following researches: the analysis of the atmosphericboundary layer turbulence, the contamination of the non-turbulent motions on the calculation ofair-sea fluxes by eddy covariance method, the impact of the waves on the momentum flux （windstress） and the parameterization of the drag coefficient.The Hilbert-Huang transform （HHT）, a powerful tool for the time-frequency analysis ofnon-linear and non-stationary data, is applied to analyzing the turbulent time series obtainedwithin the atmospheric boundary layer over the ocean. The time series are decomposed into asum of modes with different characteristic frequencies, corresponding to a sum of turbulentvortexes with different time scales. After the energy spectrum analysis, these modes areassociated with instrument noise, inertial range, and energy containing eddies.A new method based on the HHT is then introduced to reduce the influence of non-turbulentmotions on the eddy-covariance based fluxes by removing the non-turbulent modes from thetime series. By calculating the contributions of different modes on the eddy-covariance basedflux, the scale dependence of the flux is established. A gap mode is identified to distinguishbetween turbulent modes and non-turbulent modes by examining the scale dependence of theflux. After removing the non-turbulent modes from the time series, the contamination of thenon-turbulent motions are considered to be reduced or removed. The effectiveness of this newmethod is confirmed by compareing with three conventional methods （block average, moving-window average, and multi-resolution decomposition） on the basis of the datasets fromthe buoy. The results show that the new method based on the HHT is effective in removing thenon-turbulent motions and it is much better than the other three methods.The next problem focused is the impact of the waves on the momentum flux （wind stress）and the parameterization of the drag coefficient. By processing the data from the buoy and theAWAC, we obtain a waves dataset synchronized with the eddy-covariance based fluxes. Basedon the above dataset and the other five ones, six parameterization schemes of roughness or dragcoefficient are evaluated. They present great consistency with measurement when frictionvelocityu*<0.5m/s (approximately corresponding to10m wind speed U₁₀<12m/s) and largedeviation from measurement whenu*≥0.5m/s (approximately corresponding to U₁₀≥12m/s).In order to improve this deviation, we construct a two-dimensional simple harmonic modeland roughly derive the impact of wave on the momentum flux. The wave steepness is consideredto be an important parameter for this impact. Then, based on the similarity theory, the Charnockrelationship and the Toba3/2power law, we derive a new parameterization of the dragcoefficient, which is useful in the condition of0<U₁₀<20m/s. It considers both the wavesteepness and the wind-sea Reynolds number. After testing its performance, it shows that thedeviation between the measurement and the estimation in U₁₀≥12m/s has been significantlyimproved. And the new parameterization has much smaller standard error compared with theother six ones. However, the new parameterization still lacks enough tests. Meanwhile, due tothe lack of data in very high winds, we have not considered that the drag coefficient levels off ordecreases with increasing wind speed in very high winds.