The Extraction And Assimilation Of Ekman Current

The Ekman theory is a fundamental theory in marine science,and the Ekman layer model has a long history in the study of physical oceanography.The vertical eddy viscosity coefficient(VEVC)is a significant parameter in Ekman layer model,which has an important effect on the structure of Ekman spiral and Ekman transport.Temporal distribution of the VEVC in an Ekman layer model is estimated using the adjoint method.Seven groups of ideal experiments are carried out to investigate the influences of seven factors on inversion results,including the optimization algorithm,initial guess,distribution of prescribed VEVCs,wind condition,boundary layer depth(BLD),observation level and data noise.The main conclusions are that(1)the gradient descent(GD)algorithm is better than the quasi-Newton conjugate gradient(CG)and limited-memory Broyden-Fletcher-Goldfarb-Shanno(L-BFGS)algorithms in the optimization of the VEVC;(2)the inversion results are sensitive to the initial guesses;in order to obtain accurate inversion results,the initial guess must be as reasonable as possible;(3)the adjoint method is a capable method to estimate different kinds of temporal distribution of the VEVC in Ekman layer model;(4)the model is applicable to different types of wind conditions,including sinusoidally-,constantly-and linearly-varying wind speeds;(5)the inversion result with thick BLD is slightly better than that with thin BLD;(6)the inversion results are much more sensitive to the observations in the upper layers than those in the lower layers;and(7)the inversion results are still acceptable when data noise exists,indicating the method has the potential to assimilating real ocean observations.The Ekman currents are extracted from the ADCP at Bermuda Testbed Mooring,and then VEVC is also inverted.The main conclusions are that(1)GD algorithm is still the best among the three ones in practice experiments;(2)the strategy which assume that VEVC is time-varying exceed the one which assumes that it is constant or depth-varying;(3)VEVC in the study area is about 0.01m2s-1.This study extends our understanding of Ekman layer model and provides an effective way to determine the temporal variations of the VEVC.