FLUX DECOMPOSITION INTO COHERENT STRUCTURES

This study examines the intermittency of the momentum flux near the Surface and the relation of such intermittency to coherent Structures. Toward this goal, variances and covariances are decomposed into coherent structures and less coherent activity. The sampled structures are identified using the Haar transform and then decomposcd into eigenvectors of the lagged covariance matrix. The methodology is applied to thc momentum flux for a relatively stationary 50-h period of strong winds measured from a 45 m tower in the Lammefjord Experiment. Events of sinking motion with strong horizontal momentum account for the majority of the flux. Such sweeping motions arrive as gust microfronts. The large momentum flux is associated with strong coherent fluctuations of the longitudinal wind component and high correlation with relatively modest fluctuations of vertical motion. In the heated case (HAPEX), a phase lag between the vertical and horizontal velocity fluctuations leads to less efficient momentum transport by the main coherent structures. The event nature of the flux is used to formulate an expression for the flux error due to sampling problems. Estimation of the momentum flux requires a significantly longer record than for the heat flux. Modulation of the flux by mesoscale variations also affects the sampling strategy.