Adaptive Decomposition: Application to Turbulence Wavelets in Geophysics

摘要

Nine hours of 45 meter tower anemometer measurements are analyzed to demonstratean adaptive method for decomposing a time series into orthogonal modes of variation. In conventional partitioning (or filtering) the cutoff scales are specified a priori to be constant throughout the record. Applying a constant cutoff scale is less effective if two different physical modes vary on overlapping scales, since the statistical partitioning is then physically ambiguous. For the turbulence data analyzed in this study, motions leading to a majority of the momentum flux intermittently occur on small scales which otherwise lead to little flux. To better separate the transporting motions from the more random motions, the cutoff scale separating these two modes is allowed to vary with record position. Local extremes in the spatial distribution of momentum flux determine the partitioning between the two small scale modes. This leads to an adaptive cutoff scale, which better isolates the transport mode responsible for a majority of the momentum flux. The spatial variation of this cutoff scale allows the time series to be decomposed into modes which are physically more pure as is verified in terms of traditional statistics for each mode. (AN).