IDENTIFYING PERIODIC COMPONENTS IN ATMOSPHERIC DATA USING A FAMILY OFMINIMUM-VARIANCE SPECTRAL ESTIMATORS

This work describes the application of a recently developed signal pro cessing technique for identifying periodic components in the presence of unknown colored noise. Specifically, the application of this techni que to the identification of strongly periodic components in meteorolo gical time series is examined, The technique is based on the unique co nvergence properties of the family of minimum variance (MV) spectral e stimators, The MV convergence methodology and computational procedures are described and are illustrated with a theoretical example. The uti lity of this method to atmospheric signals is demonstrated with a 26-y ear ( 1964-1989) time series of 70-mb wind components at Truk Island i n the equatorial Pacific. The MV method clearly shows that although eq uatorial disturbances with periods of 3-5 days have a strong signal, t hey do not show a strong periodic component. As expected, MV convergen ce illustrates that the 70-mb zonal wind series at this location has a significant periodic component at the frequency of the annual cycle. In addition, the MV technique provides evidence for a strong periodic component at the frequency of the semiannual cycle and at a frequency within the commonly accepted range of the QBO. Although the QBO is cle arly nota strictly periodic phenomenon (since its period is known to v ary), the available data suggest that it can be modeled as a periodic component of the zonal wind. This is substantiated by a simple three-s inusoid plus autoregressive order 1 noise model of the 70-mb Truk zona l wind. This parsimonious model provides a very good fit to the observ ed data.