Multiscale Temporal Variability of Warm-Season Precipitation over North America: Statistical Analysis of Radar Measurements

Abstract

Directionally averaged time series of precipitation rates for eight warm seasons (1996?2003) over the continental United States derived from Next Generation Weather Radar (NEXRAD) measurements are analyzed using spectral decomposition methods. For the latitudinally averaged data, in addition to previously identified diurnal and semidiurnal cycles, the temporal spectra show cross-scale self-similarity and periodicity. This property is revealed by a power-law scaling with an exponent of ?4/3 for the frequency band higher than semidiurnal and ?3/4 for the 1?3-day band. For the longitudinally averaged series the scaling exponent for the frequency band higher than semidiurnal changes from ?4/3 to ?5/3 revealing anisotropic properties. The dominant periods and propagation speeds display temporal variability on about 1/2, 1, 4, 11, and 25 days. Composite patterns describing periods of <5 days display the eastward propagation characteristic of classical mesoscale convective organization. The lower-frequency (>5 days) patterns propagate westward suggesting the influence of large-scale waves, and both dominant periods and propagation speeds show marked interannual variability. The implied dependence between propagation and mean-flow for <5 days is consistent with the macrophysics of warm-season convective organization, and extends known dynamical mechanisms to a statistical framework.