Multiple-Scale Spatio–Temporal Variability of Precipitation over the Coterminous United States

Abstract

The spatio?temporal variability of precipitation over the United States using a 30-yr, gridded hourly precipitation dataset is studied. Orthogonal wavelet transform is applied to the time series at each grid box to capture the temporal scales of fluctuation at 17 different timescales ranging from 2 h to 15 yr. Rotated principal component analysis is then applied to the transformed series to identify spatial coherence of the temporal scales of fluctuations. The results indicate that the energy of the fluctuations shows an approximate power-law relationship with respect to scale in most regions. The spatial organization of the temporal variability shows coherence at distinct scales identified as the subdiurnal (2?16 h), synoptic (16 h?22 days), seasonal (42 days?1 yr), and climatic mode (15 yr). The synoptic scale explains the largest spatial variance of the fluctuations in precipitation and is spatially coherent; the subdiurnal mode is spatially less coherent. The seasonal mode is dominant over the Pacific Northwest, whereas the climatic mode has large amplitude only over California. When examining the winter and summer seasons separately, it is found that the winter precipitation fluctuation is more associated with synoptic scale; the summer fluctuation is associated with shorter timescales or the subdiurnal scale. Studies of extreme summer drought and flood events over the Midwest indicate that anomalously wet or dry years are manifestations of persistent anomalous wet or dry conditions across all temporal scales, with the maximum contribution for the wet events being affected by the synoptic-scale activities.