Analyzing the Variability and Teleconnection of Seasonal Precipitation over Western U.S. Using Empirical Orthogonal Function and Nonlinear Mode Decomposition

Abstract:

Seasonal variations, decadal trends and spatial patterns of precipitation are crucial to water managements in western United States. We use the Empirical Orthogonal Function-Principal Component (EOF-PC) and Nonlinear Mode Decomposition (NMD) methods to analyze gridded monthly precipitation observation data in western US from 1971 to 2013.To differentiate trends, noises and physically-meaningful periodic patterns from the time series, we choose the wavelet-based NMD method. The results of our analysis show the leading EOFs and PC time series are capable of reconstructing the dynamic field of regional precipitation in high accuracy. The NMD method detects several amplitude-modulated periodic curves with complex waveforms. We found the first PC’s varying amplitude is correlated to Pacific Decadal Oscillation index (R=0.41), and the fourth PC's varying amplitude is correlated to Nino3.4 index (R=0.57). These results suggest that the complex spatiotemporal variability of precipitation can be quantified with the dominant spatial patterns, periodicities, trends, and noises. And the EOF-NMD method can be used as a data-mining tool to investigate physically-meaningful information and teleconnections from a dataset of a dynamic field.