Daily precipitation and tropical moisture exports across the eastern United States: an application of archetypal analysis to identify spatiotemporal structure

Abstract:

This study examines the spatiotemporal variability of two sets of daily precipitation from the ERA-Interim reanalysis across the eastern United States between 1979-2013: 1) total precipitation, and 2) precipitation originating from tropical moisture exports (TMEs), which have been linked to extremes of midlatitude precipitation. We introduce archetypal analysis (AA) as a new method to decompose and characterize structures within the spatiotemporal climate data. AA is uniquely suited to identify extremal patterns and is a complementary method to Empirical Orthogonal Function (EOF) analysis. We provide a brief comparison between AA and EOF analysis and then examine the spatiotemporal variability, circulation anomalies, and sea surface temperature teleconnections associated with the archetypes of the two precipitation variables. Markovian structure, seasonal variability, and inter-annual trends in archetype occurrence are explored using nonparametric generalized linear models (GLMs). Results show that the modes of precipitation variability and their associated teleconnections are very similar between total and TME precipitation, suggesting that TMEs can help explain prevailing modes of total precipitation variability. Both total and TME precipitation shift longitudinally conditional on the phase of the Pacific Decadal Oscillation (PDO) and sea surface temperatures in the North Atlantic, and they are inhibited during strong, negative PDO and positive Atlantic Multidecadal Oscillation (AMO) regimes. The GLM analysis reveals distinct seasonal cycles and decadal trends in archetypes likely associated with the strength and position of the North Atlantic Subtropical High (NASH).