Rainfall variability and predictability issues for North America

A multi-millennial simulation with a coupled global climatic model has been used to investigate extreme rainfall events, mainly droughts, over North America. A rainfall index, based on the US Dust Bowl region, was used to generate a time series from which the extreme events could be identified. A very wide range of drought and pluvial multiyear sequences was obtained, all attributable to internal climatic variability. This time series reproduced the basic characteristics of the corresponding observed time series. Composites of years with negative rainfall anomalies over North America from the simulation replicated the observed rainfall composite for the Dust Bowl era, both in spatial character and intensity. Examination of individual years of a simulated composite revealed not only a wide range of rainfall anomaly patterns, dominated by drought conditions, but also ENSO distributions that included El Nino events as well as the expected La Nina events. Composites for pluvial conditions over North America were associated with composited El Nino events, as expected. Correlation of the simulated Dust Bowl rainfall with global surface temperatures identified a principal connection with the ENSO region. No systematic relationship was obtained in the simulation between the Atlantic multidecadal oscillation and Dust Bowl region rainfall, with the simulated oscillation having a much more variable periodicity than that found in the limited observations. However, a marked connection was found for SST anomalies adjacent to the northeast coast of North America, but this appears to be forced by ENSO events. A scatter diagram of NINO3.4 SST anomalies with the Dust Bowl region rainfall anomalies, for observations and the simulation, revealed inconsistencies between the occurrence of an ENSO event and the “expected” rainfall anomaly. This, and other analysis, resulted in the conclusion that annual or longer term rainfall predictions over North America, with any systematic confidence, are unlikely because of stochastic influences inherent in the climatic system associated with internal climatic variability.