Global characteristics of pluvial and dry multi-year episodes, with emphasis on megadroughts



A multi-millenial simulation with a coupled atmospheric–oceanic climatic model has been used to investigate the occurrence rate and characteristics of multi-year dry and pluvial episodes. Such episodes are defined as having the same sign rainfall anomalies for the whole of their durations. Episodes of 5-, 8- and 10-year durations are shown to occur all over the globe. The occurrence rates are approximately in agreement with probabilistic expectations, although there are spatial variations associated with El Nino Southern Oscillation (ENSO) influences. The impacts of the dry and pluvial episodes are considerable, with 10-year duration dry episodes resulting in a loss of at least one year's rainfall over the oceans and much larger losses over arid regions. Similar but positive outcomes are associated with pluvial episodes. Accumulated rainfall anomalies over various multi-decadal periods indicate that regions with persistent features can occur, implying longer term impacts than might have been expected from a succession of dry and pluvial episodes. Time series of these episodes for individual model gridboxes highlight a disparity between the occurrence rates of dry and pluvial episodes and the lack of synchronicity of these episodes at the various gridboxes. These time series also clearly illustrate the rapid fall-off in the occurrence rates of the episodes as their durations are increased. Despite claims in the literature that ‘megadroughts’ are typically associated with extended ENSO events, it is shown that for a very extended La Nina event the rainfall anomalies in a number of ENSO-sensitive regions experienced interannual variability rather than a persistent anomaly. The role of a number of large-scale oscillations in generating rainfall anomalies is examined, but it is concluded that stochastic processes are needed to account for the extended duration of dry and pluvial episodes, in addition to any forcing associated with such oscillations. This adds a random aspect to the determination of whether any given, say, ENSO event will become a dry or pluvial episode. Copyright © 2010 Royal Meteorological Society