Changes in Large Precipitation Events Under Global Warming

Abstract:

Model-based studies have suggested that changes may occur in hydrologic variables associated with increased probability of high precipitation occurrences, such as the 95th percentile of daily precipitation intensity. It has been common to discuss changes in extreme events in terms of consequences of the shift of the mean of the distribution, or an increase in variance. Here we present a prototype for the behavior of changes in the distribution of precipitation event size, i.e., of the precipitation integrated from onset to termination of an event. It has been noted in observations that event size distributions exhibit a scale-free, approximately power-law range with a cutoff at large event sizes (Peters et al, 2010). A simple stochastic model for an atmospheric column moisture budget shows how this behavior arises and how the cutoff is controlled by the physical parameters of the system. A simple set of conditions implies that the frequency of the very largest events may be expected to exhibit disproportionate sensitivity to changes. Any increase in the variance of the moisture convergence, as is projected to occur in the warmer climate, is predicted to yield changes in the very largest event-size portion of the distribution. Precipitation event size distributions computed from high time-resolution output from the Community Earth System Model appear to fit this prototype well in current climate, consistent with the form of observed event size distributions. In an ensemble of 15 simulations under Representative Concentration Pathway 8.5 for anthropogenic forcing increases, the very largest event portion of the distribution shows marked sensitivity, consistent with the simple prototype, with increases over most of the globe and strong regional increases. The theory suggests a set of conditions to evaluate for other climate quantities under which one should expect such sensitivity of probability of the very largest events.