Changes in precipitation variability across time scales in multiple global climate model large ensembles

Abstract

Anthropogenic changes in the variability of precipitation stand to impact both natural and human systems in profound ways. Precipitation variability encompasses not only extremes like droughts and floods, but also the spectrum of precipitation which populates the times between these extremes. Understanding the changes in precipitation variability alongside changes in mean and extreme precipitation is essential in unraveling the hydrological cycle's response to warming. We use a suite of state-of-the-art climate models, with each model consisting of a single-model initial-condition large ensemble (SMILE), yielding at least 15 individual realizations of equally likely evolutions of future climate state for each climate model. The SMILE framework allows for increased precision in estimating the evolving distribution of precipitation, allowing for forced changes in precipitation variability to be compared across climate models. We show that the scaling rates of precipitation variability, the relation between the rise in global temperature and changes in precipitation variability, are markedly robust across timescales from interannual to decadal. Over mid- and high latitudes, it is very likely that precipitation is increasing across the entire spectrum from means to extremes, as is precipitation variability across all timescales, and seasonally these changes can be amplified. Model or structural uncertainty is a prevailing uncertainty especially over the Tropics and Subtropics. We uncover that model-based estimates of historical interannual precipitation variability are sensitive to the number of ensemble members used, with 'small' initial-condition ensembles (of less than 30 members) systematically underestimating precipitation variability, highlighting the utility of the SMILE framework for the representation of the full precipitation distribution.