Predictability of Precipitation over the Conterminous U.S. In the CMIP5 Multi-Model Ensembles

Characterizing precipitation variability in the face of future uncertainty is important for the delivery of sustainable ecosystem services. Mounted evidence exists exploring the possible scenarios of future changes in precipitation means and extremes, but rarely has literature attempted to quantify changes in precipitation stochasticity and cyclicity induced by anthropogenic pressures. Using the Colwell index of predictability (P) and the monthly mean precipitation data from the Coupled Model Intercomparison Project Phase 5 (CMIP5) multi-model ensembles, this study seeks to identify climate change hotspots in terms of changes in precipitation inter-annual and seasonal variability for the conterminous United States (U.S.). It is demonstrated that P of monthly mean precipitation is low in the East and the Intermountain West, and is generally high in the lowland regions of the West. Across the U.S. landscape, P is mostly determined by the measure of constancy (C), which inversely correlates with the fluctuations of inter-annual precipitation. On the other hand, contingency (M) – a measure of seasonality - plays a predominant role in determining P along the coast of the Pacific Northwest. Comparing the historic (1950-2005) to the future (2040-2095) periods, it is apparent that more prominent changes in P, C and M are present under Representative Concentration Pathways (RCP) 8.5 scenario than under RCP 4.5 scenario, reflecting the more significant precipitation distributional changes across time and space. Finally, there are regional hotspots in terms of how precipitation variability changes inter-annually and seasonally. The proper understanding of these changes in precipitation stochasticity and cyclicity has great socio-economic and environmental implications.