Predictability of Precipitation Over the Conterminous U.S. Based on the CMIP5 Multi-Model Ensemble.

Mingkai Jiang,Benjamin S Felzer,Dork Sahagian

doi:10.1038/srep29962

Abstract

Characterizing precipitation seasonality and variability in the face of future uncertainty is important for a well-informed climate change adaptation strategy. Using the Colwell index of predictability and monthly normalized precipitation data from the Coupled Model Intercomparison Project Phase 5 (CMIP5) multi-model ensembles, this study identifies spatial hotspots of changes in precipitation predictability in the United States under various climate scenarios. Over the historic period (1950–2005), the recurrent pattern of precipitation is highly predictable in the East and along the coastal Northwest, and is less so in the arid Southwest. Comparing the future (2040–2095) to the historic period, larger changes in precipitation predictability are observed under Representative Concentration Pathways (RCP) 8.5 than those under RCP 4.5. Finally, there are region-specific hotspots of future changes in precipitation predictability, and these hotspots often coincide with regions of little projected change in total precipitation, with exceptions along the wetter East and parts of the drier central West. Therefore, decision-makers are advised to not rely on future total precipitation as an indicator of water resources. Changes in precipitation predictability and the subsequent changes on seasonality and variability are equally, if not more, important factors to be included in future regional environmental assessment.

Highlights

Characterizing precipitation seasonality and variability in the face of future uncertainty is important for a well-informed climate change adaptation strategy
Based on the Spatially Downscaled but not Bias Corrected (SDnoBC) datasets of the Coupled Model Intercomparison Project Phase 5 (CMIP5) multimodel ensemble, this study shows current Earth system models are not yet able to accurately estimate precipitation variability, seasonality, and predictability (Fig. 2)
Analysis of the predictability of CMIP5 ensemble results with regard to precipitation indicates that precipitation is more predictable in the East and along the Pacific Northwest coast, and is generally less so in the arid Southwest

Summary

Introduction

Characterizing precipitation seasonality and variability in the face of future uncertainty is important for a well-informed climate change adaptation strategy. Current efforts assessing climate change-induced precipitation redistribution focus on long-term temporal trends and seasonal patterns Such predictions are important for well-informed adaptation and mitigation decision-making, yet there is a need to investigate whether the perturbed precipitation patterns will become more or less predictable (in terms of how seasonal precipitation pattern fluctuate inter-annually) for natural adaptation to take place. This has led to a knowledge gap in terms of an ecologically and/or hydrologically meaningful characterization of changes of precipitation variability. The term “predictability” used in this study differs from traditional understanding of predictability in that it does not reflect the predictive power of modeled precipitation based on understanding of the underlying processes and mechanisms, but rather, it is a description of the power of past precipitation attributes in both seasonality (contingency) and inter-annual variability within a specific time interval (constancy throughout 1950–2005) of observed precipitation itself to predict future precipitation

Methods

Results

Discussion

Conclusion