Quantile-based bias correction and uncertainty quantification of extreme event attribution statements

Christopher J Paciorek,Soyoung Jeon,Michael F Wehner

doi:10.1016/j.wace.2016.02.001

Abstract

Extreme event attribution characterizes how anthropogenic climate change may have influenced the probability and magnitude of selected individual extreme weather and climate events. Attribution statements often involve quantification of the fraction of attributable risk (FAR) or the risk ratio (RR) and associated confidence intervals. Many such analyses use climate model output to characterize extreme event behavior with and without anthropogenic influence. However, such climate models may have biases in their representation of extreme events. To account for discrepancies in the probabilities of extreme events between observational datasets and model datasets, we demonstrate an appropriate rescaling of the model output based on the quantiles of the datasets to estimate an adjusted risk ratio. Our methodology accounts for various components of uncertainty in estimation of the risk ratio. In particular, we present an approach to construct a one-sided confidence interval on the lower bound of the risk ratio when the estimated risk ratio is infinity. We demonstrate the methodology using the summer 2011 central US heatwave and output from the Community Earth System Model. In this example, we find that the lower bound of the risk ratio is relatively insensitive to the magnitude and probability of the actual event.

Highlights

The summer of 2011 was extremely hot in Texas and Oklahoma, producing a record of 30.26 °C for the average June–July–August (JJA) temperature (3.24 °C above the 1961–1990 mean) as measured in the CRU observational dataset (CRU TS 3.21, Harris et al, 2014)
We present an approach to extreme event attribution that addresses differences in the scales of variability between observations and model output using the methodology of quantilebased bias correction in the context of a formal statistical treatment of uncertainty
We develop a procedure for estimation and for quantifying uncertainty in the risk ratio, a measure of the anthropogenic effect on the change in the chances of an extreme event

Summary

Introduction

The summer of 2011 was extremely hot in Texas and Oklahoma, producing a record of 30.26 °C for the average June–July–August (JJA) temperature (3.24 °C above the 1961–1990 mean) as measured in the CRU observational dataset (CRU TS 3.21, Harris et al, 2014). As in most mid-latitude land regions, the probability of extreme summer heat in this region has increased due to human-induced climate change (Min et al, 2013). Extreme event attribution analyses attempt to characterize whether and how the probability of an extreme event has changed because of external forcing, usually anthropogenic, of the climate system. The fraction of attributable risk (FAR) or the risk ratio (RR) are commonly-used measures that quantify this potential human influence (Palmer, 1999; Allen, 2003; Stott et al, 2004; Jaeger et al, 2008; Pall et al, 2011; Wolski et al, 2014). We note that the commonly used term “risk ratio” is more precisely a “probability ratio” (Fischer and Knutti, 2015) but we will stick to the RR nomenclature in this study —in part because RR is the well-established terminology

Methods

Results

Conclusion