Past and future climate change in the context of memorable seasonal extremes

A new paper by ICARUS researchers and international colleagues has been published in Climate Risk Management. The paper is Open Access.

 http://www.sciencedirect.com/science/article/pii/S221209631600005X

Past and future climate change in the context of memorable seasonal extremes

T. Matthews, D. Mullan, R.L. Wilby, C. Broderick, C. Murphy

Abstract

It is thought that direct personal experience of extreme weather events could result in greater public engagement and policy response to climate change. Based on this premise, we present a set of future climate scenarios for Ireland communicated in the context of recent, observed extremes. Specifically, we examine the changing likelihood of extreme seasonal conditions in the long-term observational record, and explore how frequently such extremes might occur in a changed Irish climate according to the latest model projections. Over the period (1900–2014) records suggest a greater than 50-fold increase in the likelihood of the warmest recorded summer (1995), whilst the likelihood of the wettest winter (1994/95) and driest summer (1995) has respectively doubled since 1850. The most severe end-of-century climate model projections suggest that summers as cool as 1995 may only occur once every ~7 years, whilst winters as wet as 1994/95 and summers as dry as 1995 may increase by factors of~8 and  ~10 respectively. Contrary to previous research, we find no evidence for increased wintertime storminess as the Irish climate warms, but caution that this conclusion may be an artefact of the metric employed. It is hoped that framing future climate scenarios in the context of extremes from living memory will help communicate the scale of the challenge climate change presents, and in so doing bridge the gap between climate scientists and wider society.

Fig. 8. Evolution of z-scores in the historical and RCP experiment calculated for centred, 30-year sliding windows. The shaded region of the CMIP5 ensemble spans 5–95th percentiles, whilst the solid lines provide the median. The discontinuity between the historical and RCP 8.5 medians is because only a subset of historical model runs continues to RCP 8.5. Note that the observed series are also displayed in each panel and the different scaling on the respective y-axes.