Predictability of the European heat and cold waves

Christophe Lavaysse,Peter Salamon,Gustavo Naumann,Jürgen Vogt,Lorenzo Alfieri

doi:10.1007/s00382-018-4273-5

Abstract

Heat and cold waves may have considerable human and economic impacts in Europe. Recent events, like the heat waves observed in France in 2003 and Russia in 2010, illustrated the major consequences to be expected. Reliable Early Warning Systems for extreme temperatures would, therefore, be of high value for decision makers. However, they require a clear definition and robust forecasts of these events. This study analyzes the predictability of heat and cold waves over Europe, defined as at least three consecutive days of {text {T}}_{text {min}} and {text {T}}_{text {max}} above the quantile Q90 (under Q10), using the extended ensemble system of ECMWF. The results show significant predictability for events within a 2-week lead time, but with a strong decrease of the predictability during the first week of forecasts (from 80 to 40% of observed events correctly forecasted). The scores show a higher predictive skill for the cold waves (in winter) than for the heat waves (in summer). The uncertainties and the sensitivities of the predictability are discussed on the basis of tests conducted with different spatial and temporal resolutions. Results demonstrate the negligible effect of the temporal resolution (very few errors due to bad timing of the forecasts), and a better predictability of large-scale events. The onset and the end of the waves are slightly less predictable with an average of about 35% (30%) of observed heat (cold) waves onsets or ends correctly forecasted with a 5-day lead time. Finally, the forecasted intensities show a correlation of about 0.65 with those observed, revealing the challenge to predict this important characteristic.

Highlights

Heat or cold waves (HWs or CWs hereafter) are linked to increased risks of mortality, morbidity, and different cardiovascular and respiratory diseases (Anderson and Bell 2009; Gasparrini and Armstrong 2011; Kovats and Hajat 2008; Huynen et al 2001; Thakur et al 1987)
This study aims at assessing the predictability of heat and cold waves in Europe by using robust statistics during the hindcasts of one of the most robust extended ensemble systems, developed at ECMWF
We focus on large-scale HWs and CWs, as these are responsible for the largest share of the impacts, so we upscale all the forecast products to one square degree resolution

Summary

Introduction

Heat or cold waves (HWs or CWs hereafter) are linked to increased risks of mortality, morbidity, and different cardiovascular and respiratory diseases (Anderson and Bell 2009; Gasparrini and Armstrong 2011; Kovats and Hajat 2008; Huynen et al 2001; Thakur et al 1987). Around Europe have been responsible for the design of these warning systems and most of the European countries have implemented tailored action plans in affected areas to minimize harm to those most at risk These plans, which may include early alerts and advisories and a variety of emergency measures to mitigate the heat or cold dangers, are called heat warning systems (HWS) or cold warning system (CWS) (Lowe et al 2011; Stanojevic et al 2014; Åström et al 2014; Matzarakis 2017; Gough et al 2014; Ghosh et al 2014; Chalabi et al 2016; Hajat et al 2013; Masato et al 2015). Osman and Alvarez (2017) illustrates the potential benefit and predictability of heat waves by analysing the scores of two models for a specific case study in Southern America

Objectives

Results

Conclusion