Assessment of an extended version of the Jenkinson\u2013Collison classification on CMIP5 models over Europe

Tim Butler,Noelia Otero,Jana Sillmann

doi:10.1007/s00382-017-3705-y

Abstract

A gridded, geographically extended weather type classification has been developed based on the Jenkinson–Collison (JC) classification system and used to evaluate the representation of weather types over Europe in a suite of climate model simulations. To this aim, a set of models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5) is compared with the circulation from two reanalysis products. Furthermore, we examine seasonal changes between simulated frequencies of weather types at present and future climate conditions. The models are in reasonably good agreement with the reanalyses, but some discrepancies occur in cyclonic days being overestimated over North, and underestimated over South Europe, while anticyclonic situations were overestimated over South, and underestimated over North Europe. Low flow conditions were generally underestimated, especially in summer over South Europe, and Westerly conditions were generally overestimated. The projected frequencies of weather types in the late twenty-first century suggest an increase of Anticyclonic days over South Europe in all seasons except summer, while Westerly days increase over North and Central Europe, particularly in winter. We find significant changes in the frequency of Low flow conditions and the Easterly type that become more frequent during the warmer seasons over Southeast and Southwest Europe, respectively. Our results indicate that in winter the Westerly type has significant impacts on positive anomalies of maximum and minimum temperature over most of Europe. Except in winter, the warmer temperatures are linked to Easterlies, Anticyclonic and Low Flow conditions, especially over the Mediterranean area. Furthermore, we show that changes in the frequency of weather types represent a minor contribution of the total change of European temperatures, which would be mainly driven by changes in the temperature anomalies associated with the weather types themselves.

Highlights

Analysis of large-scale atmospheric circulation by means of synoptic weather-typing has been widely studied over the last few decades
In order to examine how closely models reproduce circulation patterns compared to reanalyses, we examine relative frequencies over the entire European domain during the reference period (1986–2005)
Models are able to reproduce the seasonal frequencies of weather types (WT), and there is a reasonable agreement between reanalyses and the model ensemble (MME)

Summary

Introduction

Analysis of large-scale atmospheric circulation by means of synoptic weather-typing has been widely studied over the last few decades. Atmospheric variability can be analysed in terms of changes of frequencies of specific synoptic weather types (WT) (Huth 2000). Numerous WT classifications have been used for a wide range of applications: human mortality (Kassomenos et al 2001); surface climate variables, such as precipitation (Goodess and Jones 2002; Trigo and DaCamara 2000; Lorenzo et al 2008; Cortesi et al 2013) or temperature (Chen 2000; Post et al 2002); extreme events, such as storms (Donat et al 2010a), or droughts (Paredes et al 2006; Vicente-Serrano and LópezMoreno 2006); environmental variables, such as wildfire. A comprehensive discussion of the different methods and their applicability can be found in Huth et al 2016 and the references therein

Objectives

Methods

Results

Conclusion