Embed Size (px)
Citation preview
Research Collection
Doctoral Thesis
Identifying and quantifying large-scale drivers of Europeanclimate change
Author(s): Kröner, Nico
Publication Date: 2016
Permanent Link: https://doi.org/10.3929/ethz-a-010793497
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
DISS. ETH NO. 23453
Identifying and quantifying large-scale
drivers of European climate change
A thesis submitted to attain the degree of
DOCTOR OF SCIENCES of ETH ZURICH
(Dr. sc. ETH Zurich)
presented by
Nico Robert Kroner
MSc ETH, ETH Zurich
born January 6, 1985
citizen of Germany
accepted on the recommendation of
Prof. Dr. Christoph Schar, examinerDr. Sven Kotlarski, co-examinerDr. Daniel Luthi, co-examinerDr. Erich Fischer, co-examiner
Prof. Dr. Roy Rasmussen, co-examiner
2016
Abstract
Climate change will affect regional climates around the world. While regional climatechange is driven by large-scale changes in the overall thermodynamic structure and generalcirculation of the atmosphere, particular regions are affected by characteristic regionalfeedbacks. The complex interactions between the large-scale drivers and the regionalfeedback processes make it difficult to understand and quantify the involved mechanisms.In this thesis a method is developed to disentangle the role of the large-scale drivers tocontribute to (1) improving our understanding of specific drivers, (2) quantifying theirrespective contribution to the regional climate change and (3) assess the robustness of theprojected changes in dependence of the drivers uncertainties.
The developed method is applied to the European climate change, by studying regionalclimate simulations that cover historical (1971-2000) and future periods (2070-2099).
The main objective of this thesis is to shed light on the driving processes/mechanismsof European temperature and precipitation changes, and thereby to contribute to morerobust and reliable climate projections and climate impact scenarios. To this end, theinfluence of three large-scale drivers of European climate change, namely thermodynamic,lapse-rate and circulation changes are elucidated and quantified by applying and extendingthe so-called surrogate (or pseudo warming) methodology.
In Chapter 3, the surrogate approach is extended and applied to the European summerclimate. The basic idea of this approach is to use a regional climate model and apply alarge-scale warming to the lateral boundary conditions of a present-day reference simula-tion, while maintaining the relative humidity (and thus implicitly increasing the specificmoisture content). In comparison to the basic approach, two important extensions areapplied: (1) a twin design is used, where the large-scale warming is not only added to thepresent day simulation, but also a corresponding cooling is subtracted from the futurescenario simulation; (2) the temperature change is applied independent and dependent ofheight in order to quantify the effect of large-scale lapse-rate changes. The influence oflapse-rate changes on regional climate change was not studied before.
It is shown that the mean warming projected for the European summer climate iscaused by the large-scale thermodynamic effect, which also increases precipitation over thewhole continent. But the peculiar amplification of the Mediterranean warming and drying,as projected by climate models, is not reproduced by the thermodynamic effect alone.Rather results show that lapse-rate changes explain some fraction of this amplification.Over the Iberian Peninsula they explain about half of the amplified warming projectedin the full climate change scenario. The large-scale circulation effect, which also includeseffects due to changes in land-sea contrast, is inherited from the driving global climatemodel (GCM) and is shown to further amplify the north-south temperature contrast.
iii
Together this shows that the peculiar Mediterranean amplification is caused by large-scale circulation and lapse-rate changes.
In Chapter 4 the extended surrogate approach is used to analyse changes in precipita-tion statistics for the winter and summer season over Europe. Beside mean precipitation,four commonly used indices are investigated: Precipitation intensity and frequency, andindices that assess heavy precipitation events (maximum day-long precipitation amount)and dry-spell length (mean consecutive dry days), respectively. For winter, it is shownthat the thermodynamic effect is dominating changes in mean precipitation and heavyevents. Only over the Mediterranean the more uncertain circulation effect is dominatingprecipitation changes. In summer also the lapse-rate effect is important over this region,becoming as strong as the circulation effect. The strong influence of the lapse-rate effectin summer over southern European (here quantified for the first time) is confirmed by anadditional analysis of the multi-model ensemble EURO-CORDEX.
In terms of the underlying mechanism it is found that the thermodynamic effect isdominating changes in precipitation intensity whereas the lapse-rate and the circulationeffect are dominating changes in precipitation frequency and are also the main drivers forchanges in dry-spell length. For summer and winter the dominating driver for changesin heavy precipitation, which are are epically important for impact modelling, is thethermodynamic effect. Overall changes in circulation are considered more uncertain thanchanges in lapse-rate and thermodynamics, thereby providing some indications regardingthe reliability of the projected changes in different regions and indices.
In summary, this thesis introduces and validates an extended surrogate climate changeapproach. By disentangling the large-scale drivers of European climate change, valuableinsight into the underlying processes was gained. In particular, the thesis revealed for thefirst time the strong influence of lapse-rate changes on European climate change.
iv
Zusammenfassung
Ausgelost durch den globalen Klimawandel werden spezifische Veranderungen fur die un-terschiedlichen Regionen der Erde vorhergesagt. Diese Veranderungen werden einen di-rekten Einfluss auf die Gesellschaft, Wirtschaft und Okosysteme der betroffenen Regionenhaben. Es ist deshalb von grossem Interesse die Prozesse, welche diese Veranderungen an-treiben, besser zu verstehen und die mit ihnen verbundenen Unsicherheiten abzuschatzen.
Die regionalen Anderungen werden zwar durch die grossskaligen Anderungen inder thermodynamischen Struktur der Atmosphare und den Zirkulationsanderungen be-stimmt, aber in jeder Region sind zusatzlich regionale Ruckkopplungen aktiv. Daskomplexe Zusammenspiel zwischen den grossskaligen Anderungen und den regionalenRuckkopplungsprozessen erschwert das Verstandniss und die Quantifizierung der invol-vierten Prozesse.
In der vorliegenden Doktorarbeit wird deshalb eine Methode entwickelt, welche esermoglicht, die verschiedenen grossskaligen Einflusse aufzutrennen. Durch diese Auftren-nung kann dazu beigetragen werden, (1) das Verstandnis der spezifischen Treiber zu ver-bessern, (2) ihren jeweiligen Einfluss auf regionale Klimaanderungen zu quantifizieren und(3) die Verlasslichkeit der Vorhersagen an Hand der mit dem jeweiligen Treiber verbundenUnsicherheit zu bewerten.
Das ubergeordnete Ziel dieser Arbeit ist es, die Prozesse/Mechanismen, welche dieAnderungen in Temperatur und Niederschlag uber Europa antreiben, besser zu vestehenund damit zu verlasslicheren Vorhersagen fur Klimaanderungen und deren Auswirkun-gen beizutragen. Um dies zu erreichen, wird der Einfluss von drei grossskaligen Treibern,namlich Anderungen in der Thermodynamik, in der atmospharischen Stabilitat und derZirkulation, auf den europaischen Klimawandel unter Verwendung einer erweiterten, so-genannten Surrogate Methode aufgeklart und quantifiziert; Dazu werden regionale Kli-mamodelle verwendet, die sowohl eine historische (1971-2000) als auch eine zukunftige(2070-2099) Zeitspanne abdecken.
In Kapitel 3 wird die erweiterte Surrogate-Methode vorgestellt und auf das europaischeSommerklima angewandt. Das grundsatzliche Konzept dieser Methode beruht auf der Ver-wendung eines regionalen Klimamodells. Fur eine
”surrogate“ Simulation wird die Tem-
peratur in den Randdaten einer gegenwartigen Klimasimulation soweit geandert, dass inder neuen surrogate Simulation die fur die Zukunft prognostizierte Klimaerwarmung rea-lisiert wird. Zusatzlich wird in diesen Simulationen die relative Feuchte konstant gehalten(dies ist gleichbedeutend mit einer Zunahme der spezifischen Feuchte). Im Unterschied zurregularen Methode werden in dieser Arbeit zwei entscheidende Erweiterungen verwendet:(1) die Erwarmung wird nicht nur auf eine Simulation des gegenwartigen Klimas ange-wandt, sondern auch eine entsprechende Abkuhlung auf die dazugehorige Simulation deszukunftigen Klimas; (2) Die Temperaturanderung wird jeweils einmal unabhangig und
v
abhangig von der Hohe angewandt. Dieses Vorgehen ermoglicht die Quantifizierung einergrossskaligen Stabilitatsanderung. Der Einfluss einer solchen Anderung auf den regionalenKlimawandel wurde bis heute noch nicht erforscht.
Es wird gezeigt, dass die mittlere vorhergesagte Erwarmung des europaischen Som-merklimas durch thermodynamische Anderungen, von jetzt an als thermodynamischerEffekt bezeichnet, verursacht wird. Dieser Effekt bewirkt ausserdem eine Zunahme desNiederschlages uber ganz Europa. Die ungewohnlich starke Erwarmung und Austrock-nung des Mittelmeerraums, welche von vielen Klimamodellen vorhersagt wird, kann vomthermodynamischen Effekt alleine nicht reproduziert werden. Vielmehr zeigen die Ergeb-nisse, dass Anderungen in der Stabilitat einen Teil dieser verstarkten Erwarmung erklarenkonnen. Fur die Iberische Halbinsel beispielsweise erklart dieser Effekt ungefahr die Halfteder projizierten Erwarmung. Der Effekt von Anderungen in der grossskaligen Zirkulati-on wird vom antreibendem globalem Klimamodel bestimmt und verstarkt den Nord-SudGradienten in der Temperatur und dessen Anderungsignal zusatzlich. Zusammenfassendzeigt sich, dass insbesondere Stabilitats- und Zirkulationsanderungen fur die besonderenVeranderungen im Klima des Mittelmeerraum verantwortlich sind.
In Kapitel 4 wird die erweiterte Surrogate-Methode verwendet um Anderungen inNiederschlagscharakter fur das europaische Winter- und Sommerklima zu untersuchen.Neben Anderungen im mittleren Niederschlag werden vier haufig verwendete Niederschla-gindizes analysiert: Niederschlagsintensitat, Niederschlagsfrequenz, der grosster taglicherNiederschlag (Starkniederschlagsereignisse) und die mittlere Anzahl aufeinanderfolgenderTrockentage (Trockenperioden). Es wird gezeigt, dass im Winter der thermodynamischeEffekt die Anderungen im mittleren Niederschlag und den Starkniederschlagen dominiert.Eine Ausnahme bilden die Anderungen im Mittelmeerraum, welche zu dieser Jahreszeitvon der mit einer grosseren Unsicherheit behafteten Zirkulationsanderung dominiert wer-den. Im Sommer ist uber dieser Region der Effekt der Stabilisationsanderungen genausostark wie der Effekt der Zirkulationsanderungen. Eine zusatzliche Analyse eines aus meh-reren Klimamodellen bestehendem Ensembles bestatigt den starken Einfluss von Stabi-litatsanderungen auf den Niederschlag im Mittelmeerraum, welcher in dieser Arbeit zumersten Mal quantifiziert wird.
Im Bezug auf die zugrunde liegenden Mechanismen kann gezeigt werden, dass derthermodynamische Effekt hauptsachlich Anderungen in der Niederschlagsintensitat verur-sacht, wahrend die Effekte der Stabilisations- und Zirkulationsanderungen hauptsachlichdie Niederschlagsfrequenz betreffen und somit auch die Anderungen in Trockenperioden.Sowohl im Winter als auch im Sommer dominiert der thermodynamische Effekt dieAnderung in den Starkniederschlagen. Dieser Zusammenhang ist besonders fur die Auswir-kungen des Klimawandels von Bedeutung. Anderungen in der Zirkulation sind generell mitgrosseren Unsicherheiten behaftet als Anderungen in der Stabilitat oder der Thermody-namik der Atmosphare. Dies lasst Ruckschlusse auf die Verlasslichkeit der vorhergesagtenAnderungen fur die unterschiedlichen Regionen und Indizes zu.
In der hier vorliegenden Doktorarbeit wurde eine erweiterte Surrogate-Methode ein-gefuhrt und validiert. Durch diese Methode konnten die grossskaligen Phanomene, welcheden europaischen Klimawandel antreiben, voneinander separiert werden. Diese Separie-rung wiederum erlaubte es, wertvolle Einsichten in die dem regionalen Klimawandel zu-grunde liegenden Prozesse zu gewinnen. Als erste Studie uberhaupt zeigen die vorliegen-den Resultate den direkten Einfluss von Stabilitatsanderungen auf den Klimawandel inEuropa.
vi
Contents
Abstract ii
Zusammenfassung v
1 Introduction 11.1 Global and European climate change . . . . . . . . . . . . . . . . . . . . . 11.2 Climate modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51.4 Drivers and feedbacks of European climate change . . . . . . . . . . . . . . 6
1.4.1 Dynamic or large-scale circulation changes . . . . . . . . . . . . . . 71.4.2 Thermodynamic changes . . . . . . . . . . . . . . . . . . . . . . . . 81.4.3 Lapse-rate changes . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.5 Separation methods and their results . . . . . . . . . . . . . . . . . . . . . 111.6 Objectives and outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2 The COSMO-CLM Model 15
3 Separating climate change signals into thermodynamic, lapse-rate andcirculation effects: Theory and application to the European summerclimate 173.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.2.1 Introduction to surrogate climate change methodology . . . . . . . 223.2.2 Surrogate setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233.2.3 Arithmetics of the surrogate approach . . . . . . . . . . . . . . . . 253.2.4 Model and setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263.3.1 The thermodynamic effect . . . . . . . . . . . . . . . . . . . . . . . 273.3.2 The lapse-rate effect . . . . . . . . . . . . . . . . . . . . . . . . . . 273.3.3 The circulation and other remaining effects . . . . . . . . . . . . . . 303.3.4 Verification of the approach . . . . . . . . . . . . . . . . . . . . . . 30
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.4.1 Interpretation of the lapse-rate effect and the Mediterranean warm-
ing amplification . . . . . . . . . . . . . . . . . . . . . . . . . . . . 313.4.2 Mediterranean drying and the land-ocean contrast . . . . . . . . . . 343.4.3 Role of greenhouse gases . . . . . . . . . . . . . . . . . . . . . . . . 363.4.4 Assessing projection uncertainties . . . . . . . . . . . . . . . . . . . 36
vii
3.4.5 Limitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403.5 Conclusions and Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4 Quantifying the drivers of European precipitation changes:Large-scalethermodynamics, lapse-rate and circulation changes 434.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 464.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2.1 Surrogate approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 474.2.2 Model-Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 484.2.3 Statistical approach . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504.4 Discussion and conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 534.5 Supplemental Information . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.5.1 Temperature scaling . . . . . . . . . . . . . . . . . . . . . . . . . . 574.5.2 Event frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5 Conclusion and outlook 635.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 635.2 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
A The lapse-rate effect on the Mediterranean warming amplification. Sen-sitivity study and the multi-model perspective 69A.1 Physical mechanism of the lapse-rate effect . . . . . . . . . . . . . . . . . . 69A.2 The lapse-rate effect in the EURO-CORDEX ensemble . . . . . . . . . . . 73
B Further analysis of the surrogate climate change ensemble 75B.1 Different seasons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75B.2 Temperature variability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
C Derivation of pressure adjustment 83
References 85
Acknowledgements 105
Curriculum Vitae 107
viii
Chapter 1
Introduction
1.1 Global and European climate change
Human activities interact with and change the global climate. Long-term measurementsshow that global mean surface temperature is increasing since the late 19th century.The first decade of the 21st century was the warmest decade in the instrumental record(Hartmann et al., 2013). Along with an increasing global mean temperature other aspectsof the climate - for example the water cycle - are changing. In the mid-latitudes of theNorthern Hemisphere precipitation and for the whole troposphere specific humidity haveincreased since 1951 (Hartmann et al., 2013). Another feature of recent climate changeis an increase in temperature and precipitation extremes (Seneviratne et al., 2012; Donatet al., 2013; Hartmann et al., 2013).
The outlined changes are very likely to continue in the future and become even stronger(Stocker et al., 2013). The temperature increase is not projected to happen uniformly butto show a distinct geographical pattern. This pattern originates from two main featuresthe so called ”polar amplification” and the land-sea contrast. The polar amplification ismainly a feature of the Northern Hemisphere and describes the amplification of globalwarming over the Arctic (e.g., Manabe and Stouffer 1980, Manabe et al. 1991). The land-sea contrast is a very robust feature (Manabe et al., 1990; Sutton et al., 2007; Joshi et al.,2008; Boer, 2011) which shows a warming ratio from land to sea of 1.4-1.7 (Lambert et al.,2011). Also the tails of the temperature distributions will be affected with an increase inhot and a decrease in cold extremes (Schar et al., 2004; Alexander et al., 2006; Orlowskyand Seneviratne, 2012; Sillmann et al., 2013).
For precipitation it is projected that heavy and extreme precipitation increases strongerthan the mean (Rajczak et al., 2013; Kharin et al., 2013), but the magnitude of changesand the underlying processes are still debated (Allen and Ingram, 2002; Emori and Brown,2005; Pall et al., 2007; Lenderink and Van Meijgaard, 2008; Berg et al., 2009; O’Gormanand Schneider, 2009; Loriaux et al., 2013; Ban et al., 2015). Certainly, changes in pre-cipitation and its extremes are partly connected to projected changes in atmosphericdynamics. In general, global projections indicate a shift of the mid-latitude jet-streamand the stromtracks to the North (Barnes and Polvani, 2013) together with a widening ofthe Hadley Cell (Tanaka et al., 2004; Mitas and Clement, 2006; Vecchi and Soden, 2007;Lu et al., 2007, 2008). In the Northern Hemisphere this signal is not as clear and theresponse is more complicated due to different feedbacks not present in the Southern Hemi-sphere (Harvey et al., 2012; Chang et al., 2012). Changes in circulation are also connected
1
2 Chapter 1: Introduction
Figure 1.1: Temperature and precipitation changes over Europe from the SRES-A1B simu-lations. Top row: Annual mean, DJF and JJA temperature change between 1980 to 1999 and2080 to 2099, averaged over 21 models. Bottom row: same as top, but for fractional change inprecipitation. (Source of Figure and Caption: IPCC AR4, Chapter11, Figure 11.5 in (Solomonet al., 2007))
to changes in atmospheric stability. The strongest warming is simulated for the uppertropical troposphere consistent with the decrease of the moist adiabatic lapse-rate withtemperature (Bony et al., 2006). Embedded in those large-scale changes are changes onthe regional-scale that are in many cases triggered and constrained by large-scale forcingsbut can be strongly modulated by regional scale feedbacks. In this thesis we will focuson the European climate and especially on the summer season. In the following we willtherefore give a short overview of the specific changes projected for the European climate.
European climate changes follows the global features in many aspects. In figure 1.1the mean temperature and precipitation changes in a multi-GCM ensemble assumingSRES A1B emissions for the annual mean, the winter and summer season are shown.In winter the change is similar to the global pattern, with a warming amplification inthe Northern regions and a decreasing north-south temperature gradient. ”In summer,unlike most other mid-latitude regions and other seasons, the projected summer warmingover Europe has the opposite gradient and exhibits a more pronounced warming in theSouth” (Kroner et al., 2016). To elucidate the processes responsible for this pattern is animportant part of this thesis. The precipitation change shows a bi-polar structure for allseasons with increasing precipitation over northern Europe and decreasing precipitationin the South. The transition area between increasing and decreasing precipitation isoscillating, moving from north to south from the winter to the summer season (Frei et al.,2006; Rajczak et al., 2013).
The shifts in the mean are accompanied by changes in temporal variability and ex-tremes. For Europe increases in summer temperature variability are projected (Scharet al., 2004) but the exact area affected is unclear and depends on the model (Fischerand Schar, 2009; Fischer et al., 2012). Maximum temperatures are projected to increase
1.2 Climate modelling 3
together with an increase in frequency and intensity of heatwaves (Beniston et al., 2007;Kjellstrom et al., 2007; Koffi and Koffi, 2008). The signal for heavy precipitation is lessclear. In the North robust increases in frequency and intensity of heavy precipitationfor the whole year are projected. For the rest of Europe, projected changes depend onthe specific region and season under consideration (Frei et al., 2006; Fowler et al., 2007;Rajczak et al., 2013). The outlined changes will have strong impacts on the Europeansociety, economies and ecosystems and make adaptation measures necessary. It is there-fore of great importance to quantify the uncertainties behind those projections which isdirectly coupled to the understanding of the responsible processes. The individual sourcesof uncertainty will be discussed in Chapter 1.3. In the following chapter the models whichare applied to produce the presented projections will be introduced.
1.2 Climate modelling
The main tool to obtain the presented projections are global climate models (GCMs).The most sophisticated type of these models, so-called general circulation models, areconstructed by discretizing the basic physical laws which govern the climate system on athree dimensional grid. Originating in the 1960s (Manabe and Smagorin, 1967) generalcirculation models have since then been continuously developed to include more and moreprocesses of the climate system (Gordon and Stern, 1982; Boer et al., 1984; Meehl, 1990;Gent et al., 2011). The most recent GCMs that form part of the ensemble used for the fifthassessment report of the Intergovernmental Panel on Climate Change (IPCC) (Stockeret al., 2013) are so-called earth system models (ESMs) (Gent et al., 2011; Giorgetta et al.,2013). These models not only include fully coupled atmosphere and ocean componentsbut also sophisticated land surface models, biogeochemical cycles as well as atmosphericchemistry and and aerosol interactions. This development was already foreseen by Lorenz(1970).
Although ESMs represent the most complete models of the Earths climate, their spatialresolution is too coarse to explicitly resolve all relevant processes in the climate system.Aspects which require physical parametrisations include for example micro-physical pro-cesses, moist convection, turbulent fluxes and land surface processes. Today, the spatialresolution of climate GCMs is about 100km-200km which results in an effective resolu-tion of around 500km-1000km (Grotch and Maccracken, 1991; Grasso, 2000). With thisresolution the general circulation of the atmosphere and ocean as well as temperatureand precipitation patterns on the sub-continental scale can be sufficiently resolved (Rum-mukainen, 2010). Nevertheless important details that are key to represent regional andlocal-scale climate processes such as heavy precipitation and local weather phenomena(e.g cold fronts, etc.) are not or only rudimentary represented. To overcome these short-comings GCM results can be downscaled to a higher resolution. In general there are twodownscaling concepts: (1) statistical downscaling where statistical relationships derivedfrom observations are used to transfer the GCM output to the required spatial resolution;(2) dynamical downscaling via the use of a regional climate model (RCM) which featuresa higher spatial resolution and is applied over a specific region of interest only. Besidesthese two approaches, the development of variable-resolution GCMs that are run at in-creased resolution over the domain of interest and at a coarser resolution over the rest of
4 Chapter 1: Introduction
the globe has recently been pushed (Gibelin and Deque, 2003; McGregor and Dix, 2008;Zaengl et al., 2015).
We here focus on dynamical downscaling via RCMs as this technique allows us to ad-dress the specific objectives of this thesis in a straightforward manner (see below). RCMsare limited-area models based on the same physical laws and governing equations asGCMs, although most RCMs do not feature ocean and sea-ice components (Giorgi, 2008;Laprise, 2008; Rummukainen, 2010). Only in recent years the development of so-calledregional earth system models began. They were applied to better understand climatechange processes for instance in the North Sea and Baltic Seas as well as in the Mediter-ranean (Schrum et al., 2003; Sevault et al., 2014), where regional-scale ocean-atmosphereinteractions play an important role. The spatial resolution for current regional modelsfor century-long experiments spans 12.5-50 km. The higher resolution in comparison to aGCM can improve the simulation of precipitation due to a better representation of topog-raphy (Frei et al., 2003; Schmidli et al., 2006; Chan et al., 2012; Torma et al., 2015) alsoan improvement in local dynamics and small scale temperature variability for examplealong complex coastlines has been found (Feser et al., 2011; Hall, 2014).
The principal mechanism in regional modelling is that GCM data are used to drivean RCM over a limited area. The GCM data are provided to the RCM within a lateralboundary zone which is mostly 4-10 grid points wide (Rummukainen, 2010). A relaxationprocedure is used to merge the boundary data with the actual solution of the RCM inthe boundary zone (Davies, 1976). Typically, the variables provided by the GCM areair temperature, wind components, pressure, humidity and, depending on the region, seasurface temperature and sea-ice (Rummukainen, 2010). This boundary forcing cannotonly be derived from GCM experiments but also from atmospheric re-analyses. In generalthe boundary data can be manipulated and changed, which makes RCMs a powerful toolfor sensitivity and process studies (Schar et al. 1999, Fischer et al. 2007, e.g.). Thispossibility will be extensively used in the frame of the present thesis.
The idea of limited are models is not new. In numerical weather prediction such modelsare successfully used on a regular bases for quite some time (Bauer et al., 2015). Actuallymany RCMs were developed from numerical weather prediction models for example theRCM COSMO-CLM from the numerical weather prediction model COSMO (Baldaufet al., 2011). Especially non-hydrostatic RCMs are often validated and used in the contextof numerical weather prediction.
With increasing computer power spatial resolution is becoming finer and in recentyears convection-resolving simulations became feasible (Kendon et al., 2012; Ban et al.,2014, 2015; Leutwyler et al., 2016). Nevertheless also very-high resolution RCMs stillsuffer from the problem that not all processes can be resolved. For example, shallowconvection or turbulent fluxes still happen on much finer scales than resolved by theRCM grid. In the end, the resolution will never be high enough to resolve all processes.Nevertheless, resolving deep convection is a promising attempt to remove one of the mostcrucial parametrizations.
In summary the mentioned caveats can be associated with uncertainty in the regionalclimate change projections. This uncertainty is called model uncertainty which is one ofthree important sources of uncertainty established in climate science modelling. Theseare presented in the next section.
1.3 Uncertainty 5
1.3 Uncertainty
In Chapter 1.1 the projected changes to the European climate were presented. Thesechanges are subject to considerable uncertainties, originating from three distinct aspects:scenario uncertainty, internal variability and model uncertainty.
• Scenario uncertainty is related to the unknown evolution of anthropogenic green-house gas (GHG) emissions. These depend on a highly uncertain technological,socio-economic and population development. To tackle this uncertainty a scenariostrategy was adopted. Preceding IPCCs Fifth Assessment Report (Stocker et al.,2013) four representative concentration pathways (RCPs) were developed. Rep-resentative in that way that different scenarios for the future development of thehuman society which have similar GHG and aerosol emissions are represented by oneRCP without preferring one of the scenarios. Pathways because the RCPs providethe time-dependent evolution of GHG emissions and not only a specific long-termconcentration (Moss et al., 2008; van Vuuren et al., 2011). From the four RCPs onestands for low emission scenarios the RCP2.6, two for stabilization scenarios theRCP4.5 and the RCP6 and one for high emission scenarios the RCP8.5 (Moss et al.,2010). The numbers represent the changes in radiative forcing (W/m2) reached in2100 relative to pre-industrial conditions. For near term projections emission sce-nario uncertainty is rather small. But for long term projections scenario uncertaintycan become the dominant term (Hawkins and Sutton, 2009).
• Internal variability is a property of many complex non-linear systems (Lorenz,1963). Already small perturbations in the initial conditions can change the tra-jectory of a system in phase space completely, making exact long-term projectionsimpossible. This is important for weather forecasting as it sets a natural limit to thepredictability of weather. In climate projections one tries to circumvent this prob-lem by looking at weather statistics instead of single events. Nevertheless for nearand mid-term climate projections, internal variability remains an important sourceof uncertainty (Hawkins and Sutton, 2009; Deser et al., 2012b; Rowell, 2012; Knuttiand Sedlacek, 2013; Shepherd, 2014). This is caused by features of the global climatesystem which produce internal variability at interannual, decadal and multi-decadaltime scales (e.g. North Atlantic Oscillation (NAO) or El Nino-Southern Oscillation(ENSO)). When moving from global to regional scales internal variability remainsan important source of uncertainty even for long-term projections (Deser et al.,2012b,a).
• Model uncertainty was already introduced in Section 1.2 in connection with thephysical parametrisations. Climate models are per definition an imperfect represen-tation of the real climate system. Several reasons lead to this imperfection. RCMs oreven large eddy simulations (LES) do not provide resolutions high enough to explic-itly resolve all governing processes. These processes have to be parametrised whichoften requires empirical tuning of model parameters. This tuning is often donesubjectively based on expert judgement (although Bellprat et al. (2012) recentlydeveloped a method to perform this tuning in an objective way) and is afflictedwith considerable uncertainties. Another problem which stands beside the techni-cal limitation related to the grid spacing is the problem of process understanding.
6 Chapter 1: Introduction
Several processes are not understood well enough for that a reliable parametrizationcould be constructed. But the lack of understanding is not confined to the unre-solved processes; also the complex interplay of the resolved processes still possesunsolved questions. All of the outlined uncertainties are summarized in the termmodel uncertainty.
In climate science the general problem of uncertainty is typically addressed by us-ing ensemble studies. The projections presented in Section 1.1 are based on a so-calledmulti-model ensemble (MME). MMEs are constructed by using climate simulations fromdifferent climate models (GCMs and/or RCMs). They can sample all three types of uncer-tainty. But how exactly such an ensemble should be constructed and also evaluated is stilldebated (Knutti et al., 2010; Sanderson and Knutti, 2012; Sanderson et al., 2015). Recentcomprehensive GCM-based MMEs were the Coupled Model Inter-comparison Projects(CMIP) 3 (Meehl et al., 2007) and 5 (Taylor et al., 2012) which provided the basis forIPCCs AR4 (Meehl and Stocker, 2007) and AR5 (Stocker et al., 2013), respectively. Forthe European continent two high resolution GCM-RCM MMEs already exist and a thirdone is currently being constructed. The two completed ones are the PRUDENCE (Chris-tensen and Christensen, 2007) and the ENSEMBLES (van der Linden and Mitchell, 2009)ensembles. The ongoing EURO-CORDEX (Jacob et al., 2014) project runs under the um-brella of the CORDEX (Giorgi et al., 2009) initiative which aims at providing RCM MMEsfor all terrestrial regions of the world.
MMEs are a well-accepted method to represent the uncertainties related to climateprojections, although they are unlikely to sample the whole uncertainty range (Knutti,2010; Sanderson et al., 2015). Also, the large variety of models and their interdependencymakes it difficult to use MMEs for process studies. To elucidate certain processes itis often more fruitful to design special experiments which allows modellers to reduce thecomplexity of the climate system. This can be done by using much simpler approaches, forexample single-column convective equilibrium models or intermediate complexity modelsor by varying only one feature in a complex model (so-called sensitivity experiments). Inthis thesis the last approach is applied.
In the following Section the drivers and feedbacks governing the European climatechange are closer examined and a strategy is proposed to improve our understanding andthe relative importance of these drivers.
1.4 Drivers and feedbacks of European climate
change
In this thesis we are mainly interested in how different large-scale drivers effect Europeanclimate change and how their influence is modulated by regional feedbacks. For examplea circulation change induced drying could lead to stronger and more severe heat wavesvia soil-moisture temperature interactions.
A common classification of the large-scale drivers of climate change is a separationinto thermodynamic and dynamic changes (Emori and Brown, 2005; Rowell and Jones,2006; Seager et al., 2010; Kendon et al., 2010). We will follow this approach but will add athird large-scale driver: lapse-rate changes. Large-scale lapse-rate changes were until nownot considered in connection to European climate change, although strong changes in the
1.4 Drivers and feedbacks of European climate change 7
Figure 1.2: Large-scale drivers of European climate change
atmospheric lapse-rate are also simulated for the mid-latitudes (Frierson, 2006). Anotherlarge-scale feature which is regarded to have important influence on European climatechange is the land-sea warming contrast (Feudale and Shukla, 2011). Unfortunately theland-sea contrast is directly connected to dynamic changes (Brayshaw et al., 2009; Cattoet al., 2011; Woollings et al., 2012) and therefore we will not quantify it separately.
The benefits of disentangling these three large-scale driver are threefold: (1) we couldimprove our understanding of specific drivers, (2) we could quantify their respective con-tribution to the total climate change signal and (3) we could separate the climate changesignal into rather certain and rather uncertain aspects; for example changes attributed todynamic changes generally have a higher uncertainty than pure thermodynamic effects.In the following, the three main large-scale drivers are presented in more detail and theirchanges with climate change are examined.
1.4.1 Dynamic or large-scale circulation changes
For the European region the most important large-scale circulation phenomena are theNorth Atlantic Oscillation (NAO), extra tropical cyclones (ETCs) and blockings (Hurrellet al., 2003; Folland et al., 2009; Dole et al., 2011; Mariotti and Dell’Aquila, 2012). These3 phenomena are not independent but do influence each other. Nevertheless we follow theIPCC AR5 in considering them separately (Collins et al., 2013).
The NAO is a prominent teleconnection pattern which can explains large parts ofatmospheric circulation variability. It manifest itself as a spatial pattern in the NorthernHemisphere’s geopotential height field describing a pressure seasaw between the Azoresand the Iceland regions in the North Atlantic. Its positive and negative phase go alongwith an above-average and below-average pressure difference between these two regionsand can considerably influence the European climate (Hurrell et al., 2003; Wallace andHobbs, 2006). In general the pattern and ist influences are more pronounced in winter thanin summer. The future trend of the NAO is not clear; the CMIP2, 3 and 5 ensemblesshow contradicting trends (Stephenson et al., 2006; Boe et al., 2009; Gillett and Fyfe,
8 Chapter 1: Introduction
2013). Therefore also its contribution to future European climate change is uncertain(Christensen et al., 2013).
ETCs build in the strong baroclinic environment between the warm tropical and coldpolar air masses and are a common phenomenon in the mid-latitudes (Eady, 1949; Pet-terss and Smebye, 1971; Davis and Emanuel, 1991). The whole North Atlantic ETCstogether form the north Atlantic storm track and are especially important in the winterseason (Chang et al., 2002; Wallace and Hobbs, 2006). The ETCs are connected to thepassage of fronts over the European continent and contribute strongly to the day-to-dayweather variability (Chang et al., 2002). On a global perspective a robust poleward shiftof the storm tracks is simulated but the North Atlantic storm track is a special case.Many models simulate increased storm activity and an extension into Europe instead ofa latitudinal shift (Bengtsson et al., 2006; Ulbrich et al., 2008; McDonald, 2011). Thiswould mean more and stronger ETCs over Europe in winter. But these results have to beregarded with caution as GCMs have several problems with simulating the storm tracksin the current climate. Simulated storm tracks are often dislocated (Woollings, 2010),too zonal and cyclone intensity is underestimated (Colle et al., 2013; Zappa et al., 2013).In terms of storm track changes due to climate change the uncertainties are even larger.The simulated changes depend on a variety of factors from the horizontal model reso-lution (Colle et al., 2013) to their capability to accurately represent subtropical uppertropospheric temperatures (Haarsma et al., 2013). Because of this, the IPCC assesses thesimulated changes to be of low confidence (Collins et al., 2013).
Blocking patterns are quasi-stationary high pressure systems which, for instance, candeflect ETCs from their normal path over Europe (Legras and Ghil, 1985; Wallace andHobbs, 2006). They are connected to very stable weather situations over Europe and canbe associated with severe summertime heat waves (Dole et al., 2011; Quesada et al., 2012;Vautard et al., 2013; Miralles et al., 2014). Similar to the changes in the storm tracksthe IPCC assess the projected changes for blocking to be of low confidence (Christensenet al., 2013).
In the summer season another large-scale circulation feature becomes important forthe European climate: the Hadley Circulation. The Hadley cell is formed by the up-drafts in the inter tropical convergence zone and their counterpart, the down-drafts in thesubtropics. It is projected that the latitudinal width of the Hadley cell is increasing (e.g.Lu et al. 2007; Collins et al. 2013), which translates into a shift of the subtropical dryzones towards the pole (Seidel et al., 2008; Scheff and Frierson, 2012; Collins et al., 2013).Although this feature is rather robust in general the circulation changes over Europe areregarded as rather uncertain and an important reason hindering more reliable climatechange projections in this region (Woollings, 2010).
1.4.2 Thermodynamic changes
The thermodynamic effect is mainly caused by the increase of greenhouse gas (GHG)concentrations in the atmosphere and the global greenhouse effect. Because of higherGHG levels more long-wave radiation is trapped in the atmosphere which leads to ageneral rise of global mean temperature. This rise of temperature has several directconsequences, two of them directly affecting temperature itself: The snow/ice albedoeffect and the land-ocean contrast. The snow/ice albedo effect is a positive feedback.Increasing temperatures lead to a decrease in snow and ice cover which leads to a lower
1.4 Drivers and feedbacks of European climate change 9
surface albedo and less reflection of solar energy, hence a further increase of temperature(Wallace and Hobbs, 2006). The land-ocean contrast contrary to previous theories, isnot primarily related to the large thermal inertia of the ocean but to boundary layer,lapse-rate and cloud feedbacks (Joshi et al., 2008; Collins et al., 2013; Sherwood and Fu,2014). Although a more recent study, so far only considering one GCM, challenges theminor role of the ocean heat uptake (Sejas et al., 2014).
Also the water cycle is effected by the temperature increase because the water holdingcapacity of air is directly linked to its temperature via the Clausius-Clapyeron (CC)relation. The CC relation connects temperature T to the saturation water vapour pressurees:
∂es∂T
=Lv ∗ esRvT 2
(1.1)
where Lv is the latent heat of vaporization and Rv the universal gas constant for air(Hartmann, 1994). The CC relation implies an increase of specific moisture of roughly 6-7%/K for constant relative humidity, and on the global scale models actually project thatrelative humidity will stay constant (Allen and Ingram, 2002; Held and Soden, 2006). Themoistening of the atmosphere has direct effects on precipitation but one has to distinguishbetween mean and extreme precipitation. On the global scale mean precipitation is notprojected to increase by the same amount as specific humidity. The increase is limited bythe energy balance between radiative cooling and latent heating to 1-3%/K (Allen andIngram, 2002; Held and Soden, 2006; Schneider et al., 2010). On the regional scale thechanges in mean precipitation can substantially differ from that scaling, because here alsosmall-scale thermodynamic and dynamic features become important (O’Gorman et al.,2012). The situation is different for extreme precipitation.
A robust response to increasing atmospheric greenhouse-gas concentrations is thatheavy precipitation events increase stronger than the mean (Allen and Ingram, 2002;Trenberth et al., 2003; Kharin et al., 2013; Rajczak et al., 2013; Fischer et al., 2014). Theexact increase depends on the time-scale considered. For daily precipitation extremesmost studies suggested that they follow the Clausius-Clapeyron (CC) relation (Frei et al.,1998; Allen and Ingram, 2002; Allan and Soden, 2008; Ban et al., 2015) which impliesan increase of about 6-7%/K. For sub-daily time-scales some observational and modellingstudies suggest a higher so called super-adiabatic scaling (Lenderink and Van Meijgaard,2008; Kendon et al., 2012; Berg et al., 2013; Loriaux et al., 2013). Although a more recentmodelling study suggest a CC-scaling also for hourly precipitation (Ban et al., 2015).
1.4.3 Lapse-rate changes
Lapse-rate changes describe changes of the static stability, one of the most importantquantities describing the structure of the atmosphere. The warming maximum in theupper tropical troposphere is a robust feature of climate change projections in the CMIP3and CMIP5 ensembles (Fig. 1.3). The simulated changes are consistent with theoreticalargumentations for a upper tropospheric warming maxima (Bony et al., 2006; Allan andSoden, 2008; Johnson and Xie, 2010). Most of these arguments are connected to themoist adiabatic lapse-rate which is decreasing with increasing temperatures (Schneideret al., 2010). Nevertheless there is still a discussion going on in the scientific communitywhether upper tropospheric warming maxima should already be detectable in current
10 Chapter 1: Introduction
Figure 1.3: CMIP5 multi-model changes in annual mean zonal mean temperature in the atmo-sphere for 20812100 relative to relative to 19862005 under the RCP2.6 (left), RCP4.5 (centre)and RCP8.5 (right) forcing scenarios. Hatching indicates regions where the multi-model meanchange is less than one standard deviation of internal variability. Stippling indicates regionswhere the multi-model mean change is larger than two standard deviations of internal variabilityand where at least 90% of the models agree on the sign of change (see Box 12.1). (Source ofFigure and Caption: IPCC AR5, Chapter12, Figure 12.12 in (Collins et al., 2013))
measurements (Hartmann et al., 2013). In terms of its effects on climate change, thelapse-rate change was mostly discussed as a global scale negative feedback (Bony et al.,2006) and in terms of its connection to the change of the Hadley circulation (Mitas andClement, 2006; Collins et al., 2013). However, Frierson (2006) showed that lapse-ratechanges are not only important in the tropics but also for the mid-latitudes. In this thesistheir direct influence on European climate change will be quantified for the first time.
The presented large-scale drivers are modulated by regional and local-scale feedbackswhich result finally in the projected changes of climate variables including the meanstate, the variances and the extremes (Christensen et al., 2013). Next we will give a shortoverview of the most important feedbacks for the European summer climate acting onthe local to regional-scale:
• Soil-moisture temperature interactions: Soil-moisture is the crucial variable deter-mining the partitioning of incoming solar radiation into sensible and latent heat.A decrease in soil moisture can reduce latent cooling considerably and can increasesensible heating (Fischer, 2007; Seneviratne et al., 2010). This in turn can consid-erably increase near-surface temperatures. It was shown that via this mechanismsoil moisture is a crucial variable for explaining changes in temperature variabilityand warm temperature extremes over Europe (Seneviratne et al., 2006a; Jaeger andSeneviratne, 2011; Boe and Terray, 2014).
• Soil-moisture precipitation feedback: The feedback between soil-moisture and pre-cipitation is complex and has a temporal as well as spatial component (Ruscicaet al., 2015; Guillod et al., 2015). It involves interactions between boundary layerprocesses and advected moisture (Schar et al., 1999). Conventional climate modelsmostly simulate a positive feedback in which a decrease in soil moisture prohibits
1.5 Separation methods and their results 11
further precipitation which decreases soil moisture even further and so on (Pal andEltahir, 2001; Koster et al., 2004; Seneviratne et al., 2006b). But high resolutionstudies question this feedback and can even simulate an opposite feedback sign,i.e. increasing precipitation sums with decreasing soil moisture (Hohenegger et al.,2009). Recent studies by Boe (2013) and Froidevaux et al. (2014) argue that thefeedback depends on the background flow. This would fit into the theory of Scharet al. (1999) that the advected moisture also plays an important role. The questionwhich sign is more realistic for climate change studies remains open so far.
• Cloud radiation interactions: Clouds play an important role by altering the radiationbalance of the Earth by reflecting incoming solar radiation and trapping outgoinglong-wave radiation. Observations (Tang et al., 2012) and models studies (Cattiauxet al., 2013b; Boe and Terray, 2014) suggest that clouds play an important directrole in shaping European temperature change.
In the next section different methods to separate large-scale drivers are summarized andthe influence of these drivers on the European summer climate is presented.
1.5 Separation methods and their results
The separation of regional climate change signals into components that can be linked tospecific large-scale drivers is a matter of ongoing research. Several methods are appliedto perform this disentangling. One way to classify the different methods is to examineon which driver they focus. Three groups can be identified: (A) methods that focus onthe dynamic drivers and treat thermodynamic influences as residuals, (B) methods that,vice versa, quantify the influence of thermodynamic drivers and that account for dynamicinfluences as residuals, and (C) methods which consider both classes simultaneously. Themethods focusing on the dynamic drivers (class A) can be applied to a range of variables(temperature, precipitation, etc.) and mostly rely on daily surface or upper-level pres-sure fields (Vautard, 1990; Huth et al., 2008; Cattiaux et al., 2013b; Deser et al., 2016).Conversely, Class B methods are mainly applied to precipitation because they typicallyuse the CC-relation (see section 1.4) as a thermodynamic proxy (Pall et al., 2007; Ra-dermacher and Tomassini, 2012). The third type of methods (Class C) also mostly focuson precipitation changes. They compute changes to the moisture or energy balance andattribute them to thermodynamic or dynamic drivers often using changes in the verticalwind ∆w as proxy for dynamic changes and changes in the vertically integrated specifichumidity ∆dqc/dp as thermodynamic proxy (e.g. Emori and Brown 2005; Seager et al.2010; Chou et al. 2012). All of the mentioned methods so far use standard climate modeloutput and perform the disentangling a posteriori. In contrast to this, the surrogate cli-mate change method developed by Schar et al. (1996) achieves the disentangling prior tothe model simulation. This is done by manipulating the GCM-derived lateral boundaryforcing for RCMs. The basic method would fall under category B but it can be expandedto quantify more drivers (Rowell and Jones, 2006; Kendon et al., 2010; Rasmussen et al.,2011). The method was picked up by the community and applied to different researchquestions (Frei et al., 1998; Seneviratne et al., 2002; Rowell and Jones, 2006; Kendonet al., 2010; Im et al., 2010; Rasmussen et al., 2011; Sorland and Sorteberg, 2015). In
12 Chapter 1: Introduction
this thesis the basic method of Schar et al. (1996) will be expanded. A more detaileddescription of the extended surrogate approach is presented in Chapter 3.
Next, the findings of the influence of the large-scale drivers on the European climateobtained with the methods presented before are summarized. First precipitation changesare examined. In general the studies discriminate between mean and heavy/extremeprecipitation changes. For mean precipitation changes it was found that, for most seasonsand subregions of Europe, the thermodynamic effect is dominating the projected changes(Emori and Brown, 2005; Held and Soden, 2006; Seager et al., 2010). Only in certainsubregions and seasons the circulation changes are dominating. For the European summerclimate for example several studies found that the widening of the Hadley cell and ageneral increase of anticyclonic weather or circulation types are the dominate drivers fora precipitation decrease over southern Europe (Vautard and Yiou, 2009; Seager et al.,2014; Santos et al., 2016). This is seemingly contradicting the analyses of Rowell andJones (2006) and Kendon et al. (2010) who found that the general warming explains mostof the precipitation decrease in this area. These contradictions can be explained by thedifferent definitions of thermodynamic and dynamic responses. In the study of Rowelland Jones (2006) and Kendon et al. (2010) the large-scale thermodynamic effect includedalso the large-scale latitudinal temperature change and therefore also a change in thelarge-scale circulation was included in the thermodynamic response (see Chapter 3). Thispoint shows that great care has to be taken in any separation effort, especially in thedefinition of the individual large-scale drivers.
The governing mechanisms and remaining open questions concerning heavy and ex-treme precipitation were already discussed in section 1.4 and do mostly also apply forEuropean climate change.
In terms of temperature changes several studies showed that circulation changes arenot directly responsible for the typical European temperature response pattern (Plavcovaand Kysely, 2013; Cattiaux et al., 2013b; Belleflamme et al., 2015). Most studies so farinvoke regional feedbacks to explain the pronounced Mediterranean warming such as thesoil-moisture temperature feedback (Seneviratne et al., 2006a; Fischer et al., 2007; Vidaleet al., 2007; Boe and Terray, 2014) or cloud-radiation interactions (Rowell and Jones, 2006;Tang et al., 2012; Cattiaux et al., 2013b; Boe and Terray, 2014). However, it could so farnot be quantified which large-scale changes drive those regional feedbacks. To our knowl-edge the influence of lapse-rate changes on European climate change was not quantifiedso far, although several studies already mention that changes in the atmospheric stabilitycould have important influences (Emori and Brown, 2005; Radermacher and Tomassini,2012; Cattiaux et al., 2013b). Until now only O’Gorman and Schneider (2009) developeda framework to quantify its influence but applied it to global scale heavy precipitationchanges only. They showed that their scaling approach including lapse-rate changes issuited to explain heavy precipitation changes in GCM simulations.
In summary, the dependency of mean precipitation changes on thermodynamic anddynamic drivers is rather robust. Open issues remain, for instance concerning the questionhow the different drivers affect precipitation frequency and intensity and concerning theresponse of heavy and extreme precipitation. For temperature changes it is establishedthat changes in large-scale circulation do not contribute directly to them but it remainsopen which large-scale effect is driving the obviously important local feedback processes.The largest unknown are the lapse-rate changes.
1.6 Objectives and outline 13
1.6 Objectives and outline
In the previous sections it was shown that strong changes are projected for the Europeanclimate which calls for the development of adaptation measures in various sectors. It istherefore of high importance to maximise the robustness of projections which ultimatelyrequires to understand the processes which drive the projected changes. It was outlinedthat European climate change is governed by a huge variety of processes which act on localto global scales. A promising strategy to tackle this complexity is the disentangling of thedifferent large-scale drivers. It was illustrated that the different methods to achieve thisseparation yield interesting results but that they are also restricted in terms of climatevariables (focus on precipitation) and large-scale drivers (no quantification of the lapse-rate effect). In this thesis the surrogate climate change method is expanded to addressthose caveats. The precise objectives are:
1. To establish an extended surrogate method that goes beyond previous works and totest its applicability for European climate change.
2. To quantify the influence of large-scale thermodynamic (TD), lapse-rate (LR) andlarge-scale circulation (CO) changes on the European summer climate. To quantifytheir respective contributions to changes in temperature and precipitation patterns.
3. To identify the processes behind the typical bipolar European precipitation changepattern in terms of changes in precipitation intensity and frequency. To evaluate thehypothesis of stronger thermodynamic control on heavy and extreme precipitation.
The thesis is structured in five chapters which are outlined in the following:
• Chapter 2: The regional climate model COSMO-CLM
In this chapter the main tool of this study - the regional climate model COSMO-CLM - is described.
• Chapter 3: Separating climate change signals into thermodynamic, lapse-rate and circulation effects: Theory and application to the Europeansummer climate. (Kroner et al. 2016, Climate dynamics)
The aim of this study is twofold: First, the extended surrogate approach is intro-duced and validated. Second, the new approach is used to quantify the contributionfrom TD, LR and CO changes on the European summer climate. The study showsthat the new approach provides an elegant way to separate the full climate changesignal into the individual contributions. Further, it confirms the strong influence ofthe CO on European drying and it shows for the first time the strong influence ofthe LR effect on the Mediterranean warming amplification.
• Chapter 4: Quantifying the drivers of European precipitation changes:Large-scale thermodynamics, lapse-rate and circulation changes. (Kroneret al 2016, in preparation)
In this study the extended surrogate approach introduced in Chapter 3 is used toanalyse the changes in European precipitation statistics. Including precipitationfrequency and intensity as well as heavy precipitation and prolonged dry spells. We
14 Chapter 1: Introduction
find that the TD effect is mainly increasing precipitation intensity whereas the COand LR effect are mainly decreasing precipitation frequency. We also shown thatfor heavy precipitation the influence of the TD effect is stronger than on the meanwhereas for the CO and LR effect it is the opposite.
• Chapter 5: Conclusions and Outlook
This chapter summarizes the main conclusions of the thesis and provides suggestionsand ideas for further research.
• Appendix A: The Mediterranean warming anomaly and the lapse-rateeffect. A multi-model perspective
In this figure outline a additional sensitivity experiment to elucidate the mechanismbehind the LR effect in the Mediterranean warming amplification is presented. Fur-ther it is shown, that the lapse-rate effect on the Mediterranean warming anomalyis not only found in our surrogate ensemble but is a robust feature in the newEURO-CORDEX ensemble.
• Appendix B: Further analysis of the surrogate climate change ensemble(other seasons and temperature variability)
This appendix gives an overview of several further analyses carried out with thesurrogate ensemble which complements the analysis shown in Chapters 3 and 4.First the separation of TD, LR and CO effects on temperature and precipitationchanges is also shown for the other seasons. Second the influence of TD, LR and COon summer temperature variability measured as inter-annual variability and totaldaily variability is shown. It can be seen that without CO changes temperaturevariability is not or only marginally increasing in the mid-latitudes.
• Appendix D: Derivation of the pressure adjustment applied to the bound-ary data
In the surrogate climate change method the temperature and specific humidity ofthe boundary data is changed. The procedure to maintain hydrostatic balance inthe pressure field of the boundary data is derived in this appendix.
Chapter 2
The COSMO-CLM Model
The main research tool of this thesis is the RCM COSMO-CLM (CCLM) (Rockel et al.,2008). CCLM is the climate version of the numerical weather prediction model COSMO(Doms et al., 2011; Doms and Schattler, 2015) developed and maintained by the ClimateLimited-area Modelling-Community (CLM-Community, http://www.clm-community.eu).The COSMO model was developed by the Consortium of Small-scale Modelling (COSMO,http://www.cosmo-model.org/) out of the Local Model (LM) from the German weatherservice DWD. CCLM features a dynamical core which solves the non-hydrostatic andfully-compressible equations describing atmospheric motion. It uses the Runge-Kuttatime-integration scheme together with a third-order horizontal advection scheme. Furtherthe model system includes a whole suite of parametrizations to represent subgridscaleprocesses like radiation, convection or micro-physics as well as interactions with the oceanand land surfaces.
We use the model in a well-tested version for regional climate applications(cosmo4.8 clm17). The same version of CCLM was used to carry out simulations con-tributing to the EURO-CORDEX multi-model-ensemble (Kotlarski et al., 2014; Jacobet al., 2014; Keuler et al., 2016). In the mentioned version, among others, the followingparametrizations schemes are applied: For radiation the δ-two-stream radiation scheme(Ritter and Geleyn, 1992), for convection the Tiedtke scheme (Tiedtke, 1989), a one-moment scheme for cloud micro-physics which includes cloud liquid water and ice andprognostic equations for rain and snow (Reinhardt and Seifert, 2006; Doms et al., 2011),and the surface and soil model TERRA-ML (Doms et al., 2011) which features a multi-layer soil. In the setup applied here, the soil consists of ten non-equidistant layers with thecentre of the lowermost layer located at 11.5 m depth. The relaxation procedure of Davis(1976) is used to impose the large-scale forcing at the lateral boundaries. The spongezone around the domain has a width of 10 grid points. The model has 40 vertical layerswhich stretch from the surface to a height of 22.7 km where the nine uppermost layersstarting from 11 km serve as damping layers for the model top. The vertical coordinateis a height-based terrain following hybrid Gal-Chen coordinate.
At ETH Zurich the CCLM is applied over a wide range of grid-spacing (100 m −100 km) and times (1 day − 100 years). Recently it has been used for high resolutionconvection-permitting experiments (Ban et al., 2014, 2015; Keller et al., 2015) and forregional climate change studies, for instance, in the frame of EURO-CORDEX (Kotlarskiet al., 2014; Keuler et al., 2016). The model can also be used in idealized settings (Schlem-mer et al., 2011; Hassanzadeh et al., 2014). As one of the first RCMs, CCLM can also be
15
16 Chapter 2: The COSMO-CLM model
run on new computer architectures utilizing graphical processor units (GPUs) (Leutwyleret al., 2016).
In this thesis the COSMO-CLM is used to apply a extended surrogate framework themethod is described in detail in Chapter 3 and in Chapter C an derivation for the pressureadjustment can be found.
Chapter 3
Separating climate change signalsinto thermodynamic, lapse-rate andcirculation effects: Theory andapplication to the European summerclimate
17
19
Separating climate change signals into thermodynamic,lapse-rate and circulation effects:
Theory and application to the European summer climate∗
Nico Kroner1, Sven Kotlarski 1,2, Erich Fischer1, Daniel Luthi1,Elias Zubler1,3 and Christoph Schar1
Abstract
Climate models robustly project a strong overall summer warming across Europe show-ing a characteristic north-south gradient with enhanced warming and drying in southernEurope. However, the processes that are responsible for this pattern are not fully under-stood. We here employ an extended surrogate or pseudo-warming approach to disentanglethe contribution of different mechanisms to this response pattern. The basic idea of thesurrogate technique is to use a regional climate model and apply a large-scale warming tothe lateral boundary conditions of a present-day reference simulation, while maintainingthe relative humidity (and thus implicitly increasing the specific moisture content). Incomparison to previous studies, our approach includes two important extensions: First,different vertical warming profiles are applied in order to separate the effects of a meanwarming from lapse-rate effects. Second, a twin-design is used, in which the climatechange signals are not only added to present-day conditions, but also subtracted from ascenario experiment. We demonstrate that these extensions provide an elegant way toseparate the full climate change signal into contributions from large-scale thermodynamic(TD), lapse-rate (LR), and circulation and other remaining effects (CO). The latter inparticular include changes in land-ocean contrast and spatial variations of the SST warm-ing patterns. We find that the TD effect yields a large-scale warming across Europe withno distinct latitudinal gradient. The LR effect, which is quantified for the first time inour study, leads to a stronger warming and some drying in Southern Europe. It explainsabout 50% of the warming amplification over the Iberian Peninsula, thus demonstratingthe important role of lapse-rate changes. The effect is linked to an extending Hadleycirculation. The CO effect as inherited from the driving GCM is shown to further amplifythe north-south temperature change gradient. In terms of mean summer precipitation theTD effect leads to a significant overall increase in precipitation all across Europe, which iscompensated and regionally reversed by the LR and CO effects in particular in southernEurope.
∗Climate Dynamics, 2016, doi:10.1007/s00382-016-3276-31Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland2now at: Federal Office of Meteorology and Climatology MeteoSwiss, Zurich, Switzerland3Center for Climate Systems Modeling (C2SM), ETH Zurich, Zurich, Switzerland
20 Chapter 3: Separating climate change signals
3.1 Introduction
The record-breaking summer heatwaves in 2003, 2006 and 2010 have raised interest ofthe scientific community in the European summer climate (Schar et al., 2004; Rebetezet al., 2009; Barriopedro et al., 2011). Observational studies show that European summertemperatures were rising in the last decades, especially in the upper percentiles (Della-Marta et al., 2007; Parey et al., 2010). Model simulations project that this trend willcontinue if greenhouse-gas concentrations continue to rise (Kjellstrom et al., 2007; Fischerand Schar, 2009, 2010; Orlowsky and Seneviratne, 2012; Kharin et al., 2013; Collins et al.,2013). Along with mean temperature changes, a strong drying for the Mediterraneanregion (Rowell and Jones 2006, ROW06 henceforth; Sanchez-Gomez et al. 2009; Kendonet al. 2010, KEN10 henceforth), changes of precipitation intensity and frequency (Rajczaket al., 2013; Ban et al., 2015) and increased temperature variability (Schar et al., 2004;Fischer et al., 2012; Cattiaux et al., 2015) are projected for many European regions. Thesechanges are expected to have strong socio-economic and ecological impacts across Europe(Field et al., 2012; IPCC, 2014).
Among the most peculiar characteristics of the European climate change signal is thepronounced north-south warming gradient, which includes a substantially stronger sum-mer warming and drying in the Mediterranean and Southern Europe in comparison toCentral and Northern Europe. This pattern is robust across the overwhelming major-ity of global climate simulations (IPCC, 2013). Indeed, unlike most other mid-latituderegions and other seasons which are projected to see a stronger warming at high lati-tudes (referred to as polar amplification) and an associated reduction of the troposphericnorth-south temperature contrast, the projected summer warming over Europe has theopposite gradient and exhibits a more pronounced warming in the south. This could alsohave strong implications for temperature variability and extreme events (Schneider et al.,2015). The reasons for this anomalous latitudinal warming pattern have been investigatedin a number of studies, but are still not fully understood. Some studies have invoked large-scale changes in circulation including an expansion of the Hadley cell and shift in stormtracks (Mbengue and Schneider, 2013; Collins et al., 2013), some have addressed the ther-modynamic effects of a generally warmer and moisture atmosphere (Rind et al., 1990;Manabe et al., 1992), some have pinpointed an enhanced land-sea contrast due to the de-layed oceanic warming (Manabe et al., 1992; Gregory et al., 1997; Rowell and Jones, 2006;Boe and Terray, 2014), and others have argued with regional land-atmosphere feedbackprocesses such as a stronger spring soil-moisture drying over Southern Europe and/or re-gional soil-moisture atmosphere temperature and precipitation feedback processes (Scharet al., 1999; Seneviratne et al., 2006a; Vidale et al., 2007; Boe and Terray, 2014).
In terms of their origins, climate change effects are often classified into thermodynam-ically and dynamically-driven features. Thermodynamic changes include the large-scalewarming of the atmosphere (Stocker et al., 2013) as well as an increase of specific humid-ity following the Clausius-Clapeyron relationship (Raval and Ramanthan, 1989; DelGenioet al., 1991; Allen and Ingram, 2002; Soden et al., 2005). These thermodynamic changesvary depending on latitude, longitude and height (Woollings, 2008; Collins et al., 2013).The dynamic changes manifest themselves in changes of the atmospheric circulation andare more uncertain than the thermodynamic ones (Hall, 2014; Shepherd, 2014). The mostrobust projected changes include a shift of the storm tracks and a general shift of the massdistribution in the atmosphere (Barnes and Polvani, 2013; Collins et al., 2013). Changes
3.1 Introduction 21
in temperature lapse-rates represent yet another factor, and these are usually categorizedwith either the thermodynamic or the dynamic effects. Models project that the warmingin the tropics (and to a lesser extent in the mid-latitudes) proceeds faster in the upperthan in the lower troposphere (Tett et al., 1996; Held and Soden, 2000; Bony et al., 2006).This behaviour is due to convection (as the moist adiabatic lapse-rate decreases withtemperature), and it implies a stabilization of the dry atmospheric stratification. Whilelapse-rate changes are broadly discussed in relation to the tropics, they have received onlylittle attention in the assessment of European climate change. Here we will address thisgap and try to quantify the role of lapse-rate changes for European climate change.
Several studies have tried to objectively discriminate between thermodynamic effects,dynamic effects and local feedbacks (e.g. ROW06; KEN10). Such efforts are compli-cated by the fact that all contributing factors are strongly interdependent. For instance,changes in the large-scale circulation may also cause changes in cloud cover and willtherefore have strong influence on the surface radiative balance (van Ulden et al., 2007).Likewise, a recent analysis of the EURO-CORDEX (Jacob et al., 2014) experiments forexample suggests that the occurrence of European summer heat waves results from com-plex interactions between soil moisture and large-scale circulation (Vautard et al., 2013).
In this study we exploit an extended version of the surrogate climate change or pseudo-warming methodology (SCC hereafter) to disentangle the different factors. An earlyversion of this approach was introduced by Schar et al. (1996). The SCC gives thepossibility to modify a given synoptic-scale atmospheric evolution by applying verticalwarming and moistening profiles. The method was used to study heavy precipitation overEurope at a time where no comprehensive ensembles of climate projections were available(Frei et al., 1998). SCC scenarios were further used to investigate the processes behindsummer drying in the mid-latitudes (Seneviratne et al. 2002, ROW06; KEN10). Im et al.(2010) used the method to quantify the role of thermodynamic feedbacks over the Alpineregion. A surrogate approach was also employed by Rasmussen et al. (2011) to studythe accelerated hydrological cycle over Colorado and by Sorland and Sorteberg (2015) tostudy monsoon low-pressure systems under climate change. This study will apply theSCC approach to systematically quantify the influence of individual large-scale driverson projected changes of the European summer climate. For this purpose, state-of-the-artclimate models are employed and the SCC is extended to a larger domain compared toprevious works and to climatological time scales.
Im et al. (2010) and Frei et al. (1998), for instance, used much smaller modelling do-mains covering only parts of the European continent. This study thereby complements theextensive works of ROW06 and KEN10 who also looked at the entire European continentover climatological time scales using one individual atmospheric model at a comparativelylow resolution. The generalisation of the results of such surrogate experiments is oftenlimited because only one specific climate model was used. Therefore it is of great interestto perform similar studies with different climate models and this study contributes to thisaim.
An very important methodological extensions with respect to previous work is, that wequantify the influence of large-scale lapse-rate changes by using the SCC methodology withdifferent vertical profiles. Also we extend the idea of ROW06 in which the climate changesignal is not only added to present-day conditions, but the different contributions are alsosystematically subtracted from the scenario experiment representing future conditions.This twin-design provides a very elegant way to separate the different effects.
22 Chapter 3: Separating climate change signals
The paper is structured as follows: In the second section a thorough explanation ofthe surrogate approach is given together with a description of the model setup. Section 3presents the results of the surrogate experiments including a validation of the approach.Section 4 discusses the results, before ending with conclusions and an outlook.
3.2 Methods
3.2.1 Introduction to surrogate climate change methodology
The basic idea of the surrogate approach is to apply a large-scale temperature change tothe lateral boundary conditions of a present-day reference simulation of a regional climatemodel (RCM), while maintaining the relative humidity and thus increasing the specificmoisture content. The relative humidity is held constant because observations and modelstudies suggest it would behave like this in a warmer climate (Cess et al., 1990; Ross et al.,2002; Allen and Ingram, 2002; Soden et al., 2005). The mean temperature perturbationis taken from a mean climate change signal of the driving GCM run. The resultingsurrogate climate follows the large-scale circulation of the reference period but with awarmer and more humid climate. In principle the approach allows to study the regionaland local climate response and feedbacks in response to a thermodynamic change of theatmosphere, whereas the more uncertain dynamic response to climate change is excluded.
Because of dynamical constraints, the specified temperature change cannot be chosenarbitrarily. Schar et al. (1996) showed that by using ∆T = ∆T (p), i.e. the temperaturechange is a function of pressure only, the equations of motions are not altered by thismanipulation, and the flow remains unchanged except for implied changes and feedbacks,such as latent heating or changes in cloud cover. It is important to note that changes inhorizontal temperature gradients are not included, because following the adjustment ofthe thermal wind relation the large-scale circulation would be altered. Structures thatentail changes in the horizontal temperature gradients are thus considered to be part ofthe dynamic rather than thermodynamic changes.
Concerning the vertical dependency of ∆T (p), previous studies have shown that warm-ing signals vary with height and are stronger at the tropopause than on the ground (Sodenet al., 2005; Santer et al., 2008; Collins et al., 2013; Vallis et al., 2015). This is true forthe tropics up to latitudes of 60◦ north. Nevertheless, some of the early surrogate stud-ies mentioned before used a vertically constant temperature perturbation (Schar et al.,1996; Seneviratne et al., 2002; Im et al., 2010), ignoring the lapse-rate change, whereasother studies accounted for it (ROW06; KEN10). However, none of these studies actuallytested whether this lapse-rate change has an important influence on the European climatechange. In this study we will exploit a general form of the surrogate framework. First,consideration will be given to a height-dependent temperature perturbation, in order toassess the importance of lapse-rate effects for the projected climate change signal. Sec-ond, in addition to warming experiments we will also employ experiments that add somecooling and drying to scenario experiments. As will be seen, the combination of warmingand cooling experiments yields some validation of the methodology.
3.2 Methods 23
(a) (b)
Figure 3.1: (a) Mean annual cycle and (b) vertical structure of the warming signal SCEN-CTRL. The two cases ∆T = const and ∆T = ∆T (z) are shown by black and blue lines, respec-tively.
3.2.2 Surrogate setup
Our ensemble of surrogate experiments consists of six members. Two of them are ref-erence runs CTRL and SCEN, in which the regional climate model CCLM (see Section2.4) is nested into the GCM MPI-ESM-LR in conventional downscaling mode. The fourother members are surrogate experiments. In order to make our surrogate experimentscomparable to the reference runs, we calculate ∆T (p) as the area mean climate changesignal extracted from the reference simulations:
∆T = TSCEN − TCTRL (3.1)
In a first step we omit the height dependence of the warming (i.e., we do not considerlarge-scale lapse-rate changes) and choose the warming at 850hPa as representative for thewhole atmosphere. In a second step we consider ∆T to be a function of height (as definedby the vertical model levels), explicitly accounting for changes in the large-scale lapse-rate. As the model considered doesn’t use pressure coordinates, we used ∆T (z) insteadof ∆T (p), but the associated violation of mass balance is negligible. Figure 3.1b shows∆T (z) for both approaches. The variability of ∆T with the seasonal cycle is accountedfor by calculating a smoothed mean annual cycle of ∆T (Figure 1a). As a smoothingalgorithm a spectral filter was used because of its superior performance compared to amoving mean (Bosshard et al., 2011). For ∆T (z) the smoothing was done on every levelseparately. On the 13 model level corresponding roughly to the 850 hPa level the sameannual cycle is used for both cases, i.e. for ∆T = const and ∆T = ∆T (z). The horizontalaveraging is performed over the whole computational domain.
In practical terms, the approach works as follows: At all grid points and levels of theatmospheric boundary data, ∆T is added and the specific moisture content is increasedaccording to the Clausius Clapyeron relation. To maintain physical consistency also thepressure on every model level has to be adapted. This is achieved by integrating thehydrostatic balance
dp(z)
dz= −g ∗ p(z)
R ∗ T (z)(3.2)
24 Chapter 3: Separating climate change signals
numerically from the bottom to the top of the atmosphere, using the surface pressurefrom the parent simulation (CTRL or SCEN, respectively).
All other variables in the lateral boundary data relaxation zone, in particular thehorizontal wind, are used without changes. To keep the atmosphere at the boundariesin balance with the ocean, the sea surface temperatures (SSTs) are increased by thesame ∆T as the lowest atmospheric level. The same procedure was also applied bySchar et al. (1996) and Im et al. (2010). The soil moisture is left unchanged and canevolve freely during the simulation, as the present study was not aiming at separating theinfluence of soil moisture feedbacks on climate change. The atmospheric concentrationsof CO2 and further greenhouse gases are taken from the driving scenario representingRCP8.5. For all simulations (CTRL, SCEN and surrogate) the aerosol climatology fromTanre et al. (1984) is used. Through this method ∆T was added to the lateral boundaryconditions of the CTRL period, but also subtracted from lateral boundary conditions ofthe SCEN period (twin-design). This provides a future circulation in a climate with thesame mean temperature as today and, hence, an independent estimate of the large-scalethermodynamic effect. A similar approach was used by ROW06 and KEN10.
The four surrogate experiments are shown in Figure 3.2 and referred to as homogeneouswarming (HW; applied to CTRL) and homogeneous cooling (HC; applied to SCEN) andthe same for the vertically-dependent approach with vertically-dependent warming (VW;applied to CTRL) and vertically-dependent cooling (VC; applied to SCEN).
Figure 3.2: Schematic of the six simulations conducted. CTRL and SCEN refer to the tworeference simulations for the periods 1971-2000 and 2070-2099, respectively. The four surrogateexperiments are homogeneous warming (HW), homogeneous cooling (HC), vertically-dependentwarming (VW) and vertically-dependent cooling (VC). Blue and red colors refer to present-dayand future temperatures, respectively, whereas the grey scale encodes the state of the circulation:Light gray for present-day circulation and dark grey future circulation. The arrows representhow the surrogate experiments were constructed.
3.2 Methods 25
3.2.3 Arithmetics of the surrogate approach
The numerical experiments defined above enable the separation of the climate changesignal into thermodynamic, lapse-rate and circulation plus other remaining effects. Herewe provide the respective definitions. A validation of the associated interpretations willfollow a posteriori.
Large-scale thermodynamic effect (TD): This effect is meant to represent changesthat are related to the large-scale homogeneous warming of the atmosphere and the ac-companying increase of specific moisture, also including the direct radiative effect of futurechanges in CO2 and other greenhouse gases. Due to the twin design, there are two waysto quantify this effect. It can either be calculated as the difference HW−CTRL, or as thedifference SCEN−HC. The temperature perturbations applied in these two experimentsare the same but with opposite sign. Later we will investigate these two estimates andfind that they are indeed similar. Following this reasoning, we define the thermodynamiceffect as the average of the two estimates, i.e.
TD :=(HW − CTRL) + (SCEN −HC)
2(3.3)
Lapse-rate effect (LR): The lapse-rate effect is defined as the difference of the twosurrogate approaches with homogeneous and vertically-dependent warming or cooling. Inthe homogeneous surrogate experiments, the mean change in the lapse-rate is excluded,whereas it is accounted for in the vertically-dependent experiments. There are again twooptions to diagnose the effect, either as VW −HW or as HC − V C, and following (3.3)we define it as the average:
LR :=(VW −HW ) + (HC − V C)
2(3.4)
Large-scale circulation and other remaining effects (CO): This effect is in-directly defined by the large-scale atmospheric and ocean features not covered in thevertically dependent surrogate experiments, i.e., as the residual between a full and a sur-rogate scenario. Note that thereby it is not possible to separate a circulation response toexternal forcing from internal variability. As above, there are two ways to calculate theCO, and the average of the two effects is considered:
CO :=(SCEN − VW ) + (V C − CTRL)
2(3.5)
Full climate change (FCC): Adding up the three contributions as defined aboveimplies a large number of cancellations and yields:
SCEN − CTRL = TD + LR + CO (3.6)
This is one of the key results of this study. It demonstrates that the climate changesignal SCEN −CTRL as simulated in the full downscaling mode can be split into large-scale thermodynamic (TD), lapse-rate (LR), and circulation plus other remaining (CO)
26 Chapter 3: Separating climate change signals
effects. Within the framework considered, the splitting is exact by design and does notinvolve any approximations. This attractive property is achieved by using both coolingand warming surrogate experiments. This set-up is responsible for the cancellations thatultimately yield the simple result (3.6).
It is important to note that the plausibility of the above definitions needs to be checkeda posteriori. In practice we will do so by comparing the two possible definitions. Forinstance, the identification of LR as a lapse-rate effect is only justified provided the twodefinitions, i.e. VW−HW and HC−V C yield a qualitatively similar estimate. If the twodefinitions were to differ substantially, it would mean that a simple additive separationinto the three effects as in (3.6) is not feasible.
3.2.4 Model and setup
We here use the regional climate model COSMO-CLM (Consortium for Small-Scale Mod-elling in Climate Mode), hereafter referred to as CCLM. CCLM is a non-hydrostaticand fully-compressible regional climate model, based on the numerical weather predic-tion model COSMO (Baldauf et al., 2011). Here we use model version 4.8 clm17 (Rockelet al., 2008), which has also been heavily tested and applied in the framework of EURO-CORDEX. A more detailed description of the model and its parameterization schemescan be found in Vautard et al. (2013) and Kotlarski et al. (2014). We do not presenta validation for our CTRL simulations, since the good performance of CCLM over theEuropean continent was demonstrated in previous studies (Jaeger et al., 2008; Bellpratet al., 2012; Montesarchio et al., 2014; Ban et al., 2014; Kotlarski et al., 2014). TheCCLM model domain corresponds to the EURO-CORDEX domain. Using a rotatedlatitude-longitude grid, it approximately covers the area from 27N-72N, and 33W-45E,which covers the larger European area including the North-Atlantic, parts of Greenlandand Northern Africa. The resolution is 0.44 degrees (50 km).
In all experiments CCLM is driven by the global climate model (GCM) MPI-ESM-LR(Stevens et al. 2013) at the lateral boundaries. The driving GCM is forced with historicalradiative forcing until 2005, and the Representative Concentration Pathway (RCP) 8.5thereafter (Moss et al., 2010). To maximize the signal to noise ratio, our analysis focuseson the time slices 1971-2000 and 2070-2099. These 30-year-long simulation periods arereferred to as CTRL and SCEN, respectively. The two CTRL and SCEN simulationsare standard CORDEX experiments and part of a transient simulation covering the timeperiod 1950-2100. The other simulations, i.e. HW, VW, HC and VC, are 31-year longtime slices specifically conducted for the current paper. These simulations use a modifiedrestart file of the respective reference simulations as initial conditions. As mentionedabove the soil moisture is not changed, but the soil temperatures are adjusted by a ∆Trepresenting the actual difference between the two start dates. The first year is than usedas spin up.
3.3 Results
In the following section the influences of large-scale thermodynamic, lapse-rate and large-scale circulation changes plus other remaining changes on the mean European summer
3.3 Results 27
climate are presented. Figure 3.3 shows the main results which will be presented in thefollowing subsections.
3.3.1 The thermodynamic effect
The large-scale thermodynamic effect leads to a rather homogeneous warming over theEuropean continent (Fig. 3.3a). Over eastern Europe the warming signal is weaker,whereas parts of northern Africa show a slightly stronger temperature increase than thedomain mean (Fig. 3.3b). In general warming over sea surfaces reflects the prescribedboundary forcing of 4.1◦C and is stronger than over land. It seems that there is a process inplace which dampens the warming applied at the boundaries over the land areas (see endof this section). Precipitation is increasing over most of Europe (Fig. 3.3c). The strongestincrease is seen at the western boundaries and over the North Atlantic. While absoluteprecipitation changes over the Mediterranean are small, relative changes are substantialbut somewhat poorly defined due to the small reference precipitation amounts. Thelarge-scale precipitation increase over Europe relates to the prescribed increase of specifichumidity q at the lateral boundaries. The same behaviour was also found by KEN10.
The general response in the thermodynamic surrogate experiment is similar to pre-vious experiments. Like Seneviratne et al. (2002) we see a strong increase in downwardlongwave radiation at the surface, which is primarily associated with the strong increaseof atmospheric water vapour and increased tropospheric temperatures. Incoming surfaceshortwave radiation is strongly reduced due to increased cloud cover and evaporation be-cause of higher atmospheric demand. Overall it is evident that a thermodynamic changeof the atmosphere alone cannot reproduce the distinct change patterns of the referencescenario (Fig. 3.3j-l). There is no summer drying in southern Europe, and no north-southgradient in temperature change (Fig. 3.3a-c). Thus the distinct spatial patterns projectedin the full climate change scenario must be caused by other mechanisms, such as large-scale circulation changes, lapse-rate changes, and/or effects due to land-ocean contrast.The TD effect also contains the influence of changes in CO2 and other greenhouse gasconcentrations the influence of these changes is discussed in sections 3.4.3.
For the TD effect the imposed temperature change signal is obviously dampened overthe continent (Fig. 3.3b). The responsible process can be identified by looking at thevertical change profile which is prescribed at the boundaries in comparison to the onesimulated in the model domain. Figure 3.4a-c) show that the vertically homogeneouschange profile is not preserved in the domain but also tends to become stratified with astronger warming higher up and a weaker warming at the ground. Only profiles for theregions Mid-Europe (ME), Eastern Europe (EA) and France (FR) (see e.g. Christensenand Christensen 2007) are shown but this result is also true for the other regions. Thestratification increase can be explained by a strong increase in convective activity for theTD which leads to an increase of latent heating in the higher troposphere and a coolingat the ground. Figure 3.4d shows that the share of convective precipitation on totalprecipitation increases strongly.
3.3.2 The lapse-rate effect
Comparing the results of the vertically homogeneous and the vertically dependent sur-rogate experiments (see eq.3.4) allows us to quantify the effect of a lapse-rate change on
28 Chapter 3: Separating climate change signals
Figure 3.3: Separation of the JJA climate change signal. The first three rows show (a-c)the large scale thermodynamic effect (TD) (d-f) the lapse-rate effect (LR) and (g-i) the large-scale circulation plus other remaining effects (CO), respectively. The fourth row shows the fullclimate change signal (FCC = SCEN −CTRL). The left and the middle column show changesin 2m seasonal mean temperature, absolute and as anomalies to the domain mean respectively.The right column shows relative changes in total precipitation (in the southern regions overthe Sahara the absolute precipitation amounts are very low (< 0.2mm/day) therefore relativechanges can become very high and are not robust).
3.3 Results 29
Figure 3.4: TD effect on convective precipitation. (a-c) Vertical profile of temperature changefor (a) Mid-Europe (ME), (b) Eastern Europe (EA) and (c) France (FR) (see e.g. Christensenand Christensen 2007). The solid line represents the actual temperature change simulated overthe region, the dashed line represents the change profile prescribed at the lateral boundariesin the homogeneous warming case, (d) relative change in the ratio of convective to grid scaleprecipitation in comparison to CTRL for the same three regions ME, EA and FR. The meanchange in the region is shown as red line, the spatial variability is represented, by the boxescontaining all values between the 0.25 and 0.75 percentile and by the whiskers containing allvalues between 0.05 and 0.95 percentile.
future temperature (Fig. 3.3d-e) and precipitation (Fig. 3.3f) changes, respectively. Theresults will also be compared with the results of ROW06 and KEN10, who also used avertically-dependent surrogate approach. The temperature change pattern shows a clearnorth-south gradient with an amplified warming in the Mediterranean region and easternEurope. ROW06 found a similar pattern with an amplification of the warming in theMediterranean region. In our case the amplification is especially strong over the IberianPeninsula and western North Africa around the Atlas mountains (Fig 3.3e). The simu-lated temperature change also exhibits some land-sea contrast, which is mainly caused bydifferences in the prescribed atmospheric and sea-surface ∆T (Fig. 3.3d-e). Recall thatthe SSTs have the same ∆T as the lowest atmospheric level and that the homogeneousand vertically dependent experiments have a different warming at these levels (Fig 3.1b).
Also the precipitation signal shows a distinct north-south gradient (Fig. 3.3f). Unlikethe thermodynamic response TD which showed precipitation increases almost throughoutthe domain, the lapse-rate effect causes a reduction in precipitation over most of Europe,and in particular over the Mediterranean. The Mediterranean drying is further associatedwith a decrease in soil moisture and evaporative fraction, along with reduced cloud coverand increased short wave radiation for the warming regions (not shown). In comparison toROW06 and KEN10 the amplitude of the precipitation decrease over continental Europe(41N-50N,0E-31E) is much weaker. ROW06 estimated that their warming effect (whichincludes also some circulation effects and the lapse-rate changes) accounts for 42.5% of
30 Chapter 3: Separating climate change signals
the precipitation decrease seen in the full climate change signal for this area. Over thesame area the lapse-rate effect in our experiments accounts for 12.5% of the precipitationdecrease.Overall the results show that the lapse-rate effect can reproduce parts of the distinctchange pattern as projected by the full climate change simulations. In particular, anorth-south temperature change gradient is simulated, which accounts for about 30% ofthe total change in gradient. Over the Iberian peninsula the lapse-rate effect accountsfor more than half of the warming seen in the full climate change signal (Fig. 3.3j).This shows that lapse-rate changes can significantly alter the climate change signal overEurope.
3.3.3 The circulation and other remaining effects
The combined effect of the large-scale circulation change and other remaining changesis shown in Figure 3.3g-i). The temperature changes show a strong land-ocean contrastas well as a strong north-south gradient. The Atlantic, especially in the north, shows aweaker warming than the Mediterranean. This SST pattern is part of CO by design. TheMediterranean land areas show a clear amplification of the warming. Part of this warmingamplification reflects the northward shift of the storm tracks predicted by most GCMs(Collins et al., 2013), which brings warmer air to southern Europe. Also precipitationchanges show a distinct north-south pattern. The general effect of CO is a decreasein precipitation over most of Europe, with strong reductions in Southern Europe andslight increases in Scandinavia (Fig. 3.3i). ROW06 and KEN10 isolated a European-scale reduction in precipitation due to circulation changes. For continental Europe theyreport circulation changes accounting for less than 10% of the precipitation decrease.We find that the combined effect of changes in circulation and the land-ocean contrastaccounts for more than 80% of the full change. The difference between the two studieshighlights an important role of the modified land-ocean contrast: Because the Atlanticwarms substantially slower than the continent, air that is advected from sea to landexperiences considerable warming and drying. This process evidently explains a significantfraction of the reduction in southern European precipitation.
3.3.4 Verification of the approach
The applied surrogate approach relies on two main assumptions, which need to be verified.First, the approach assumes that the changes in the boundary conditions are imposed ontothe RCM in such a way that the large-scale circulation does not significantly change. Infigure 3.5b) it can be seen that the differences in the mean geopotential height at 500hPaare small between the VW experiment and its reference simulation CTRL, and the sameis true for the HW experiment (Fig. 3.5c). The change from CTRL to the future circu-lation of SCEN is much stronger (Fig. 3.5a). In figure 3.5a) 50% of the points show asignificant difference, whereas in Figs. 3.5b-c) only around 5% of the points are signifi-cant. In addition, the differences in figures 3.5b-c) exhibit a large-scale pattern that canbe linked to large-scale differences in low-level atmospheric temperature. This is also truefor the difference in mean geopotential height between SCEN and the cooling surrogateexperiments HC and VC (not shown). These results suggest that the large-scale circula-tion is only slightly altered by the thermodynamic changes to the boundaries. It confirms
3.4 Discussion 31
Figure 3.5: Differences in 500hPa geopotential height anomalies; Gridpoints where the differ-ence is significant at the 5% level are shaded. Panels show (a) SCEN-CTRL, (b) VW-CTRL,(c) HW-CTRL
our assumption of largely unchanged flow condition between references simulation andsurrogate experiment.
Second, the surrogate approach assumes that TD, LR and CO are separable and noimportant interactions occur. Note, however, that the averaged terms of the twin design(Equations 3.3 to 3.5) cannot be used to verify this assumption, as combining theseequations exactly yields the SCEN-CTRL climate change signal by design. As mentionedin section 3.2.3 there is another way to verify the approach: The two possibilities toquantify each individual effect should lead to similar results, that is:
TD ≈ HW − CTRL ≈ SCEN −HC (3.7)
LR ≈ VW −HW ≈ HC − V C (3.8)
CO ≈ SCEN − VW ≈ V C − CTRL (3.9)
Figure 3.6) shows both options of (3.7-3.9) for 2m temperatures, and figure 3.7) for pre-cipitation changes. For temperature the spatial change patterns are very similar (spatialcorrelation of at least 0.95) but the amplitude differs slightly (Fig. 3.6). Over most partsof Europe the lapse-rate effect is stronger for HC-VC in comparison to VW-HW (Fig.3.6c-d), suggesting that the lapse-rate effect is stronger for a future circulation. For totalprecipitation the spatial correlation is smaller but still acceptable (at least 0.85) (Fig.3.7). In general the analysis yields a good agreement between the two ways of quantifyingthe three effects. This is true for JJA, but also for the other seasons (not shown). Theseresults indicate that interactions are overall too weak to distort the separation of theeffects with the surrogate method, and that the method - as such - is well applicable.
3.4 Discussion
3.4.1 Interpretation of the lapse-rate effect and the Mediter-ranean warming amplification
Concerning the amplification of the warming in the Mediterranean region, ROW06 founda similar pattern as in the current study (Fig. 3.3d). They argue that the extra warming
32 Chapter 3: Separating climate change signals
Figure 3.6: Changes in 2m temperature for JJA; the two columns represent the two differentways to calculate the different effects according to (3.7-3.9). The first row shows the thermody-namic effect (as anomalies to the domain mean change) as (a) HW-CTRL and (b) SCEN-HC,the second row the lapse-rate effect as (c) VW-HW und (d) HC-VC, and the third row thelarge-scale circulation plus other remaining effects as (e) SCEN-VW and (f) VC-CTRL.
3.4 Discussion 33
Figure 3.7: Same as in Figure 3.6, but for relative changes in precipitation.
34 Chapter 3: Separating climate change signals
over the Mediterranean is caused by less advection of cold maritime air, which increasesthe influence of radiative effects. In our study we see a similar behaviour, but we canattribute these changes to the lapse-rate effect, which was implicitly also included byROW06.
The role of the lapse-rate effect is consistent with the existing literature. It is wellunderstood that the projected large-scale warming and moistening will lead to increasesin the moist-adiabatic lapse rate, which in turn implies changes in stratification. This iscritical as theoretical and modelling studies suggest that large-scale stratification deter-mines the extent of the Hadley circulation. In particular the Hadley circulation is believedto extend polewards up to the latitude where the vertical shear becomes baroclinicallyunstable (Lu et al., 2007; Levine and Schneider, 2015), and thus is expected to expandin response to increasing stratification and an increasing depth of the extratropical tro-posphere. As a result of the extension, the Mediterranean and Southern Europe wouldbecome more strongly affected by subsidence in the descending branch of the Hadley cir-culation, which would tend to suppress convection and lead to less precipitation, strongerinversions, and warmer surface temperatures (in response to decreased vertical exchange).
Studies in the literature also show that there is a further amplification by positive feed-backs that involve reduced relative humidity, decreased cloud cover, increased incomingshort-wave radiation, ampflified subsidence, and reduced soil moisture.
The role of the ensuing feedback processes is apparent from studies of European heatwaves. In particular, the linkage between soil moisture deficits and heatwaves has beenwell established (e.g. Fischer et al. 2007), and also a strengthening of inversions due toincreased subsidence has been documented (Miralles et al., 2014). The current studyincludes these amplification effects and raises attention to the role of lapse-rate changes,but we do not attempt to quantify the ensuing feedback processes.
Overall our results show that including the lapse-rate change into the thermodynamiceffect more realistically replicates the behaviour seen in scenario simulations. For furthersurrogate experiments it is hence an important decision whether to include the lapse-ratechange or not.
In the next subsections we provide further analysis and investigate additional factorsthat may contribute to the Mediterranean drying.
3.4.2 Mediterranean drying and the land-ocean contrast
In section 3.3.2 it was shown that the lapse-rate effect yields a distinct drying and warm-ing pattern over the Mediterranean. We define the Mediterranean area to include thetwo regions Iberian Peninsula (IP) and Mediterranean (MD) (see Christensen and Chris-tensen 2007). When comparing our results to existing studies, certain differences arefound that partly relate to the setup. ROW06 and KEN10 identified, what they call thethermodynamic effect to be dominating the Mediterranean summer drying. In contrastwe identified the large-scale circulation plus other remaining effects to be the dominatingfactor (Fig. 3.3i). A fully quantitative comparison is not possible as no domain averagesare given by ROW06 and KEN10 beside for an area in continental Europe by ROW06and we are restricted to qualitatively compare area plots. As alluded to above, besidesimilarities in the design (all three studies use a vertically dependent ∆T ), there are alsoimportant differences in the surrogate setup, regarding how the lateral and lower bound-ary conditions are imposed. In the design of ROW06 and KEN10, changes in land-ocean
3.4 Discussion 35
temperature contrast as well as horizontal temperature gradients were included into thelarge-scale thermodynamic change effect. The SSTs in the surrogate experiment wheretaken from the scenario and ∆T was not just a function of height but also of longitude andlatitude ∆T (x, y, z). In the present study the land-sea contrast as well as the horizontaltemperature gradient are not included in the large-scale thermodynamic effect. ∆T is afunction of height only and the SSTs are raised by the same amount as the temperatureon the lowest atmospheric level.
Especially the difference in the treatment of the land-ocean warming contrast ap-pears important. In the literature there is a growing number of studies which arguethat the land-ocean warming contrast has an important contribution to the projectedMediterranean drying (Fasullo, 2010; Boe and Terray, 2014; Sherwood and Fu, 2014).The differences between the present study and ROW06 and KEN10 do not contradictthis hypothesis. In the present work the land-ocean warming contrast is not included inthe large-scale thermodynamic effect (TD), and as a consequence the associated Mediter-ranean drying is much weaker than in the study of ROW06 and KEN10, who includedthis feature in their thermodynamic effect. In the present study the land-ocean warmingcontrast is part of the large-scale circulation and other remaining effects (CO). Hence theassociated simulations show a strong drying over the Mediterranean as well as the CentralEuropean region (Fig. 3.3i). This confirms that the land-ocean warming contrast is likelyan important factor for the Mediterranean drying. The question remains whether thegeneral land-ocean warming contrast or the specific geographical pattern of SST changes(less warming in the North Atlantic and stronger warming in the Mediterranean) is themore important factor.
To investigate this question, two additional surrogate experiments were carried out.One is based on the VW experiment (VW SST) one on the VC experiment (VC SST).In these experiments the SSTs are not increased/decreased by the same amount as thelowest atmospheric level, but by the domain mean SST difference between SCEN andCTRL. This translates to a ∼1K smaller SST increase/decrease for JJA and a strongerland-ocean warming contrast. The lateral boundary variables are exactly the same as inVW or VC respectively. With those two additional surrogate experiments the effect of amean land-ocean contrast (LOC) can be calculated in the same fashion as the other threeeffects:
LOC :=(VW SST − VW ) + (V C − V C SST )
2(3.10)
Figure 3.8 shows that the LOC is mainly decreasing temperature through out thedomain but has almost no influence on the spatial change pattern and can not explainthe Mediterranean warming amplification. In contrast to temperature the influence onprecipitation especially over the Mediterranean is stronger although not as strong as forthe lapse-rate effect. Nevertheless one can conclude that the mean land-ocean contrastcontributes to the Mediterranean drying. We argued before that the land-ocean contrastcould explain part of the contrasting results between our study and ROW06 but thequestion remains whether the general land-ocean warming (tested in Fig.3.8) or spatialSST changes are more important. The presented results suggests that the spatial changepattern in the SST (i.e. the warming minima over the North Atlantic) makes an importantcontribution to the Mediterranean drying. However, the influence of these changes is more
36 Chapter 3: Separating climate change signals
difficult to quantify because they are coupled to changes in the atmospheric circulationvia the thermal wind relation.
Another important conceptual difference in the setup concerns soil moisture effects.ROW06 and KEN10 conducted specific experiments to test the role of spring and summersoil moisture feedbacks. In those experiments they showed that soil moisture changes area significant component of the reduction in summer precipitation. We did not conductsuch experiments, but use the soil-moisture to diagnose the Mediterranean drying. To thisend, Figure 3.9) shows the JJA soil moisture signal for our experiments. The figure showsthat most of the summer soil moisture change shows up in the large-scale circulation andother remaining effects (CO), with some minor imprints also in the lapse-rate experiment(LR). This shows that in our setting the soil moisture effect on summer drying is implicitlyincluded in the CO effect. Note that additional experiment would be needed to furtherdisentangle the regional feedbacks caused by the CO effect.
3.4.3 Role of greenhouse gases
The TD effect also contains the influence of changes in CO2 and other greenhouse gasconcentrations following the RCP8.5 scenario. To quantify the influence of these concen-tration changes we performed an additional vertical warming experiment (VW GHG) inwhich atmospheric greenhouse gas concentrations are prescribed according to the CTRLsimulation. Influence of CO2 and other greenhouse gases is than calculated as the differ-ence between VW and (VW GHG). Figure 3.10 shows that the influence of these changeson our regional climate simulation are very weak. These results confirm the findings ofearlier surrogate experiments (Seneviratne et al., 2002).
3.4.4 Assessing projection uncertainties
Projected changes in the atmospheric circulation in response to greenhouse gas forcingare considered rather uncertain (Christensen et al., 2013; Shepherd, 2014; Hall, 2014).Therefore the discrimination between thermodynamic and dynamic effects is often usedto discriminate between more and less reliable aspects of the projections.
The surrogate approach offers a direct possibility to address this discrimination incomparison to other more statistical analyses, such as studies using climate analogues(Cattiaux et al., 2013a). The surrogate approach provides an opportunity to separatethermodynamic (TD), lapse-rate (LR) and circulation plus other remaining (CO) effects,and thereby assess projection uncertainties in climate change scenarios. A problem of thisapproach is that the CO effect combines effects with different associated uncertainties, thetwo most important are the large-scale circulation changes and the land-ocean contrast.The change in large-scale circulation as already mentioned above is regarded as ratheruncertain, whereas the general land-ocean warming contrast is a very robust feature ofclimate change projections (Joshi et al., 2008; Lambert et al., 2011). However, in section3.4.2 we show that the general land-ocean contrast is not such an important driver whichsuggests that the spatial change pattern in the SST is very important and this pattern aswell as the associated circulation changes are more uncertain. These arguments justify toassess the CO effect as more uncertain than the TD and LR effects.
3.4 Discussion 37
Figure 3.8: Effect of the land-ocean contrast (LOC) on (a) 2m temperature, (b) total precip-itation (%) and (c) soil moisture (%). Soil moisture integrated over the seven topmost layers(corresponding to the top 1.5 m of the soil).
38 Chapter 3: Separating climate change signals
Figure 3.9: Same as in Figure 3.3, but for relative changes in soil moisture integrated over theseven topmost layers (corresponding to the top 1.5 m of the soil).
3.4 Discussion 39
Figure 3.10: Effect of CO2 and other greenhouses gas concentrations on surrogate simulationsfor the summer season JJA. Calculated as the difference between the vertical warming experiment(VW) with a greenhouse gas forcing according to the RCP8.5 scenario and a VW experimentwith historical greenhouse gas forcing. (a) 2m temperature, (b) total precipitation %. Note that(a) uses a finer colour scale than other figures.
40 Chapter 3: Separating climate change signals
Our results (see Fig. 3.3) show that the general warming is driven primarily by ther-modynamic effects (TD). Still large parts of the spatial temperature change pattern arestrongly influenced by the CO and the LR.
Following the argumentation of Hawkins and Sutton (2011) and Shepherd (2014),precipitation is strongly influenced by dynamic and thermodynamic effects, and the de-pendence upon dynamic effects thus implies considerable uncertainties. Studies of largeensembles, for example Cattiaux et al. (2013a), show that the precipitation signal forthe summer season over mid Europe is strongly depending on the GCM considered. Thetransition line between precipitation increases in the north and decreases in the southdepends significantly on the selected model.
Our study indicates that this uncertainty in response pattern of precipitation resultsfrom the opposing influence of thermodynamic, lapse-rate and dynamic effects. In North-ern Europe, TD and CO effects lead to an increase in precipitation, while in SouthernEurope LR and CO (and to a lesser extent TD) effects lead to a precipitation decrease,whereas in Central Europe TD and CO have opposing signs. Our study adds LR effectsto this discussion. In particular, we consider the LR effects as less uncertain than the COeffects, as the temperature-dependence of the moist adiabatic lapse-rate can be considereda fundamental property. That said, we are not aware of a detailed analysis of changes inlapse-rates over Europe in the CMIP5 ensemble.
3.4.5 Limitations
Our study has a number of limitations. First of all, we considered only one realizationof one RCM and its driving GCM. To some extent this limitation is counteracted bythe special twin-design, which allows assessing the robustness of the effects relative totwo different background circulations. In addition, we did not aim at separating thelarge-scale circulation changes in a forced response of circulation and internal variability.Internal variability has been found to play an important role for trends in circulationeven at multi-decadal time scales. Note however, that the role of internal variability isparticularly dominant in near-term climate change whereas we here look at long-termchanges by the end of the century (Hawkins and Sutton, 2011).
3.5 Conclusions and Outlook
In this study the surrogate (or pseudo-warming) approach was extended in a European-scale study. The primary research question was to quantify the contributions of large-scalethermodynamic, lapse-rate and dynamical changes to the European summer climate.
We have applied both warming and cooling profiles to the CTRL and SCEN simu-lations, respectively. Results show that this twin design allows for a clear separation ofthree main effects: large-scale thermodynamic, lapse-rate and dynamical effects. Morespecifically, we find that applying a warming to CTRL has a similar effect as applying acooling to SCEN. This demonstrates that the basic response to large-scale climate changesis not overly dependent upon the reference climatology.
We find that the thermodynamic effect alone, which manifests itself in an overallwarmer and moister summer climate, cannot reproduce the typical north-south warminggradient and Mediterranean drying of climate simulations. When accounting for a lapse-
3.5 Conclusions and Outlook 41
rate change some fraction of the typical southern European summer drying and amplifiedwarming can be reproduced, even though the prescribed lapse-rate change is horizontallyhomogeneous. About half of the amplified warming over the Iberian Peninsula can beexplained by the lapse-rate effect. The influence of the lapse-rate effect is linked to anextending Hadley circulation. Large-scale circulation changes and other remaining effectsmake a further important contribution to the Mediterranean amplification, i.e. theyamplify the warming and drying over southern Europe.
After establishing the current methodology for mean climate changes, it would be ofinterest to consider changes in temperature variability and climate extremes. Especiallyfor those phenomena, the possibility to disentangle thermodynamic, lapse-rate and dy-namic effects could give valuable insights into the processes dominating the changes inhigher order statistics. For a better estimate of the reliability of the described effects, asimilar study with a different RCM would be very helpful.
Acknowledgements The COSMO-CLM simulations studied in this work were car-ried out at the Swiss National Supercomputing Centre (CSCS). We are thankful to theCOSMO and CLM communities who maintain and develop the model. We also wantto acknowledge the work of Stauffer et al. (2015) who provided the tool to generate theapplied colour scale. We thank Tapio Schneider for his helpful comments on this workand its relation to the Hadley circulation. We also want to thank the two anonymousreviewers for their constructive and useful comments.
Chapter 4
Quantifying the drivers of Europeanprecipitation changes:Large-scalethermodynamics, lapse-rate andcirculation changes
43
45
Quantifying the drivers of European precipitation changes:Large-scale thermodynamics, lapse-rate and circulation changes∗
Nico Kroner1, Daniel Luthi1, Erich Fischer1, Sven Kotlarski 1,2,and Christoph Schar1
Abstract
The large-scale driver for changes in European precipitation statistics (mean, inten-sity, frequency, heavy precipitation and dry days) are examined. Regional climate modelensembles suggest a bipolar climate change pattern over Europe, with decreasing (increas-ing) mean precipitation and wet-day frequency in the south (north). Increases in intensityand heavy precipitation show a similar pattern but the increases extend further south. Anextended surrogate or pseudo-global warming approach is applied to disentangle the influ-ence of large-scale thermodynamic, circulation and lapse-rate changes on the projections.Additionally a multi-model ensemble (EURO-CORDEX) is utilized to evaluate the find-ings. The thermodynamic effect is found to increase precipitation intensity, but to haveno influence on the precipitation frequency. Its influence on heavy precipitation events isstronger than on mean precipitation. The large-scale circulation in contrast is decreasingthe precipitation frequency and has only a small influence on precipitation intensity. Ingeneral its influence becomes weaker for heavy precipitation events. The lapse-rate ef-fect is important in summer over southern Europe. For this region and season, its effectis as strong as the large-scale circulation effect, and it is also decreasing precipitationfrequency. The strong influence of the lapse-rate effect on Mediterranean precipitationchange is quantified for the first time in this study. It is shown that the lapse-rate effectexplains about half of the Mediterranean drying in a multi-model ensemble.
∗in prep.1Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland2now at: Federal Office of Meteorology and Climatology MeteoSwiss, Zurich, Switzerland
46 Chapter 4: Quantifying drivers of European precipitation changes
4.1 Introduction
Along with increasing global mean temperatures, climate models project a general in-crease in precipitation in the high-latitudes and the tropics and a decrease in the subtrop-ics (Stocker et al., 2013). A robust response to increasing atmospheric greenhouse-gasconcentrations is that heavy precipitation events increase stronger than the mean (Allenand Ingram, 2002; Trenberth et al., 2003; Kharin et al., 2013; Rajczak et al., 2013; Fis-cher et al., 2014). Both the changes in mean precipitation and heavy precipitation haveimportant socio-economic impacts (Field et al., 2012). It is therefore of great interest tounderstand the driving processes. The increase in daily precipitation extremes is suggestedto follow the Clausius-Clapeyron (CC) relation (Allen and Ingram, 2002; Pall et al., 2007;Allan and Soden, 2008) which implies an increase of precipitable water by about 6-7%/Kunder constant relative humidity. Whether or not and for what time scales heavy precipi-tation indeed scales according to CC, and what temperature it scales with is, however, notentirely clear and a topic that is under constant debate (Lenderink and Van Meijgaard,2008; O’Gorman and Schneider, 2009; Muller et al., 2011; Loriaux et al., 2013; Ban et al.,2015). Global mean precipitation is expected to increase by about 1-3%/K, much slowerthan the CC scaling, since it is limited by the energy balance between radiative coolingand latent heating (e.g.Allen and Ingram 2002; Stephens and Ellis 2008). However, onsub-continental scales precipitation changes can differ from the global scale changes (seefor example European summer precipitation below). One likely reason for this discrep-ancy are dynamic factors such as changes in the large-scale circulation or the temperaturelapse-rate (e.g.O’Gorman et al. 2012).
To quantify and better understand the key drivers of regional changes in mean andheavy precipitation a common strategy is to disentangle the effect of thermodynamicand dynamic changes. Different methods to realize this separation were applied, rangingfrom moisture budget analyses (Seager et al., 2010) to CC-scaling as thermodynamicproxy (Pall et al., 2007; Radermacher and Tomassini, 2012), to weather-type analyses(Santos et al., 2016) and, finally, to a model-based and physically consistent separationof the effects (e.g Rowell and Jones 2006; Rasmussen et al. 2011). For most regions thethermodynamic effect is dominating but there are regions, for example on the polar sideof the subtropical subsidence regions, where the dynamic effect is more important (Seageret al., 2010; Scheff and Frierson, 2012).
In this work we focus on precipitation changes over the European continent. Previ-ous studies based on regional climate model (RCM) projections found a bi-polar changepattern over Europe, with increasing precipitation in northern Europe and decreasing pre-cipitation over the Mediterranean (Frei et al., 2006; Nikulin et al., 2011; Rajczak et al.,2013; Jacob et al., 2014). The transition area between increasing and decreasing precipi-tation is oscillating with the season, moving from north to south from winter to summer.Rajczak et al. (2013) showed that also frequency changes are subject to a clear bi-polarpattern whereas intensity changes are positive for almost whole Europe.
Several studies separately assessed thermodynamic and dynamically driven precipita-tion changes for Europe (Kendon et al., 2010; Radermacher and Tomassini, 2012; Seageret al., 2014) with conflicting findings. Kendon et al. (2010) attributed most of the summerdrying over the Mediterranean to thermodynamic changes, whereas Seager et al. (2014)argued that circulation changes are most important. Most of these apparent contradic-tions are probably caused by different definitions of thermodynamic and dynamic effects.
4.2 Method 47
In the present work, we complement these previous studies with the general aim tobetter quantify thermodynamic and dynamic influences on future precipitation changesover the European continent. For this purpose we apply the framework introduced byKroner et al. (2016) (from now on KR16) who developed a method to disentangle dynamicand thermodynamic changes, and in addition to quantify the role of stability changes, anissue that is often mentioned as an important aspect but rarely quantified (Emori andBrown, 2005; O’Gorman and Schneider, 2009). More specifically, KR16 extended thesurrogate approach that has been introduced by Schar et al. (1996) and that is able torepresent a warmer climate without changes in the large-scale circulation.
We here use this extended approach (1) to attribute projected changes in precipitationfrequency and intensity to thermodynamic, dynamic or lapse-rate changes, and (2) tostudy the influence of these three factors on changes in heavy precipitation events andprolonged dry spells. With these analyses we aim to shed light on the processes responsiblefor the general precipitation change pattern seen in European-scale climate projections.
The paper is structured as follows: First we will give a short summary of the methodused by KR16 as well as the model and the statistics used in this study. In section 4.3 theresults of our analysis are presented followed by a discussion and a conclusion in section4.4
4.2 Method
4.2.1 Surrogate approach
KR16 extended the surrogate climate change method introduced by Schar et al. (1996).The basic idea of the surrogate approach is to apply a large-scale warming to the lat-eral boundary conditions of a present-day RCM simulation, while maintaining relativehumidity (and thus implicitly increasing the specific moisture content). In the extended”twin design” of KR16 the large-scale warming ∆T is not only applied to a present-daysimulation but, also, a corresponding large-scale cooling is subtracted from the future sce-nario simulation. The respective warming (cooling) is calculated as the mean differencebetween the present-day (CTRL) and the scenario (SCEN) simulation averaged over thewhole domain and subject to an annual cycle (Fig. 4.5). KR16 applied this temperatureperturbation in two versions: First, ∆T is considered to be constant with height, andsecond ∆T(z) is applied as a function of height (Fig. 4.5). This procedure results in a setof six RCM experiments: the reference CTRL and SCEN simulations and four surrogateexperiments. These four experiments are: homogeneous warming HW, vertical warmingVW, homogeneous cooling HC and vertical cooling VC. The overall aim is to provide a setof simulations where the large-scale thermodynamic (TD), lapse-rate (LR) and dynamicchanges can be separated. As further discussed in KR16 the dynamic changes includechanges to the large-scale circulation and other remaining effects (for example the land-ocean contrast). Following KR16 we will use CO as the abbreviation for this effect. Thesix experiments can be combined to exactly yield the three effects and the full climatechange (FCC):
TD :=(HW − CTRL) + (SCEN −HC)
2(4.1)
48 Chapter 4: Quantifying drivers of European precipitation changes
LR :=(VW −HW ) + (HC − V C)
2(4.2)
CO :=(SCEN − VW ) + (V C − CTRL)
2(4.3)
FCC := SCEN − CTRL = TD + LR + CO (4.4)
Here the acronyms are meant to represent the effect of the respective changes and canfor instance be interpreted to refer to surface warming, mean precipitation changes, orsome other quantity of interest. The definitions in equation 4.1-4.3 are plausible if the twoways to calculate each effect yield similar results. For example, VW - HW and HC - VC,both used in eq. 4.2, should yield similar estimates. The validity of their assumption willbe validated a posteriori. In Figure 4.1 the result of the separation and the full climatechange signal (FCC) for seasonal mean precipitation changes according to the model setupdescribed in the following Section are shown.
4.2.2 Model-Setup
KR16 performed their study using the regional climate model COSMO-CLM (CCLM) inversion cosmo4.8 clm17 (Rockel et al., 2008). CCLM is a non-hydrostatic limited areamodel developed from the numerical weather prediction model COSMO of the GermanWeather Service DWD and maintained by the CLM-Community. The model is hereused in its well-tested configuration applied for the EURO-CORDEX (Jacob et al., 2014)ensemble. A more detailed description on the employed parametrisation schemes canbe found in Vautard et al. (2013) and Keuler et al. (2016). Lateral boundary data arederived from a transient simulation of the global general circulation model (GCM) MPI-ESM-LR (Giorgetta et al., 2013), which is also part of the CMIP5 (Taylor et al., 2012)ensemble. The CTRL and SCEN simulations are time slice experiments covering theperiods 1971-2000 and 2070-2099, respectively. For the climate change scenario greenhousegas concentrations according to the RCP8.5 (Moss et al., 2010) are assumed. The modeldomain is the same as the one used for the EURO-CORDEX integrations and is shownin Figure 4.1 excluding the lateral boundary zone of 10 grid cells. The resolution of thesimulations is 0.44 on a rotated grid, roughly corresponding to 50 km grid cell spacing.
4.2.3 Statistical approach
When addressing changes in precipitation statistics, we focus on four distinct Europeanregions that represent a subset of the eight widely used PRUDENCE regions (Christensenand Christensen, 2007). These sample the spatial pattern of the precipitation climate overthe European continent: Scandinavia (SC), Mid-Europe (ME), Alps (AL) and the IberianPeninsula (IP) (see boxes in Fig. 4.1g). We disentangle the mean precipitation changesignal into intensity and frequency of wet days (daily precipitation > 1mm). As intensitymeasure we use the simple daily intensity index (INT also known as SDII), corresponding
4.2 Method 49
Figure 4.1: Relative changes in total precipitation for DJF (left column) and JJA (right col-umn). The first three rows show (a-b) the large scale thermodynamic effect (TD), (c-d) the lapse-rate effect (LR), and (e-f) the large scale circulation and other remaining effects (CO), respec-tively. The fourth row shows the full climate change signal (FCC = SCEN − CTRL). (In thesouthern regions over the Sahara the absolute precipitation amounts are very low (< 0.2mm/day)therefore relative changes can become very large and are not robust). In (g) also the four analy-ses regions (Scandinavia (SC), Mid-Europe (ME), Alps (AL) and Iberian Peninsula (IP) fromnorth to south) are marked.
50 Chapter 4: Quantifying drivers of European precipitation changes
to the mean wet-day intensity (see Table 4.1). We further quantify the influence of thethree effects TD, LR and CO on dry spell length and extreme precipitation. To this endwe use the seasonal maximum number of consecutive dry days CDD and the seasonalmaximum one-day precipitation Rx1day (Zhang et al., 2011). All of the five measuresintroduced above are calculated for each grid point and each individual experiment. Then,the TD, LR and CO effects are calculated according to Eq. (4.1 - 4.3) and divided bythe respective value in the CTRL experiment, i.e., relative effects are computed. In a laststep these relative effects are spatially averaged for the respective region. In this workwe will consider the winter and summer season DJF and JJA respectively and will giveseasonal means where appropriate.
ID Full name DefinitionRmean seasonal mean precipitationINT Simple daily intensity index The ratio of seasonal total precipitation to
the number of wet days (≥ 1mm)FRE Wet-day frequency The ratio of the number of wet days (≥
1mm) to the total number of daysCDD Consecutive dry days 30year mean of seasonal maximum num-
ber of consecutive dry days with precipi-tation < 1mm
Rx1Day Max 1-day precipitation amount 30year mean of seasonal maximum 1-dayprecipitation
Table 4.1
4.3 Results
In Figure 4.1, seasonal mean precipitation changes are shown for TD, LR, CO and thefull climate change signal. The TD effect is very similar in winter and summer and leadsto an increase of precipitation between about 5% and 25% over most of Europe in bothseasons. In contrast, the LR effect has almost no influence in winter. In summer it leadsto a decrease in precipitation especially over southern Europe. Also the CO effect showssome variation with the season. In winter it is associated with a pronounced precipitationdecrease in the Mediterranean region and slight decreases over the rest of the continent.In contrast, CO shows a strong drying signal in summer of more than 15% over mostparts of Europe and precipitation increases of similar magnitude in the far North. Asalready shown in KR16, the three effects add up to the full climate change signal FCC(Fig. 4.1g and h). In winter the TD effect (i.e., large-scale thermodynamic changes) isdominating the mean precipitation change signal. In summer on the other hand the LReffect (lapse-rate changes) and the CO effect (large-scale circulation and other remainingchanges) overcompensate the TD effect, leading to an overall precipitation decrease overlarge parts of Europe. Although using a different separation method, Seager et al. (2014)found similar CO and TD effects for mean summer precipitation changes over Europe,but did not consider the lapse-rate effect.
In the following we further analyse possible causes of projected changes in mean pre-cipitation in the four regions introduced in section 4.2.3. For SC, ME and AL the increase
4.3 Results 51
in mean winter precipitation is mainly caused by an increase in intensity (INT) whereasthe frequency of rain events remains almost unchanged (Fig 4.2). The INT increase ismainly caused by the TD effect which explains around 90% of the FCC signal. Overthe IP, the CO compensates most of this precipitation increase by reducing precipitationfrequency and intensity (Figure 4.1c)). Changes in winter dry spell length (CDD) aresmall in all regions. Extreme precipitation (Rx1day) increases stronger than the meanin most regions (namely ME, AL and IP). This stronger change in extreme precipitationis mainly caused by the TD effect (i.e. by thermodynamic effects) which has a strongerinfluence on extreme than on mean precipitation (see also Fig. 4.6).
Figure 4.2: Seasonal and regional mean changes for DJF precipitation characteristics; Rmean= mean precipitation, INT = Wet-day intensity, FRE = wet-day frequency, CDD = meandryspell length, Rx1Day = seasonal maximum precipitation. The four bars represent the fullclimate change signal (FCC), the large-scale thermodynamic (TD), the lapse-rate (LR) and thelarge-scale circulation and other remaining (CO) effects. They are calculated according to Eq.4.1 - 4.4. The grey triangles show the two different ways to calculate each effect (see section4.2). Black stars indicate the results of the EURO-CORDEX multi-model ensemble (table 4.2),which are available for FCC only.
In summer (Fig. 3) mean precipitation is decreasing in all regions except for SC.Similar to winter, the TD effect would enhance precipitation along with an increase ofINT. However, in summer reduced precipitation frequency (FRE) overcompensates thisincrease in INT. This FRE decrease in summer is mainly caused by the CO effect. OverIP, the LR effect further decreases the frequency of precipitation days (Fig. 4.3) andis almost as important for the mean changes as the CO effect. Along with a tendencyto fewer wet days, the dry spell length (CDD) is increasing for all regions except SC,indicating increased risk for meteorological drought conditions in those regions. The COand LR effects, which decrease the number of precipitation days, are the main driversfor the increase in CDD. The TD effect enhances extreme precipitation (Rx1day) butis offset by CO and in IP also by LR, similar as for mean precipitation. However, theinfluence of CO is weaker, and that of TD stronger, than for the mean. Consequentlythis leads to a decrease in extreme precipitation in AL and IP, which is substantially less
52 Chapter 4: Quantifying drivers of European precipitation changes
pronounced than the reduction in mean precipitation. For ME, the change even becomespositive for heavy precipitation. The changes in INT, FRE and Rx1day described so farare consistent with the changes found by Frei et al. (2006); Rajczak et al. (2013) whoanalysed the PRUDENCE and ENSMBELES ensembles respectively.
Figure 4.3: As Figure 4.2 but for the summer season JJA.
In our special experiment design, we only use one GCM-RCM chain. To put ourmodel results into perspective, we also analysed the available EURO-CORDEX ensemblemember. Table 4.2 summarizes the models included in this ensemble. In Figures 4.2and 4.3 the individual EURO-CORDEX simulations are shown as stars. Note that onlythe FCC signal can be assessed with this additional ensemble, as no dedicated surrogateexperiments are available. Including EURO-CORDEX, however, presents an opportunityto assess the ensemble spread and to put the FCC signal of our specific model chainMPI-ESM-LR-CCLM into a larger context. In winter our model chain mostly producesan average signal. In summer it is typically located on the dry end of the ensemble,producing the smallest precipitation increase over SC and very strong decreases over ME,AL and IP. Also the changes in CDD are quite strong and sometimes exceed those ofthe other EURO-CORDEX members substantially (AL). It should be noted that manystudies found increases in JJA heavy events, while Rx1day is here projected to decreasein AL and IP. The results appear nevertheless consistent. In particular, Ban et al. (2015)found a shift from intensity decrease to increase at the 99th percentile. This percentilecorresponds to higher return period than the Rx1day indicator considered here.
The previous results showed that the lapse-rate effect has a strong influence on pre-cipitation changes over southern Europe in summer. Although Frierson (2006) alreadyshowed that lapse-rate changes are an eminent feature of European climate projections,their effect on individual variables was never quantified in a larger ensemble of climatesimulations. In the light of our findings we studied the lapse-rate effect on precipitation inthe EURO-CORDEX ensemble. We find a strong correlation between lapse-rate changesand changes in mean and heavy precipitation (Fig. 4.4). Almost 50% of the spread inmean and heavy precipitation over the Iberian Peninsula and the Mediterranean region
4.4 Discussion and conclusion 53
can be explained by lapse-rate changes. The lapse-rate for the two periods was calculatedas T500hPa − T850hPa, where T500hPa and T850hPa are calculated as annual mean over therespective 30 years over the whole model domain.
Figure 4.4: Correlations between Lapse-rate changes and changes in (a) Rmean (p <0.05, R2 = 0.41) and (b) Rx1Day (p < 0.05, R2 = 0.47) are shown for the Iberian Peninsulain summer (JJA). The ensembles is a sub-ensembles from (table 4.2) only including the sim-ulation done with the RCM RCA4 because only for those simulations temperature on differentpressure level was available during the study. For every model run the driving GCM is providedin the figure.
4.4 Discussion and conclusion
We have shown that for winter the thermodynamic effect plays a dominant role for theprojected mean precipitation increase, by enhancing the precipitation intensity. In sum-mer the intensity increase induced by the thermodynamic effect is of similar magnitudebut offset by a general decrease of precipitation frequency, leading to decrease in meanprecipitation. For extreme precipitation, in both seasons the thermodynamic effect is
54 Chapter 4: Quantifying drivers of European precipitation changes
stronger and the large-scale circulation effect weaker than for the mean, leading to sub-stantially less pronounced decreases or even an increase in extremes despite decreases inthe mean. In summer over IP and MD (not shown) the effect of lapse-rate changes is asstrong as the one of circulation changes.
In general, the influence of the CO effect on precipitation in winter and summer foundin this study are similar to previous studies (Riediger and Gratzki, 2014; Seager et al.,2014; Santos et al., 2016). In a GCM-RCM chain, the large-scale circulation is providedby the GCM (Rummukainen, 2010), and in our study only one GCM realisation wasused. This introduces uncertainties to the large-scale circulation response because even adifferent realisation of the same GCM (or another GCM) could show a different circulationchange signal (Deser et al., 2016). The TD and the LR effect are less affected by thisuncertainty, which can be shown by considering the two different ways to calculate them(see eq. 4.1-4.3), the difference between the two being the background circulation. Infigure 4.2 and 4.3 the two terms are marked as triangles and it can be seen that thedifferences are rather small. Only for Rx1Day in the summer season, the differences arelarger for the dry regions AL and IP. This could be related to the fact that Rx1dayin this season and regions is very sensitivity to small projected changes. Concerningthermodynamically induced changes, the summer and winter TD effects are very similar.At first glance this might be surprising, as previous modelling studies suggested a dryingover land driven by large-scale temperature increase (Rind et al., 1990; Gregory et al.,1997; Joshi et al., 2008; Sherwood and Fu, 2014). The idea behind this argumentation isthat the moisture sources over land cannot keep up with an increased atmospheric demandand therefore the local source for precipitation is decreased which is especially importantin summer. Such an effect cannot be seen in our study. One important difference toprevious works, however, is the treatment of land-ocean contrasts. In the setup of KR16,the SST in the surrogate experiments is assumed to increase by the same amount asthe near-surface temperatures over land, i.e. the land-sea temperature contrasts remainunchanged in these experiments. Hence, combining the mentioned studies with our ownwork suggests that not the general warming is the most important factor for decreasingsummer precipitation over land, but rather that changes in land-sea contrast are mostlyresponsible. Testing this hypothesis would go beyond the scope of this study but is, ingeneral, possible by extending the applied surrogate framework.
Comparing winter and summer precipitation changes, the difference in the LR effectis striking. Almost no effect in winter is seen, whereas in summer LR makes an importantcontribution especially in the Mediterranean regions. This behaviour can be explainedwith the sensitivity of different precipitation types with respect to changes in the atmo-spheric stability. The areas showing the strongest LR effect show the largest convectivefraction (IP, MD). Our results suggest that especially convective precipitation is sensitiveto changes in atmospheric stability and that increases in atmospheric stability, suppressespecially this type of precipitation.
The general precipitation change pattern over Europe also in terms of FRE and INTchanges were already shown in the past (Frei et al., 2006; Nikulin et al., 2011; Rajczaket al., 2013). The driver for those changes in FRE and INT are less clear. Using acompletely different method and looking at the tropics, Chou et al. (2012) found thatthermodynamic changes exert strong control over FRE and INT changes. In a studyover Europe more similar to our own, Kendon et al. (2010) also found a strong controlof the thermodynamic component and weaker contributions from large-scale circulation
4.4 Discussion and conclusion 55
changes. In our study, in contrast, we found a clear separation of the drivers on etherprecipitation measure with the thermodynamic changes controlling changes in INT, andlarge-scale circulation changes and the lapse-rate changes controlling changes in FRE. Thecontrasts with Kendon et al. (2010) can probably be explained by the different definitionsof thermodynamic and large-scale circulation changes as they further separated the in-fluence of soil-moisture changes. In our study, in contrast, soil-moisture is a dependentvariable; for a detailed discussion see KR16. The contrasting results to Chou et al. (2012)could simply be due to the different region studied, but the different method used opensa more interesting line of thought. Chou et al. (2012) applied a more mechanistic way toseparate thermodynamics and dynamics. In principal they argued that rain generatingprocesses could be described by −〈w ∗ ∂qc/∂p〉, with w being the vertical wind and qcthe specific moisture, and therefore dynamic changes could be identified by changes in wand thermodynamic by changes in ∂qs/∂p a similar approach was also used by Emori andBrown (2005) and O’Gorman and Schneider (2009). An important difference between ourand the more mechanistic method is that in the pseudo-warming approach the separa-tion is done by conducting different simulations whereas for the mechanistic method it isdone posterior. Further insight into the mechanisms driving the change in FRE and INTcould be gained by combining the methods; in this case the surrogate approach wouldprovide the controlled environment in which the mechanistic approaches could be testedand used to further study the regional response to the large-scale drivers. In this sense, afuture combination of both approaches to test and expand our knowledge on the drivingprocesses of precipitation changes would be of interest.
In terms of uncertainty one can regard the changes due to thermodynamics and thelapse-rate as more certain than changes in the large-scale circulation for Europe (Chris-tensen et al., 2013). In light of this, the increase of precipitation intensity, and even morethe increase in heavy precipitation, can be regarded as certain. In contrast changes in thegeneral frequency of rain events are much more uncertain, as well as the increase of CDD.An exception are the FRE and CDD changes over IP and the Mediterranean, which can beregarded with higher confidence because in this regions the projected lapse-rate changestogether with the large-scale circulation changes are regarded with high confidence by theIPCC (Christensen et al., 2013).
4.5 Supplemental Information 57
4.5 Supplemental Information
Full model chain abbreviationSMHI SMHI-RCA4 NOAA-GFDL-GFDL-ESM2M RCA4 GFDLSMHI SMHI-RCA4 NCC-NorESM1-M RCA4 NCCSMHI SMHI-RCA4 MPI-M-MPI-ESM-LR RCA4 MPISMHI SMHI-RCA4 MOHC-HadGEM2-ES RCA4 HadGEMSMHI SMHI-RCA4 MIROC-MIROC5 RCA4 MIROCSMHI SMHI-RCA4 IPSL-IPSL-CM5A-MR RCA4 CM5SMHI SMHI-RCA4 ICHEC-EC-EARTH RCA4 ECSMHI SMHI-RCA4 CSIRO-QCCCE-CSIRO-Mk3-6-0 RCA4 CSIROSMHI SMHI-RCA4 CNRM-CERFACS-CNRM-CM5 RCA4 CNRMSMHI SMHI-RCA4 CCCma-CanESM2 RCA4 CCCmaKNMI KNMI-RACMO22E MOHC-HadGEM2-ES RACMO HadGEMKNMI KNMI-RACMO22E ICHEC-EC-EARTH RACMO ECIPSL-INERIS IPSL-INERIS-WRF331F IPSL-IPSL-CM5A-MR WRF IPSLICTP ICTP-RegCM4-3 MOHC-HadGEM2 RegCM4 HadGEMHMS HMS-ALADIN52 CNRM-CERFACS-CNRM-CM5 ALADIN52 CNRMDMI DMI-HIRHAM5 ICHEC-EC-EARTH HIRHAM5 ECCNRM CNRM-ALADIN53 CNRM-CERFACS-CNRM-CM5 ALADIN53 CNRM
Table 4.2: EURO-CORDEX (50km resolution ensemble) regional climate models used in thisstudy
4.5.1 Temperature scaling
Figure 4.6 shows the scaling of the highest daily precipitation percentiles (99 to 99.98)with temperature for the TD and FCC experiments. The LR and CO are not shownbecause they are designed in such a way that the mean temperature change is zero,therefore a temperature scaling would be pointless. In winter, all percentiles show anincrease, which does not strongly depend on the individual percentile, and which in mostcases approximately corresponds to the Clausius-Clayperon scaling of 6.5%/K. FCC andTD effects are close to each other, although the scaling is typically less pronounced inFCC. These results confirm that in winter the thermodynamic effect is controlling theoverall precipitation response.
Figure 4.5: a) mean annual cycle of ∆T at 850hPa, b) mean height profile of ∆T for JJA, c)mean height profile of ∆T for DJF
58 Chapter 4: Quantifying drivers of European precipitation changes
Figure 4.6: Temperature scaling for TD and FCC: Spatial average of the percentage change ofhigh percentiles of the wet day precipitation distribution divided by the corresponding temperaturechange. Left column : DJF, right column: JJA. In the four rows the 4 regions SC (a-b), ME(c-d), AL (e-f) and IP (g-h) are shown respectively. The dashed line indicates a precipitationincrease according to the Clausius-Clapeyron relation (7%/K).
4.5 Supplemental Information 59
In summer the thermodynamic effect is increasing towards the higher percentiles,reaching and eventually surpassing CC-scaling. The FCC signal is reduced in comparisonto the TD signal which indicates an important contribution of the CO and LR effectseven on the higher percentiles. Towards very high percentiles also the FCC signal showsan increase and is approaching the CC-scaling.
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80DJF CORDEX intensity for percentile 99.8 in SC pr
(a)
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80DJF CORDEX intensity for percentile 99.8 in ME pr
(b)
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80DJF CORDEX intensity for percentile 99.8 in AL pr
(c)
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80DJF CORDEX intensity for percentile 99.8 in IP pr
(d)
Figure 4.7: Relative changes of the 99.8th percentile for the EURO-CORDEX Ensemble (Tab.4.2) in winter DJF for (a) SC, (b) ME, (c) AL and (d) IP. TD is calculated by applying CC-scaling to the temperature change in the respective region. CO+LR is the residuum between theFCC and TD.
Pall et al. (2007) assumed that extreme precipitation should scale with the Clausius-Clapyeron relation and that all deviations from this scaling are caused by dynamic drivers.In Figure 4.6 it is shown that for the surrogate framework around the 99.8th percentile, theassumption is fulfilled and the thermodynamic effect scales with the Clausius-Clapyeronrelation. In Figure 4.7 and 4.8 this CC-relation was used to separate thermodynamicand dynamic drivers in the EURO-CORDEX ensemble (Table 4.2). The lapse-rate effectcan not be quantified separately with this method. The separation shows that in thewinter season differences in temperature change contribute equally to the model spread(in extreme precipitation changes) as differences in the dynamic factors (Fig. 4.7). Incontrast in summer for ME,AL and IP most of the model spread has to be caused bydynamic factors (Fig. 4.8).
4.5.2 Event frequencies
In Figure 4.9 the simulated changes of individual precipitation frequencies for the TD, LRand CO effects and for the FCC are shown. In winter (left column), the FCC indicatesfrequency increases for event intensities up from about 8±4 mm/day depending on theregion under consideration. For small intensities and except for IP, this signal is pickedup by the TD effect which, however, surpasses the FCC increase for high intensities but
60 Chapter 4: Quantifying drivers of European precipitation changes
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80JJA CORDEX intensity for percentile 99.8 in SC pr
(a)
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80JJA CORDEX intensity for percentile 99.8 in ME pr
(b)
TD FCC CO+LR
Inte
ns
ity
ch
an
ge
[%
]
-80
-60
-40
-20
0
20
40
60
80JJA CORDEX intensity for percentile 99.8 in AL pr
(c)
TD FCC CO+LRIn
ten
sit
y c
ha
ng
e [
%]
-80
-60
-40
-20
0
20
40
60
80JJA CORDEX intensity for percentile 99.8 in IP pr
(d)
Figure 4.8: As Fig. 4.7 but for JJA
still qualitatively agrees with FCC. In contrast, LR and CO show only small frequencyincreases or even decreases for mid to high intensities. These findings indicate the domi-nating influence of thermodynamic effects on changes in the distribution of daily winterprecipitation intensities. Positive summer frequency changes (right column) in FCC areobtained in SC and for high intensities in ME only, and are negative in AL and IP overthe entire intensity spectrum. This decrease is mostly dominated by the CO effect, i.e.,by large scale circulation changes. In both winter and summer the TD effect is subject tothe strongest dependency of frequency changes on the specific intensity class consideredand, except for ME in winter and IP in summer, is typically associated with strongerfrequency increases for high precipitation intensities. High intensities hence seem to bemost affected by thermodynamic changes.
4.5 Supplemental Information 61
Figure 4.9: Ratio of frequencies for daily precipitation intensities. On the horizontal axes thebins of the intensity classes (mm/day) are shown. The uppermost class includes daily eventswith more than 50 mm/day. On the vertical axes the ratio of frequencies for the three effectsTD, LR, CO and the FCC is shown. Left column: DJF, right column: JJA. In the four rows the4 regions SC (a-b), ME (c-d), AL (e-f) and IP (g-h) are shown respectively. The distributionsare calculated for every grid-point separately and then averaged over the specific region.
Chapter 5
Conclusion and outlook
5.1 Conclusion
This thesis identifies and quantifies the large-scale drivers of European climate change.For this purpose an extended surrogate approach is developed and implemented withthe regional climate model COSMO-CLM. The three main drivers identified here are (1)large-scale thermodynamic (TD), (2) lapse-rate (LR) and (3) large-scale circulation andother remaining (CO) changes.
In Chapter 3 this new approach is applied to the European summer climate. It isshown that the separation is plausible and that the novel twin design, in which bothwarming and cooling profiles are applied to CTRL and SCEN simulations, respectively, isan elegant way of separating the three main drivers. The large-scale thermodynamic effectleads to a warmer and moister climate but does not account for the characteristic climatechange pattern over the European continent (i.e. drying and amplified warming overthe Mediterranean). We show that this strong north-south warming gradient is causedby the large-scale circulation - in our framework including also the effects of land-seacontrast and spatial variations of the SST warming patterns - and the lapse-rate effect.Indeed, the lapse-rate effect, quantified for the first time in this study, accounts for halfof the warming amplification simulated over the Iberian peninsula. Both the large-scalecirculation and the lapse-rate effect contribute to the Mediterranean drying. While thelapse-rate effect is confined to the Mediterranean, the large-scale circulation effect extendsfar into Mid-Europe.
In Chapter 4 the new approach is used to elucidate the mechanism behind the seasonalnorth- and southward shift of the precipitation-response pattern, with a northern Euro-pean precipitation increase and a southern European drying. Furthermore, the large-scaledrivers of the changes in precipitation intensity and frequency for the European continentare examined. We find that the thermodynamic effect enhances precipitation intensity insummer and winter throughout the continent. The varying contributions of large-scalecirculation and lapse-rate effects modulate this pattern and are responsible for the oscil-lating precipitation change pattern over the seasons. The large-scale circulation as well asthe lapse-rate changes have a strong effect on summer precipitation. Both have only smalleffects on precipitation intensity but are strongly decreasing precipitation frequency. Thelapse-rate effect is confined to the Mediterranean region because it particularly dampensconvective precipitation events, which represent a large fraction of summer precipitationin this region.
63
64 Chapter 5: Conclusion and outlook
The European climate was subject to climatic changes in the last century and it isprojected that those changes continue in the future. The projected changes are governedby a complex interplay between the large-scale climate changes and regional and local scalefeedbacks. Regional climate models driven by global climate models are at the moment themost robust way to simulate and capture these interaction in the future. Nevertheless stillconsiderable uncertainties remain with respect to model uncertainty, internal variabilityand scenario uncertainty (Section 1.3). The topic of disentangling thermodynamic anddynamic drivers for climate change is becoming increasingly popular and is used to shedlight on the processes driving specific regional changes (see Section 1.5). This thesis makesan contribution to the related discussion by further advancing a methodology which incomparison to the previous diagnostic works can objectively separate the effects. It wasshown that despite the high complexity and interactions shaping the European climate,insight into the processes can be gained by separating its large-scale drivers, which leadsto one important conclusion from this thesis: The effort to disentangle the large-scaledrivers is very worthwhile. Not only is the new method very well suited to study theregional changes of different climate variables over the European continent, but it canalso be used to separate robust from uncertain large-scale drivers.
In general the changes connected to the thermodynamics are regarded as very ro-bust whereas the changes due to large-scale circulation changes have higher uncertainties(Stocker et al., 2013). The analysis of the changes in precipitation statistics revealed aclear separation between the drivers with the thermodynamic effect mainly increasing in-tensity and the circulation effect decreasing precipitation frequency. In the Mediterraneanregion the lapse-rate effect was revealed as a third important driver, also decreasing precip-itation frequency. In accordance with these findings it was shown that the thermodynamiceffect has stronger influences on heavy precipitation than the mean, whereas the oppositeis found for circulation and the lapse-rate effect. This leads to the conclusion that theincreases in precipitation intensity, especially for heavy precipitation events, is a ratherrobust feature of European climate changes whereas changes in precipitation frequencyand prolonged dry-spells are more uncertain except for the Mediterranean climate wherealso the more robust lapse-rate effects are identified as important driver.
The overall influence of the lapse-rate change on European climate was quantifiedthe first time in this thesis. It was shown that lapse-rate changes substantially influencethe amplified warming in the Mediterranean region and contribute to the decrease inprecipitation. In conclusion, the lapse-rate changes should be accounted for in the analysisof future multi-model studies and further research should be conducted to understand themechanism behind its influence on regional climate change.
5.2 Outlook
The work carried out in the frame of this thesis presents an important step towardsunderstanding and quantifying the influence of large-scale drivers on regional climatechange patterns over the European continent. Still, our approach is subject to a numberof limitations and open questions remain. In the following, possible next steps to improveon this and to deal with some of the caveats of this thesis are outlined.
• The caveat of using only one single model setup
5.2 Outlook 65
The method presented in this thesis was only applied to one realization of oneparticular GCM-RCM chain. However, the large-scale circulation and other ef-fects, as it is quantified in this thesis, depends on how the large-scale circulationis projected to change in this specific GCM realisation, and also on how realistichistorical circulation patterns are represented. Also, note that we did not aim atseparating the large-scale circulation changes into a forced response and non-forcedinternal variability. This could be done by analysing different realizations of thesame GCM-RCM chain. With only one GCM used here, we are unable to quantifythe uncertainty of the CO effect. However, by performing several experiments usingcompletely different GCMs as driving data we could further quantify the uncertaintyin the circulation effect as well as in the thermodynamic and the lapse-rate effects.For the last two effects part of the uncertainty is addressed by the twin design andthe results give some confidence that the lapse-rate effect and the thermodynamiceffect do not overly depend on the background circulation. In a general sense, thequantification of all three effects could be dependent on the specific RCM and uponthe computational domain used. To address this uncertainty further experimentswith different RCMs would be required.
• Synergy with other methods
In contrast to most other (purely diagnostic) separation methods the experimen-tal design of the surrogate approach is a priori targeted towards a separation ofthe large-scale drivers. On the one hand this confines the methods applicability todedicated surrogate studies and excludes the use of standard regional climate sce-narios as available from large ensemble projects. On the other hand the separationcan be designed and carried out more strictly. A promising approach for furtherresearch could be a combination of the diagnostic and the surrogate method. Insuch a framework the diagnostic methods (see section 1.5) could be validated in asurrogate ensemble. For example one could examine if extreme precipitation scaleswith the CC-scaling for the TD effect validating the assumption of CC-scaling asa proxy for thermodynamic changes (Pall et al., 2007). In a similar way also otherdiagnostic methods could by validated.
• Factor interactions
In Chapter 3 we presented six surrogate experiments and the methodology to com-bine them to yield the TD, LR and CO effects. The surrogate approach is notthe only way to disentangle the large-scale drivers; further possible methods wereshortly summarized in Chapter 1.5. Overall, these methods can be seen as a sub-group of sensitivity studies, i.e. of the more general effort in climate science toquantify the influence of certain factors on climate simulations. For this purposeStein and Alpert (1993) introduced a general approach to formalise the differentmethods and to also take interactions between different factors into account. Tofurther test the robustness of our results (the twin design already proved plausibil-ity) one could translate our six-member ensemble to the factor separation method.In this way interactions between the different large-scale drivers, i.e. their mutualdependencies, could be tested. We followed this approach and the results can befound in (Schar and Kroner, 2016).
66 Chapter 5: Conclusion and outlook
• Convection-permitting surrogate simulations
In the framework developed in this thesis a strong influence of lapse-rate changes onconvective precipitation was found (compare Chapter 4). However, convective pre-cipitation is parametrized in the simulations used in this study, which can introduceconsiderable uncertainties (Grell et al., 2000; Chan et al., 2014; Ban et al., 2015). Toreduce this uncertainty and to further address the influence of lapse-rate changes,higher resolution simulations would be needed. Fortunately convection-permittingclimate simulations have recently become feasible on sub-European and even on theEuropean scale (Kendon et al., 2012; Ban et al., 2014, 2015; Leutwyler et al., 2016).Together with the extended surrogate approach developed in this thesis the under-standing of the influence of lapse-rate changes on heavy convective precipitationevents could be improved. Several studies - although on smaller domains - alreadycombined convective permitting simulations and alternative surrogate approachesRasmussen et al. (2011), Attema et al. (2014) and Sorland and Sorteberg (2015).
• Extending the surrogate framework
With the extended surrogate approach the lapse-rate effect was isolated, in additionto the usual separation in large-scale circulation and large-scale thermodynamiceffects. But the new method can be further extended. The special twin designallows the user to apply a large variety of large-scale forcings (SST, relative humiditypatterns, etc) like a construction kit from which the total climate change signal isbuilt piece by piece. In chapter 3 the land-ocean contrast was present as one suchan extension. In the future it would be of great interest to also separately quantifythe influence of slow and fast varying changes of the large-scale circulation.
• Connection to impact studies
Climate impact studies are typically using multi-model ensemble projections to as-sess the direct impacts of global climate change on the human society. A hugeproblem to deal with is the large uncertainty of the climate projections. Withthe surrogate method robust and less robust changes (for example thermodynamicagainst circulation changes) can in principle be separated and this information couldalso be used to improve the robustness of impact assessments. For instance, onecould discriminate between impacts expected with high confidence and with lowconfidence.
• Temperature variability
In this thesis mainly changes in climate mean were addressed. But especially for theEuropean summer climate, temperature variability is an interesting topic. Althoughthis issue has recently been addressed by a large number of studies (e.g.,Schar et al.2004; Fischer and Schar 2010; Vautard et al. 2013) which identified several impor-tant processes such as soil-moisture temperature interactions (Seneviratne et al.,2006b; Fischer and Schar, 2009; Boe and Terray, 2014), an increased north-southtemperature gradient (Schneider et al., 2015) and cloud effects (Tang et al., 2012;Boe and Terray, 2014). Still, the relative influence of the different processes andtheir large-scale drivers remain unknown and the projections show considerable un-certainty (Fischer et al., 2012). The extended surrogate framework could be used to
5.2 Outlook 67
better quantify the influence of the different large-scale drivers on the regional feed-backs controlling future changes in temperature variability. In a first step changes insummer inter-annual and daily temperature variability were calculated (Fig. B.4).
• The lapse-rate effect
The strong influence of lapse-rate changes on the changes in Mediterranean precipi-tation and temperature has been demonstrated for the first time in this thesis. TheLR effect on precipitation due to its suppression of vertical motion and convectiveup-drafts is relatively straight-forward, whereas its influence on the Mediterraneanwarming anomaly is more complex. It is possible that the warming amplification isnot directly caused by the lapse-rate effect but that it is a secondary effect followingthe precipitation decrease via soil-moisture temperature interactions. However, aspecial experiment with fixed soil-moisture showed that soil-moisture temperatureinteractions can explain only part of the effect (see Appendix A). Over the IberianPeninsula further mechanisms obviously contribute to the amplified warming. Forinstance, the stronger warming aloft could be mixed down by convective mixing andthereby amplify the near-surface warming. A possible strategy to study this idea isto perform additional surrogate experiments with varying lapse-rates.
• Generalisation of the surrogate findings
This thesis identified the lapse-rate effect as an important element of Europeanclimate change. In Chapter 4 a first step toward generalizing the findings in thecontext of a multi-model ensemble was performed. It would be of great interest toalso validate the identified Mediterranean warming amplification in a multi-modelensemble framework (See Appendix A).
In summary, the extended surrogate approach developed and applied in this thesis revealsa great potential for sensitivity studies to improve our understanding of atmospheric pro-cesses. The fast development of computer resources not only allows us to further increasethe resolution of climate projections, but also makes process studies with intermediateresolution affordable. The presented results clearly show that it is worth to invest in thistype of study and the open questions and further lines of research presented motivatefurther research.
Appendix A
The lapse-rate effect on theMediterranean warmingamplification. Sensitivity study andthe multi-model perspective
In Chapter 3 it was shown that about half of the warming amplification over the IberianPeninsula simulated in the full climate change scenario is caused by the lapse-rate ef-fect. The first part of Appendix A summarizes further analyses and experiments whichwere conducted to explain the physical mechanism behind the influence of the lapse-rateeffect and the second part studies the connection between the lapse-rate effect and theMediterranean warming in the multi-model EURO-CORDEX ensemble.
A.1 Physical mechanism of the lapse-rate effect
To elaborate the physical mechanism of the lapse rate effect additional surrogate ex-periments were performed. Because of computational constraints those additional ex-periments could only be performed for one of the two ways to quantify the lapse rateeffect. Therefore in the following the lapse-rate effect is quantified as VW-HW not [(VW-HW)+(HC-VC)]/2. This approximation is justified by the finding presented in Chapter3 that VW-HW and HC-VC are very similar (Fig 3.6).
Figure A.1 summarizes the lapse-rate effect for the summer season JJA (VW-HWbased on the standard surrogate experiments introduced in Chapter 3) on temperature,precipitation, soil-moisture and the surface fluxes of sensible and latent heat. Fischer(2007) showed that a reduced latent heat flux and an increased sensible heat flux have awarming effect on 2m temperature. Soil-moisture in turn is the crucial variable determin-ing the partitioning of incoming solar energy into sensible and latent heat, which leads tostrong temperature soil-moisture interactions. Several studies so far showed that this soil-moisture temperature interaction is important for European climate change (Seneviratneet al., 2010; Jaeger and Seneviratne, 2011; Fischer et al., 2013). Indeed, it can be seenthat the LR effect decreases the surface latent heat flux (i.e. decreases evaporation) andincreases the surface sensible heat flux (Fig. A.1d,e) and that the spatial pattern of thechanges in the surfaces fluxes is very similar to the temperature change pattern. The LR
69
70 Chapter A: The lapse-rate effect
effect on soil-moisture is shown in Figure A.1c); its spatial pattern roughly correspondsto the change in surface fluxes.
These results suggest that the LR effect on soil-moisture due to decreased precipita-tion (Fig. A.1b) is also responsible for the specific temperature change signal via theaforementioned soil-moisture temperature interactions. An additional surrogate experi-ment (VW fix soil) has been carried out to review this theory. This experiments setupis identical to the VW experiment described in Chapter 3 with the exception that soil-moisture is not allowed to evolve freely but is prescribed as the soil-moisture of the HWexperiment (to allow the soil-moisture to interact with the actual weather the two up-permost soil layers can evolve freely). The HW experiment and therefore the VW fix soilexperiment have the higher mean soil-moisture content.
Figure A.2 shows the difference between the regular VW experiment and VW fix soil.It can be seen that the warming amplification is slightly influenced by the soil-moistureconditions. This influence is stronger in Eastern Europe but in general it is considerablysmaller than the influence of the entire lapse-rate effect (compare to figure A.1a)). Inter-estingly the soil-moisture influence on the surface fluxes is quite strong and almost all fluxchanges in the regular LR surrogate experiment can be explained by soil-moisture changes.This, in turn, means that soil-moisture changes and changes in the surface fluxes are notthe primary driver of the warming amplification in the Mediterranean region caused bythe LR effect.
Another theory to explain the warming amplification over the Mediterranean is thatfor the vertically dependent warming the stronger warming at higher altitudes is mixeddown to the surface over the Mediterranean which would mean the warming amplificationis not driven from below (soil-moisture changes, surface flux changes) but from above. InFigure A.3 the mean vertical temperature change profiles for the summer season over fourdifferent regions are shown (Scandinavia (SC), Mid-Europe (ME), Alps (AL) and IberianPeninsula (IP)). These regions correspond to the European sub-domains defined by thePRUDENCE project (Christensen and Christensen, 2007). Shown are the actual changeprofiles simulated over the regions (solid lines) as well as the change profiles applied to theboundaries (dashed lines). The tilting of the HW profiles in comparison to the prescribedhomogeneous profiles was already discussed in Chapter 3. Over IP the VW profile alsodeviates from the prescribed one and shows a near-surface warming amplification whichreaches up to 4000m (Fig. A.3d) and is accompanied by a strong cooling above 4000m.The other three regions do not show such a strong warming amplification and actuallythe profiles over Scandinavia and Mid-Europe show a weak cooling throughout the profile(Fig. A.3a,b). For AL the cooling above 4000m is stronger although not as strong as overIP and also in comparison to SC and ME below 4000m almost no cooling is simulated butrather a slight near-surface warming (Fig. A.3c). The vertical profile over IP would be inaccordance with the down-mixing theory but as this could also be the result of anotherprocess we cannot fully confirm the theory. Heating rates due to convective mixing wouldbe well suited to confirm or disproof the theory but those are not part of the normalmodel output and therefore not readily available.
A.1 Physical mechanism of the lapse-rate effect 71
Figure A.1: Lapse-rate effect for JJA estimated as VW-HW on a) 2m temperature, b) relativeprecipitation, c) relative soil-moisture d) sensible heat flux and e) latent heat flux. ( d) and e)positive fluxes are defined upward)
72 Chapter A: The lapse-rate effect
Figure A.2: Soil-moisture contribution to the lapse-rate effect estimated as VW−VW fix soilon a) 2m temperature, b) relative precipitation, c) sensible heat flux and d) latent heat flux. (c) and d) positive fluxes are defined upward)
Figure A.3: Vertical profile of JJA temperature difference to CTRL a) SC, b) ME, c) AL andd) IP. Green for HW, orange for VW and red for SCEN. Dashed lines represent the changeprofile imposed on the boundaries, full lines the change profiles realised in the domain.
A.2 The lapse-rate effect in the EURO-CORDEX ensemble 73
A.2 The lapse-rate effect in the EURO-CORDEX en-
semble
In this thesis a strong effect of lapse-rate changes on the Mediterranean warming am-plification was found but only one GCM-RCM chain was applied to quantify this effect.This introduces considerable uncertainty to the effect as it may depend on the appliedGCM-RCM chain. Therefore it is of great interest to study the lapse-rate effect in a largermulti-model ensemble. Figure A.4 shows the correlation between the lapse-rate changeand the southern European warming anomaly in the EURO-CORDEX multi-model en-semble. In figure A.4a) the available EURO-CORDEX simulations at a resolution of11.25 km (EUR-11) are used (see table A.1). In figure A.4b) the EURO-CORDEX sub-ensemble at 50 km (EUR-44) carried out by one specific RCM only (SMHI-RCA4) isshown. The lapse-rate change was calculated in the same way as in Section 4.3. Thesouthern warming anomaly is the difference between the area mean climate change signalover the Mediterranean region defined by the longitude-latitude box [-10,25,35,44] and thearea mean climate change signal over South- and Mid-Europe defined by [-10,25,35,59].In both areas only land points are used to calculate the mean. For both ensembles sig-nificant correlations are found, supporting the finding of the surrogate climate changeexperiments.
Full model chain abbreviationSMHI SMHI-RCA4 MPI-M-MPI-ESM-LR MPI RCA4SMHI SMHI-RCA4 MOHC-HadGEM2-ES HadGEM RCA4SMHI SMHI-RCA4 IPSL-IPSL-CM5A-MR IPSL RCA4SMHI SMHI-RCA4 ICHEC-EC-EARTH EC RCA4SMHI SMHI-RCA4 CNRM-CERFACS-CNRM-CM5 CNRM RCA4KNMI KNMI-RACMO22E ICHEC-EC-EARTH EC RACMOIPSL-INERIS IPSL-INERIS-WRF331F IPSL-IPSL-CM5A-MR IPSL WRFDMI DMI-HIRHAM5 ICHEC-EC-EARTH EC HIRHAM5CLMcom CLMcom-CCLM4.8-17 MPI-M-MPI-ESM-LR MPI CCLMCLMcom CLMcom-CCLM4.8-17 MOHC-HadGEM2-ES HadGEM CCLMCLMcom CLMcom-CCLM4.8-17 ICHEC-EC-EARTH EC CCLMCLMcom CLMcom-CCLM4.8-17 CNRM-CERFACS-CNRM-CM5 CNRM CCLM
Table A.1: EURO-CORDEX (11.25km resolution ensemble; EUR-11) regional climate modelsused in this study
74 Chapter A: The lapse-rate effect
Figure A.4: Correlations between large-scale lapse-rate changes and southern European (SE)warming anomaly. (a) EURO-CORDEX EUR-11 ensemble (table A.1) with r=0.62, correlationsignificant at the 5% level for every model run the driving GCM and RCM are provided inthe figure) and (b) EURO-CORDEX SMHI-RCA4 EUR-44 ensemble (table A.2 with r=0.77,correlation significant on the 5% level only the driving GCM is provided in the figure becauseonly one RCM (RCA4) is used in this sub-ensemble.
Full model chain abbreviationSMHI SMHI-RCA4 NOAA-GFDL-GFDL-ESM2M GFDLSMHI SMHI-RCA4 NCC-NorESM1-M NCCSMHI SMHI-RCA4 MPI-M-MPI-ESM-LR MPISMHI SMHI-RCA4 MOHC-HadGEM2-ES HadGEMSMHI SMHI-RCA4 MIROC-MIROC5 MIROCSMHI SMHI-RCA4 IPSL-IPSL-CM5A-MR CM5SMHI SMHI-RCA4 ICHEC-EC-EARTH ECSMHI SMHI-RCA4 CSIRO-QCCCE-CSIRO-Mk3-6-0 CSIROSMHI SMHI-RCA4 CNRM-CERFACS-CNRM-CM5 CNRMSMHI SMHI-RCA4 CCCma-CanESM2 CCCma
Table A.2: EURO-CORDEX (50km resolution sub-ensemble, EUR-44) regional climate modelsused in this study
Appendix B
Further analysis of the surrogateclimate change ensemble
B.1 Different seasons
In Chapter 3 the focus of the work was on the European summer climate (JJA). In FigureB.1 - B.3 the analysis of the three effects is also shown for the other seasons winter (DJF),spring (MAM) and autumn (SON). It can be seen that the lapse-rate effect is much weakerin the other seasons and the large-scale circulation effect is dominating the spatial patternof the temperature change. Already in Chapter 4 it was shown that for precipitation inwinter the thermodynamic effect is dominating over the other two effects and only overthe Mediterranean the circulation effect has considerable influence. In SON and MAMthe circulation effect is extending its influence further north. An interesting effect canbe seen in MAM where over large parts of Mid-Europe the circulation effect and thethermodynamic effect are increasing precipitation, which leads to very strong increasesof mean precipitation in the full climate change signal over those regions. For SON suchan effect is not found. In general the separation of the three effects also works for theother seasons: the two ways of calculating one effect are similar also for these seasons(not shown).
B.2 Temperature variability
In this thesis mainly mean changes and their spatial patterns for the different variableswere addressed. But for the European summer climate especially changes in temperaturevariability are an important topic. In Figure B.4 the influence of the three effects onrelative changes in daily summer temperature variability (DSV) and inter-annual summertemperature variability (IASV) are shown. DSV is calculated as the standard deviation ofall JJA daily mean values of the respective 30-year time slice. This maximum is locatedfurther North in comparison to the maximum mean temperature change (Fig. 3.3). Thechanges in IASV show a similar pattern as the DSV changes but are more scattered withespecially strong changes over the Balkan area and along the coast of the Black Sea. Itcan be seen, that most of the changes in DSV and IASV are caused by the large-scalecirculation effect (Fig. B.4f). The lapse-rate effect, in contrast to the mean warmingsignal, increases both types of variability rather homogeneously over the whole continent
75
76 Chapter B: Further analysis
Figure B.1: Separation of the DJF climate change signal. The first row shows (a-b) thelarge scale thermodynamic effect (LST ) (c-d) the lapse-rate effect (LR) and (e-f) the largescale circulation effect (LSC), respectively. The fourth row shows the full climate change signal(FCC = SCEN − CTRL). The left column shows changes in 2m seasonal mean temperature,as anomalies with respect to the domain mean. The right column shows relative changes intotal precipitation (Note that in the southern regions over the Sahara the absolute precipitationamounts are very low (< 0.2mm/day) therefore relative changes can become very high and arenot robust).
B.2 Temperature variability 77
Figure B.2: As Fig. B.1 but for the spring season MAM.
78 Chapter B: Further analysis
Figure B.3: As Fig. B.1 but for the autumn season SON.
B.2 Temperature variability 79
(Fig. B.4c,d). For DSV the LST effect shows a north-south gradient with increasingtemperature variability in the south and decreasing in the north (Fig. B.4a). For IASVthe LST effect has almost no influence at all (Fig. B.4b).
In figure B.5 the influence of the three effects on changes in 2m minimum and maxi-mum temperatures are shown. On can see that for minimum temperatures the LR effectexplains a larger fraction of the projected Mediterranean warming amplification than forthe mean. For maximum temperature changes the LSC effect is stronger than for themean.
80 Chapter B: Further analysis
Figure B.4: Separation of the JJA climate change signal. The first row shows (a-b) thelarge scale thermodynamic effect (LST ) (c-d) the lapse-rate effect (LR) and (e-f) the largescale circulation effect (LSC), respectively. The fourth row shows the full climate change signal(FCC = SCEN−CTRL). The left column shows relative changes in daily summer temperaturevariability (DSV). The right column shows relative changes in inter-annual summer temperaturevariability (IASV).
B.2 Temperature variability 81
Figure B.5: As Fig. B.4 but for minimum 2m temperature (left column) and maximum 2mtemperature (right column).
Appendix C
Derivation of pressure adjustment
In the surrogate climate change method the temperature and specific humidity of theboundary data is changed. The procedure to maintain hydrostatic balance in the pressurefield is derived in this appendix:
We start with the hydro static balance
dp(z)
dz= −g · ρ(z) (C.1)
where g =earth acceleration, p =pressure, z =height and ρ =density.Using the ideal gas law
ρ =p
RT(C.2)
where T =Temperature and R =the universal gas constant.we can rewrite equation C.1 as:
dp(z)
dz= −g · p(z)
R · T (z)(C.3)
or
dln(p(z)) = − gR· 1
T (z)· dz (C.4)
By integrating equation C.4 the vertical pressure profile psur in the boundary datafor the surrogate simulations after applying a temperature anomaly ∆T to the referenceboundary data can be calculated. Because T (z) is not known equation C.4 has to beintegrated numerically. This can be done by discretizing equation C.4:
psur(k) = psur(k − 1) · exp( gR· 1
Tx(k) + ∆T·∆z(k)) (C.5)
where k = kn− 1 with kn = number of model levels, Tx(k) the ”mean” temperaturebetween the two levels k and k-1, ∆z(k) = 0.5·(hhl(k+1)−hhl(k−1)) where hhl(k) is the
83
84 Chapter C: Derivation of pressure adjustment
height of the model level k in meter and ∆T (k) the surrogate temperature anomaly whichcan be independent and dependent on k corresponding to the homogeneous or verticaldependent warming. Tx(k) can be obtained by using equation C.5 solving for T :
Tx(k) = g · ∆z(k)
R · ln(pref (k−1)
pref (k))
(C.6)
where pref is the pressure in the reference state.Finally equation C.5 can be integrated from the bottom to the top keeping the pressure
at the lowest level kn constant psur(ke) = pref (ke). Over topography, however, also thepressure psur(kn) has to adopted. This is done with the following equation:
psur(z = ke) = pref (z = kn)((Tred
Tref (ke)
Tref (ke) + ∆T
Tred+ ∆T))κ (C.7)
where ke lowest level, Tref (ke) temperature at lowest level in reference state and Tredis the hypothetical temperature at z = 0 within the topography.
Tred is calculated by assuming the dry-adiabatic lapse-rate γ within the topographyyielding:
Tred = T (z = 0) = Tref (ke)− γ(zkn − z = 0) (C.8)
with γ = 0.00065K/m.
Bibliography
Alexander, L., X. Zhang, T. Peterson, J. Caesar, B. Gleason, A. Tank, M. Haylock,D. Collins, B. Trewin, F. Rahimzadeh, A. Tagipour, K. Kumar, J. Revadekar, G. Grif-fiths, L. Vincent, D. Stephenson, J. Burn, E. Aguilar, M. Brunet, M. Taylor, M. New,P. Zhai, M. Rusticucci, and J. Vazquez-Aguirre, 2006: Global observed changes in dailyclimate extremes of temperature and precipitation. JOURNAL OF GEOPHYSICALRESEARCH-ATMOSPHERES , 111, doi:10.1029/2005JD006290.
Allan, R. P. and B. J. Soden, 2008: Atmospheric warming and the amplification of pre-cipitation extremes. SCIENCE , 321, 1481–1484, doi:10.1126/science.1160787.
Allen, M. and W. Ingram, 2002: Constraints on future changes in climate and the hydro-logic cycle. NATURE , 419, 224+, doi:10.1038/nature01092.
Attema, J. J., J. M. Loriaux, and G. Lenderink, 2014: Extreme precipitation responseto climate perturbations in an atmospheric mesoscale model. ENVIRONMENTAL RE-SEARCH LETTERS , 9, doi:10.1088/1748-9326/9/1/014003.
Baldauf, M., A. Seifert, J. Foerstner, D. Majewski, M. Raschendorfer, and T. Reinhardt,2011: Operational Convective-Scale Numerical Weather Prediction with the COSMOModel: Description and Sensitivities. MONTHLY WEATHER REVIEW , 139, 3887–3905, doi:10.1175/MWR-D-10-05013.1.
Ban, N., J. Schmidli, and C. Schaer, 2014: Evaluation of the convection-resolving regionalclimate modeling approach in decade-long simulations. JOURNAL OF GEOPHYSI-CAL RESEARCH-ATMOSPHERES , 119, doi:10.1002/2014JD021478.
— 2015: Heavy precipitation in a changing climate: Does short-term summer precip-itation increase faster? GEOPHYSICAL RESEARCH LETTERS , 42, 1165–1172,doi:10.1002/2014GL062588.
Barnes, E. A. and L. Polvani, 2013: Response of the Midlatitude Jets, and of TheirVariability, to Increased Greenhouse Gases in the CMIP5 Models. JOURNAL OF CLI-MATE , 26, 7117–7135, doi:10.1175/JCLI-D-12-00536.1.
Barriopedro, D., E. M. Fischer, J. Luterbacher, R. Trigo, and R. Garcia-Herrera, 2011:The Hot Summer of 2010: Redrawing the Temperature Record Map of Europe. SCI-ENCE , 332, 220–224, doi:10.1126/science.1201224.
Bauer, P., A. Thorpe, and G. Brunet, 2015: The quiet revolution of numerical weatherprediction. NATURE , 525, 47–55, doi:10.1038/nature14956.
85
86 Bibliography
Belleflamme, A., X. Fettweis, and M. Erpicum, 2015: Do global warming-inducedcirculation pattern changes affect temperature and precipitation over Europe dur-ing summer? INTERNATIONAL JOURNAL OF CLIMATOLOGY , 35, 1484–1499,doi:10.1002/joc.4070.
Bellprat, O., S. Kotlarski, D. Lthi, and C. Schar, 2012: Objective calibration of regionalclimate models. JOURNAL OF GEOPHYSICAL RESEARCH: ATMOSPHERES ,117, n/a–n/a, doi:10.1029/2012JD018262, d23115.
Bellprat, O., S. Kotlarski, D. Luethi, and C. Schaer, 2012: Exploring Perturbed PhysicsEnsembles in a Regional Climate Model. JOURNAL OF CLIMATE , 25, 4582–4599,doi:10.1175/JCLI-D-11-00275.1.
Bengtsson, L., K. I. Hodges, and E. Roeckner, 2006: Storm tracks and climate change.JOURNAL OF CLIMATE , 19, 3518–3543, doi:10.1175/JCLI3815.1.
Beniston, M., D. B. Stephenson, O. B. Christensen, C. A. T. Ferro, C. Frei, S. Goyette,K. Halsnaes, T. Holt, K. Jylha, B. Koffi, J. Palutikof, R. Schoell, T. Semmler, andK. Woth, 2007: Future extreme events in European climate: an exploration of regionalclimate model projections. CLIMATIC CHANGE , 81, 71–95, doi:10.1007/s10584-006-9226-z.
Berg, P., J. O. Haerter, P. Thejll, C. Piani, S. Hagemann, and J. H. Christensen, 2009:Seasonal characteristics of the relationship between daily precipitation intensity andsurface temperature. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES ,114, doi:10.1029/2009JD012008.
Berg, P., C. Moseley, and J. O. Haerter, 2013: Strong increase in convective precip-itation in response to higher temperatures. NATURE GEOSCIENCE , 6, 181–185,doi:10.1038/ngeo1731.
Boe, J., 2013: Modulation of soil moisture-precipitation interactions over France by largescale circulation. CLIMATE DYNAMICS , 40, 875–892, doi:10.1007/s00382-012-1380-6.
Boe, J. and L. Terray, 2014: Land-sea contrast, soil-atmosphere and cloud-temperature in-teractions: interplays and roles in future summer European climate change. CLIMATEDYNAMICS , 42, 683–699, doi:10.1007/s00382-013-1868-8.
Boe, J., L. Terray, C. Cassou, and J. Najac, 2009: Uncertainties in European summerprecipitation changes: role of large scale circulation. CLIMATE DYNAMICS , 33, 265–276, doi:10.1007/s00382-008-0474-7.
Boer, G., N. McFarlane, R. Laprise, J. Henderson, and J. Blanchet, 1984: The cana-dian climate centre spectral atmospheric general circulation model. ATMOSPHERE-OCEAN , 22, 397–429, doi:10.1080/07055900.1984.9649208.
Boer, G. J., 2011: The ratio of land to ocean temperature change under global warming.Climate Dynamics , 37, 2253–2270, doi:10.1007/s00382-011-1112-3.URL http://dx.doi.org/10.1007/s00382-011-1112-3
Bibliography 87
Bony, S., R. Colman, V. M. Kattsov, R. P. Allan, C. S. Bretherton, J.-L. Dufresne, A. Hall,S. Hallegatte, M. M. Holland, W. Ingram, D. A. Randall, B. J. Soden, G. Tselioudis, andM. J. Webb, 2006: How well do we understand and evaluate climate change feedbackprocesses? JOURNAL OF CLIMATE , 19, 3445–3482, doi:10.1175/JCLI3819.1.
Bosshard, T., S. Kotlarski, T. Ewen, and C. Schar, 2011: Spectral representation ofthe annual cycle in the climate change signal. HYDROLOGY AND EARTH SYSTEMSCIENCES , 15, 2777–2788, doi:10.5194/hess-15-2777-2011.
Brayshaw, D. J., T. Woollings, and M. Vellinga, 2009: Tropical and Extratropical Re-sponses of the North Atlantic Atmospheric Circulation to a Sustained Weakening ofthe MOC. JOURNAL OF CLIMATE , 22, 3146–3155, doi:10.1175/2008JCLI2594.1.
Cattiaux, J., H. Douville, and Y. Peings, 2013a: European temperatures in CMIP5:origins of present-day biases and future uncertainties. CLIMATE DYNAMICS , 41,2889–2907, doi:10.1007/s00382-013-1731-y.
Cattiaux, J., H. Douville, A. Ribes, F. Chauvin, and C. Plante, 2013b: Towards a bet-ter understanding of changes in wintertime cold extremes over Europe: a pilot studywith CNRM and IPSL atmospheric models. CLIMATE DYNAMICS , 40, 2433–2445,doi:10.1007/s00382-012-1436-7.
Cattiaux, J., H. Douville, R. Schoetter, S. Parey, and P. Yiou, 2015: Projected increasein diurnal and interdiurnal variations of European summer temperatures. GEOPHYS-ICAL RESEARCH LETTERS , 42, 899–907, doi:10.1002/2014GL062531.
Catto, J. L., L. C. Shaffrey, and K. I. Hodges, 2011: Northern Hemisphere Extratrop-ical Cyclones in a Warming Climate in the HiGEM High-Resolution Climate Model.JOURNAL OF CLIMATE , 24, 5336–5352, doi:10.1175/2011JCLI4181.1.
Cess, R., G. Potter, J. Blanchet, G. Boer, A. Delgenio, M. Deque, V. Dymnikov,V. Galin, W. Gates, S. Ghan, J. Kiehl, A. Lacis, H. Letreut, Z. Li, X. Liang, B. Mca-vaney, V. Meleshko, J. Mitchel, J. Morcrette, D. Randall, L. Rikus, E. Roeckner,J. Royer, U. Schlese, D. Sheinin, A. Slingo, A. Sokolov, K. Taylor, W. Washing-ton, R. Wetherald, I. Yagai, and M. Zhang, 1990: Intercomparison and interpre-tation of climate feedback processes in 19 atmospheric general-circulation models.JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 95, 16601–16615,doi:10.1029/JD095iD10p16601.
Chan, S. C., E. J. Kendon, H. J. Fowler, S. Blenkinsop, C. A. T. Ferro, and D. B.Stephenson, 2012: Does increasing the spatial resolution of a regional climate modelimprove the simulated daily precipitation? CLIMATE DYNAMICS , 41, 1475–1495,doi:10.1007/s00382-012-1568-9.
Chan, S. C., E. J. Kendon, H. J. Fowler, S. Blenkinsop, N. M. Roberts, and C. A. T.Ferro, 2014: The Value of High-Resolution Met Office Regional Climate Models inthe Simulation of Multihourly Precipitation Extremes. JOURNAL OF CLIMATE , 27,6155–6174, doi:10.1175/JCLI-D-13-00723.1.
Chang, E., S. Lee, and K. Swanson, 2002: Storm track dynamics. JOURNAL OF CLI-MATE , 15, 2163–2183, doi:10.1175/1520-0442(2002)015¡02163:STD¿2.0.CO;2.
88 Bibliography
Chang, E. K. M., Y. Guo, and X. Xia, 2012: CMIP5 multimodel ensemble projec-tion of storm track change under global warming. JOURNAL OF GEOPHYSICALRESEARCH-ATMOSPHERES , 117, doi:10.1029/2012JD018578.
Chou, C., C.-A. Chen, P.-H. Tan, and K. T. Chen, 2012: Mechanisms for Global Warm-ing Impacts on Precipitation Frequency and Intensity. JOURNAL OF CLIMATE , 25,3291–3306, doi:10.1175/JCLI-D-11-00239.1.
Christensen, J., K. Krishna Kumar, E. Aldrian, S.-I. An, I. Cavalcanti, M. de Castro,W. Dong, P. Goswami, A. Hall, J. Kanyanga, A. Kitoh, J. Kossin, N.-C. Lau, J. Ren-wick, D. Stephenson, S.-P. Xie, and T. Zhou, 2013: Climate Phenomena and their Rel-evance for Future Regional Climate Change, Cambridge University Press, Cambridge,United Kingdom and New York, NY, USA, book section 14. 12171308.
Christensen, J. H. and O. B. Christensen, 2007: A summary of the PRUDENCE modelprojections of changes in European climate by the end of this century. CLIMATICCHANGE , 81, 7–30, doi:10.1007/s10584-006-9210-7.
Colle, B. A., Z. Zhang, K. A. Lombardo, E. Chang, P. Liu, and M. Zhang, 2013: HistoricalEvaluation and Future Prediction of Eastern North American and Western AtlanticExtratropical Cyclones in the CMIP5 Models during the Cool Season. JOURNAL OFCLIMATE , 26, 6882–6903, doi:10.1175/JCLI-D-12-00498.1.
Collins, M., R. Knutti, J. Arblaster, J.-L. Dufresne, T. Fichefet, P. Friedlingstein,X. Gao, W. Gutowski, T. Johns, G. Krinner, M. Shongwe, C. Tebaldi, A. Weaver,and M. Wehner, 2013: Long-term Climate Change: Projections, Commitments and Ir-reversibility , Cambridge University Press, Cambridge, United Kingdom and New York,NY, USA, book section 12. 10291136.
Davies, H. C., 1976: A lateral boundary formulation for multi-level prediction models.QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY , 102,405–418, doi:10.1002/qj.49710243210.
Davis, C. and K. Emanuel, 1991: Potential Vorticity Diagnostics of Cyclo-genesis. MONTHLY WEATHER REVIEW , 119, 1929–1953, doi:10.1175/1520-0493(1991)119¡1929:PVDOC¿2.0.CO;2.
DelGenio, A., A. Lacis, and R. Ruedy, 1991: Simulation of the effect of a warmer climateon atmospheric humidity. NATURE , 351, 382–385.
Della-Marta, P. M., M. R. Haylock, J. Luterbacher, and H. Wanner, 2007: Doubled lengthof western European summer heat waves since 1880. JOURNAL OF GEOPHYSICALRESEARCH-ATMOSPHERES , 112, doi:10.1029/2007JD008510.
Deser, C., R. Knutti, S. Solomon, and A. S. Phillips, 2012a: Communication of the role ofnatural variability in future North American climate. NATURE CLIMATE CHANGE ,2, 775–779, doi:10.1038/NCLIMATE1562.
Deser, C., A. Phillips, V. Bourdette, and H. Teng, 2012b: Uncertainty in climate changeprojections: the role of internal variability. CLIMATE DYNAMICS , 38, 527–546,doi:10.1007/s00382-010-0977-x.
Bibliography 89
Deser, C., L. Terray, and A. S. Phillips, 2016: Forced and internal components of winterair temperature trends over north america during the past 50 years: Mechanisms andimplications. JOURNAL OF CLIMATE , Submitted.
Dole, R., M. Hoerling, J. Perlwitz, J. Eischeid, P. Pegion, T. Zhang, X.-W. Quan, T. Xu,and D. Murray, 2011: Was there a basis for anticipating the 2010 Russian heat wave?GEOPHYSICAL RESEARCH LETTERS , 38, doi:10.1029/2010GL046582.
Doms, G., J. F orstner, E. Heise, H.-J. Herzog, D. Mironov, M. Raschendorfer, T. Rein-hardt, B. Ritter, R. Schrodin, J.-P. Schulz, and G. Vogel, 2011: A description of thenonhydro-static regional COSMO model. Part II: Physical parameterization, DeutscherWetterdienst, Offenbach, Germany.
Doms, G. and U. Schattler, 2015: A description of the nonhydro-static regional COSMOmodel. Part I: Dynamics and Numerics., Deutscher Wetterdienst, Offenbach, Germany.
Donat, M. G., L. V. Alexander, H. Yang, I. Durre, R. Vose, R. J. H. Dunn, K. M. Willett,E. Aguilar, M. Brunet, J. Caesar, B. Hewitson, C. Jack, A. M. G. K. Tank, A. C. Kruger,J. Marengo, T. C. Peterson, M. Renom, C. Oria Rojas, M. Rusticucci, J. Salinger,A. S. Elrayah, S. S. Sekele, A. K. Srivastava, B. Trewin, C. Villarroel, L. A. Vincent,P. Zhai, X. Zhang, and S. Kitching, 2013: Updated analyses of temperature and pre-cipitation extreme indices since the beginning of the twentieth century: The HadEX2dataset. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 118, 2098–2118, doi:10.1002/jgrd.50150.
Eady, E., 1949: Long Waves and Cyclone Waves. TELLUS , 1, 33–52.
Emori, S. and S. Brown, 2005: Dynamic and thermodynamic changes in mean and extremeprecipitation under changed climate. GEOPHYSICAL RESEARCH LETTERS , 32,doi:10.1029/2005GL023272.
Fasullo, J. T., 2010: Robust Land-Ocean Contrasts in Energy and Water Cycle Feedbacks.JOURNAL OF CLIMATE , 23, 4677–4693, doi:10.1175/2010JCLI3451.1.
Feser, F., B. Rockel, H. von Storch, J. Winterfeldt, and M. Zahn, 2011: Regionalclimate models add value to global model data a review and selected examples.BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY , 92, 1181–1192,doi:10.1175/2011BAMS3061.1.
Feudale, L. and J. Shukla, 2011: Influence of sea surface temperature on the Europeanheat wave of 2003 summer. Part I: an observational study. CLIMATE DYNAMICS ,36, 1691–1703, doi:10.1007/s00382-010-0788-0.
Field, C., V. Barros, T. Stocker, D. Qin, D. Dokken, K. Ebi, M. Mastrandrea, K. Mach,G.-K. Plattner, S. Allen, M. Tignor, and P. e. Midgley, 2012: Managing the risks ofextreme events and disasters to advance climate change adaptation. Technical report,IPCC.
Fischer, E. M., 2007: The role of landatmosphere interactions for european summer heatwaves: Past, present and future.
90 Bibliography
Fischer, E. M., U. Beyerle, and R. Knutti, 2013: Robust spatially aggregatedprojections of climate extremes. NATURE CLIMATE CHANGE , 3, 1033–1038,doi:10.1038/NCLIMATE2051.
Fischer, E. M., J. Rajczak, and C. Schar, 2012: Changes in European sum-mer temperature variability revisited. GEOPHYSICAL RESEARCH LETTERS , 39,doi:10.1029/2012GL052730.
Fischer, E. M. and C. Schar, 2009: Future changes in daily summer temperature variabil-ity: driving processes and role for temperature extremes. CLIMATE DYNAMICS , 33,917–935, doi:10.1007/s00382-008-0473-8.
— 2010: Consistent geographical patterns of changes in high-impact European heatwaves.NATURE GEOSCIENCE , 3, 398–403, doi:10.1038/NGEO866.
Fischer, E. M., J. Sedlacek, E. Hawkins, and R. Knutti, 2014: Models agree on forcedresponse pattern of precipitation and temperature extremes. GEOPHYSICAL RE-SEARCH LETTERS , 41, 8554–8562, doi:10.1002/2014GL062018.
Fischer, E. M., S. I. Seneviratne, P. L. Vidale, D. Luethi, and C. Schar, 2007: Soil moisture- Atmosphere interactions during the 2003 European summer heat wave. JOURNALOF CLIMATE , 20, 5081–5099, doi:10.1175/JCLI4288.1.
Folland, C. K., J. Knight, H. W. Linderholm, D. Fereday, S. Ineson, and J. W. Hurrell,2009: The Summer North Atlantic Oscillation: Past, Present, and Future. JOURNALOF CLIMATE , 22, 1082–1103, doi:10.1175/2008JCLI2459.1.
Fowler, H. J., M. Ekstrom, S. Blenkinsop, and A. P. Smith, 2007: Estimating change inextreme European precipitation using a multimodel ensemble. JOURNAL OF GEO-PHYSICAL RESEARCH-ATMOSPHERES , 112, doi:10.1029/2007JD008619.
Frei, C., J. Christensen, M. Deque, D. Jacob, R. Jones, and P. Vidale, 2003: Dailyprecipitation statistics in regional climate models: Evaluation and intercomparison forthe European Alps. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES ,108, doi:10.1029/2002JD002287.
Frei, C., C. Schar, D. Luthi, and H. Davies, 1998: Heavy precipitation processesin a warmer climate. GEOPHYSICAL RESEARCH LETTERS , 25, 1431–1434,doi:10.1029/98GL51099.
Frei, C., R. Scholl, S. Fukutome, J. Schmidli, and P. Vidale, 2006: Future changeof precipitation extremes in Europe: Intercomparison of scenarios from regional cli-mate models. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 111,doi:10.1029/2005JD005965.
Frierson, D. M. W., 2006: Robust increases in midlatitude static stability insimulations of global warming. GEOPHYSICAL RESEARCH LETTERS , 33,doi:10.1029/2006GL027504.
Bibliography 91
Froidevaux, P., L. Schlemmer, J. Schmidli, W. Langhans, and C. Schaer, 2014: Influenceof the Background Wind on the Local Soil Moisture-Precipitation Feedback. JOURNALOF THE ATMOSPHERIC SCIENCES , 71, 782–799, doi:10.1175/JAS-D-13-0180.1.
Gent, P. R., G. Danabasoglu, L. J. Donner, M. M. Holland, E. C. Hunke, S. R. Jayne,D. M. Lawrence, R. B. Neale, P. J. Rasch, M. Vertenstein, P. H. Worley, Z.-L. Yang,and M. Zhang, 2011: The Community Climate System Model Version 4. JOURNALOF CLIMATE , 24, 4973–4991, doi:10.1175/2011JCLI4083.1.
Gibelin, A. and M. Deque, 2003: Anthropogenic climate change over the Mediterraneanregion simulated by a global variable resolution model. CLIMATE DYNAMICS , 20,327–339, doi:10.1007/s00382-002-0277-1.
Gillett, N. P. and J. C. Fyfe, 2013: Annular mode changes in the CMIP5 simulations.GEOPHYSICAL RESEARCH LETTERS , 40, 1189–1193, doi:10.1002/grl.50249.
Giorgetta, M. A., J. Jungclaus, C. H. Reick, S. Legutke, J. Bader, M. Boettinger,V. Brovkin, T. Crueger, M. Esch, K. Fieg, K. Glushak, V. Gayler, H. Haak, H.-D.Hollweg, T. Ilyina, S. Kinne, L. Kornblueh, D. Matei, T. Mauritsen, U. Mikolajew-icz, W. Mueller, D. Notz, F. Pithan, T. Raddatz, S. Rast, R. Redler, E. Roeckner,H. Schmidt, R. Schnur, J. Segschneider, K. D. Six, M. Stockhause, C. Timmreck,J. Wegner, H. Widmann, K.-H. Wieners, M. Claussen, J. Marotzke, and B. Stevens,2013: Climate and carbon cycle changes from 1850 to 2100 in MPI-ESM simulationsfor the Coupled Model Intercomparison Project phase 5. JOURNAL OF ADVANCESIN MODELING EARTH SYSTEMS , 5, 572–597, doi:10.1002/jame.20038.
Giorgi, F., 2008: Regionalization of climate change information for impact assessmentand adaptation. WMO BULLETIN , 57(2), 86–92.
Giorgi, F., J. C, and A. G. R, 2009: Addressing climate information needs at the regionallevel: the CORDEX framework. WMO Bulletin, 58, 175–183.
Gordon, C. and W. Stern, 1982: A description of the gfdl global spec-tral model. MONTHLY WEATHER REVIEW , 110, 625–644, doi:10.1175/1520-0493(1982)110¡0625:ADOTGG¿2.0.CO;2.
Grasso, L., 2000: The differentiation between grid spacing and resolution and their appli-cation to numerical modeling. BULLETIN OF THE AMERICAN METEOROLOGI-CAL SOCIETY , 81, 579–580.
Gregory, J., J. Mitchell, and A. Brady, 1997: Summer drought in northern midlatitudesin a time-dependent CO2 climate experiment. JOURNAL OF CLIMATE , 10, 662–686,doi:10.1175/1520-0442(1997)010¡0662:SDINMI¿2.0.CO;2.
Grell, G., L. Schade, R. Knoche, A. Pfeiffer, and J. Egger, 2000: Nonhydrostatic cli-mate simulations of precipitation over complex terrain. JOURNAL OF GEOPHYSI-CAL RESEARCH-ATMOSPHERES , 105, 29595–29608, doi:10.1029/2000JD900445.
Grotch, S. and M. Maccracken, 1991: The use of general-circulation models to predictregional climatic-change. JOURNAL OF CLIMATE , 4, 286–303, doi:10.1175/1520-0442(1991)004¡0286:TUOGCM¿2.0.CO;2.
92 Bibliography
Guillod, B. P., B. Orlowsky, D. G. Miralles, A. J. Teuling, and S. I. Seneviratne, 2015:Reconciling spatial and temporal soil moisture effects on afternoon rainfall. NATURECOMMUNICATIONS , 6, doi:10.1038/ncomms7443.
Haarsma, R. J., F. Selten, and G. J. van Oldenborgh, 2013: Anthropogenic changesof the thermal and zonal flow structure over Western Europe and Eastern NorthAtlantic in CMIP3 and CMIP5 models. CLIMATE DYNAMICS , 41, 2577–2588,doi:10.1007/s00382-013-1734-8.
Hall, A., 2014: Projecting regional change. SCIENCE , 346, 1461–1462,doi:10.1126/science.aaa0629.
Hartmann, D., 1994: Global Physical Climatology . Academic Print, London, United King-dom, 411 pp.
Hartmann, D., A. Klein Tank, M. Rusticucci, L. Alexander, S. Bronnimann, Y. Charabi,F. Dentener, E. Dlugokencky, D. Easterling, A. Kaplan, B. Soden, P. Thorne, M. Wild,and P. Zhai, 2013: Observations: Atmosphere and Surface, Cambridge University Press,Cambridge, United Kingdom and New York, NY, USA, book section 2. 159254.
Harvey, B. J., L. C. Shaffrey, T. J. Woollings, G. Zappa, and K. I. Hodges, 2012: Howlarge are projected 21st century storm track changes? GEOPHYSICAL RESEARCHLETTERS , 39, doi:10.1029/2012GL052873.
Hassanzadeh, H., J. Schmidli, W. Langhans, L. Schlemmer, and C. Schar, 2014: Im-pact of topography on the diurnal cycle of summertime moist convection in idealizedsimulations. METEOROLOGISCHE ZEITSCHRIFT , –, doi:10.1127/metz/2015/0653.
Hawkins, E. and R. Sutton, 2009: The potential to narrow uncertainty in regional climatepredictions. BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY , 90,1095+, doi:10.1175/2009BAMS2607.1.
— 2011: The potential to narrow uncertainty in projections of regional precipitationchange. CLIMATE DYNAMICS , 37, 407–418, doi:10.1007/s00382-010-0810-6.
Held, I. and B. Soden, 2000: Water vapor feedback and global warming. AN-NUAL REVIEW OF ENERGY AND THE ENVIRONMENT , 25, 441–475,doi:10.1146/annurev.energy.25.1.441.
Held, I. M. and B. J. Soden, 2006: Robust responses of the hydrological cycle to globalwarming. JOURNAL OF CLIMATE , 19, 5686–5699, doi:10.1175/JCLI3990.1.
Hohenegger, C., P. Brockhaus, C. S. Bretherton, and C. Schaer, 2009: The Soil Moisture-Precipitation Feedback in Simulations with Explicit and Parameterized Convection.JOURNAL OF CLIMATE , 22, 5003–5020, doi:10.1175/2009JCLI2604.1.
Hurrell, J. W., Y. Kushnir, G. Ottersen, and M. Visbeck, 2003: The North AtlanticOscillation Climatic Significance and Environmental Impact . American GeophysicalUnion, Washington DC, USA.
Bibliography 93
Huth, R., C. Beck, A. Philipp, M. Demuzere, Z. Ustrnul, M. Cahynova, J. Kysely, andO. E. Tveito, 2008: Classifications of Atmospheric Circulation Patterns Recent Ad-vances and Applications. TRENDS AND DIRECTIONS IN CLIMATE RESEARCH ,Gimeno, L and GarciaHerrera, R and Trigo, RM, ed., European Meteorol Soc, volume1146 of Annals of the New York Academy of Sciences , 105–152, 7th Annual Meeting ofthe European-Meteorology-Society (EMS) / 8th European Conference on Applicationsof Meteorology, San Lorenzo de El Escorial, SPAIN, OCT 01-05, 2007.
Im, E. S., E. Coppola, F. Giorgi, and X. Bi, 2010: Local effects of climate changeover the Alpine region: A study with a high resolution regional climate model witha surrogate climate change scenario. GEOPHYSICAL RESEARCH LETTERS , 37,doi:10.1029/2009GL041801.
IPCC, 2013: Annex I: Atlas of Global and Regional Climate Projections , Cambridge Uni-versity Press, Cambridge, United Kingdom and New York, NY, USA, book section AI.13111394.
— 2014: Climate Change 2014: Impacts, Adaptation, and Vulnerability. Part B: RegionalAspects. Contribution of Working Group II to the Fifth Assessment Report of the In-tergovernmental Panel on Climate Change. Cambridge University Press, Cambridge,United Kingdom and New York, NY, USA, 688 pp.
Jacob, D., J. Petersen, B. Eggert, A. Alias, O. Christensen, L. Bouwer, A. Braun, A. Co-lette, M. Dqu, G. Georgievski, E. Georgopoulou, A. Gobiet, L. Menut, G. Nikulin,A. Haensler, N. Hempelmann, C. Jones, K. Keuler, S. Kovats, N. Kroner, S. Kotlarski,A. Kriegsmann, E. Martin, E. Meijgaard, C. Moseley, S. Pfeifer, S. Preuschmann,C. Radermacher, K. Radtke, D. Rechid, M. Rounsevell, P. Samuelsson, S. Somot, J.-F. Soussana, C. Teichmann, R. Valentini, R. Vautard, B. Weber, and P. Yiou, 2014:Euro-cordex: new high-resolution climate change projections for european impact re-search. REGIONAL ENVIRONMENTAL CHANGE , 14, 563–578, doi:10.1007/s10113-013-0499-2.
Jaeger, E. B., I. Anders, D. Luethi, B. Rockel, C. Schaer, and S. I. Seneviratne,2008: Analysis of ERA40-driven CLM simulations for Europe. METEOROLOGISCHEZEITSCHRIFT , 17, 349–367, doi:10.1127/0941-2948/2008/0301.
Jaeger, E. B. and S. I. Seneviratne, 2011: Impact of soil moisture-atmosphere couplingon European climate extremes and trends in a regional climate model. CLIMATE DY-NAMICS , 36, 1919–1939, doi:10.1007/s00382-010-0780-8.
Johnson, N. C. and S.-P. Xie, 2010: Changes in the sea surface temperature threshold fortropical convection. NATURE GEOSCIENCE , 3, 842–845, doi:10.1038/NGEO1008.
Joshi, M. M., J. M. Gregory, M. J. Webb, D. M. H. Sexton, and T. C. Johns, 2008:Mechanisms for the land/sea warming contrast exhibited by simulations of climatechange. CLIMATE DYNAMICS , 30, 455–465, doi:10.1007/s00382-007-0306-1.
Keller, M., O. Fuhrer, J. Schmidli, M. Stengel, R. Stockli, and C. Schar, 2015: Evalua-tion of convection-resolving models using satellite data: The diurnal cycle of summerconvection over the Alps. METEOROLOGISCHE ZEITSCHRIFT , published online.
94 Bibliography
Kendon, E. J., N. M. Roberts, C. A. Senior, and M. J. Roberts, 2012: Realism of Rainfallin a Very High-Resolution Regional Climate Model. JOURNAL OF CLIMATE , 25,5791–5806, doi:10.1175/JCLI-D-11-00562.1.
Kendon, E. J., D. P. Rowell, and R. G. Jones, 2010: Mechanisms and reliability offuture projected changes in daily precipitation. CLIMATE DYNAMICS , 35, 489–509,doi:10.1007/s00382-009-0639-z.
Keuler, K., K. Radtke, S. Kotlarski, and D. Luthi, 2016: Regional climate change overEurope in COSMO-CLM: Influence of emission scenario and driving global model. ME-TEOROLOGISCHE ZEITSCHRIFT , accepted.
Kharin, V. V., F. W. Zwiers, X. Zhang, and M. Wehner, 2013: Changes in temperatureand precipitation extremes in the CMIP5 ensemble. CLIMATIC CHANGE , 119, 345–357, doi:10.1007/s10584-013-0705-8.
Kjellstrom, E., L. Barring, D. Jacob, R. Jones, G. Lenderink, and C. Schar, 2007: Mod-elling daily temperature extremes: recent climate and future changes over Europe.CLIMATIC CHANGE , 81, 249–265, doi:10.1007/s10584-006-9220-5.
Knutti, R., 2010: The end of model democracy? CLIMATIC CHANGE , 102, 395–404,doi:10.1007/s10584-010-9800-2.
Knutti, R., R. Furrer, C. Tebaldi, J. Cermak, and G. A. Meehl, 2010: Challenges inCombining Projections from Multiple Climate Models. JOURNAL OF CLIMATE , 23,2739–2758, doi:10.1175/2009JCLI3361.1.
Knutti, R. and J. Sedlacek, 2013: Robustness and uncertainties in the newCMIP5 climate model projections. NATURE CLIMATE CHANGE , 3, 369–373,doi:10.1038/NCLIMATE1716.
Koffi, B. and E. Koffi, 2008: Heat waves across Europe by the end of the 21stcentury: multiregional climate simulations. CLIMATE RESEARCH , 36, 153–168,doi:10.3354/cr00734.
Koster, R., P. Dirmeyer, Z. Guo, G. Bonan, E. Chan, P. Cox, C. Gordon, S. Kanae,E. Kowalczyk, D. Lawrence, P. Liu, C. Lu, S. Malyshev, B. McAvaney, K. Mitchell,D. Mocko, T. Oki, K. Oleson, A. Pitman, Y. Sud, C. Taylor, D. Verseghy, R. Vasic,Y. Xue, T. Yamada, and G. Team, 2004: Regions of strong coupling between soilmoisture and precipitation. SCIENCE , 305, 1138–1140, doi:10.1126/science.1100217.
Kotlarski, S., K. Keuler, O. B. Christensen, A. Colette, M. Deque, A. Gobiet, K. Goergen,D. Jacob, D. Luethi, E. van Meijgaard, G. Nikulin, C. Schar, C. Teichmann, R. Vautard,K. Warrach-Sagi, and V. Wulfmeyer, 2014: Regional climate modeling on Europeanscales: a joint standard evaluation of the EURO-CORDEX RCM ensemble. GEOSCI-ENTIFIC MODEL DEVELOPMENT , 7, 1297–1333, doi:10.5194/gmd-7-1297-2014.
Kroner, N., S. Kotlarski, E. Fischer, D. Luthi, E. Zubler, and C. Schar, 2016: Separatingclimate change signals into thermodynamic, lapse-rate and circulation effects: the-ory and application to the european summer climate. CLIMATE DYNAMICS , 1–16,doi:10.1007/s00382-016-3276-3.
Bibliography 95
Lambert, F. H., M. J. Webb, and M. M. Joshi, 2011: The Relationship between Land-Ocean Surface Temperature Contrast and Radiative Forcing. JOURNAL OF CLI-MATE , 24, 3239–3256, doi:10.1175/2011JCLI3893.1.
Laprise, R., 2008: Regional climate modelling. JOURNAL OF COMPUTATIONALPHYSICS , 227, 3641–3666, doi:10.1016/j.jcp.2006.10.024.
Legras, B. and M. Ghil, 1985: Persistent anomalies, blocking and variations in atmo-spheric predictability. JOURNAL OF THE ATMOSPHERIC SCIENCES , 42, 433–471,doi:10.1175/1520-0469(1985)042¡0433:PABAVI¿2.0.CO;2.
Lenderink, G. and E. Van Meijgaard, 2008: Increase in hourly precipitation extremesbeyond expectations from temperature changes. NATURE GEOSCIENCE , 1, 511–514,doi:10.1038/ngeo262.
Leutwyler, D., O. Fuhrer, L. X., D. Luthi, and C. Schar, 2016: Towards european-scaleconvection-resolving climate simulations on gpus. GEOPHYSICAL MODEL DEVEL-OPMENT , In preparation.
Levine, X. J. and T. Schneider, 2015: Baroclinic Eddies and the Extent of the HadleyCirculation: An Idealized GCM Study. JOURNAL OF THE ATMOSPHERIC SCI-ENCES , 72, 2744–2761, doi:10.1175/JAS-D-14-0152.1.
Lorenz, E., 1963: Deterministic nonperiodic flow. JOURNAL OF THE ATMOSPHERICSCIENCES , 20, 130–141, doi:10.1175/1520-0469(1963)020¡0130:DNF¿2.0.CO;2.
Lorenz, E. N., 1970: Climatic Change as a Mathematical Problem. JOUR-NAL OF APPLIED METEOROLOGY , 9, 325–329, doi:10.1175/1520-0450(1970)009¡0325:CCAAMP¿2.0.CO;2.
Loriaux, J. M., G. Lenderink, S. R. De Roode, and A. P. Siebesma, 2013: Understand-ing Convective Extreme Precipitation Scaling Using Observations and an EntrainingPlume Model. JOURNAL OF THE ATMOSPHERIC SCIENCES , 70, 3641–3655,doi:10.1175/JAS-D-12-0317.1.
Lu, J., G. Chen, and D. M. W. Frierson, 2008: Response of the Zonal Mean AtmosphericCirculation to El Nino versus Global Warming. JOURNAL OF CLIMATE , 21, 5835–5851, doi:10.1175/2008JCLI2200.1.
Lu, J., G. A. Vecchi, and T. Reichler, 2007: Expansion of the Hadley cell under globalwarming. GEOPHYSICAL RESEARCH LETTERS , 34, doi:10.1029/2006GL028443.
Manabe, S., K. Bryan, and M. Spelman, 1990: Transient-response of aglobal ocean atmosphere model to a doubling of atmospheric carbon-dioxide.JOURNAL OF PHYSICAL OCEANOGRAPHY , 20, 722–749, doi:10.1175/1520-0485(1990)020¡0722:TROAGO¿2.0.CO;2.
Manabe, S. and J. Smagorin, 1967: Simulated climatology of a general circulation modelwith a hydrologic cycle .2. analysis of tropical atmosphere. MONTHLY WEATHERREVIEW , 95, 155–&, doi:10.1175/1520-0493(1967)095¡0155:SCOAGC¿2.3.CO;2.
96 Bibliography
Manabe, S., M. Spelman, and R. Stouffer, 1992: Transient responses of acoupled ocean atmosphere model to gradual changes of atmospheric co2 .2.seasonal response. JOURNAL OF CLIMATE , 5, 105–126, doi:10.1175/1520-0442(1992)005¡0105:TROACO¿2.0.CO;2.
Manabe, S. and R. Stouffer, 1980: Sensitivity of a global climate model to an increase ofco2 concentration in the atmosphere. JOURNAL OF GEOPHYSICAL RESEARCH-OCEANS AND ATMOSPHERES , 85, 5529–5554, doi:10.1029/JC085iC10p05529.
Manabe, S., R. Stouffer, M. Spelman, and K. Bryan, 1991: Transient responsesof a coupled ocean atmosphere model to gradual changes of atmospheric co2 .1.annual mean response. JOURNAL OF CLIMATE , 4, 785–818, doi:10.1175/1520-0442(1991)004¡0785:TROACO¿2.0.CO;2.
Mariotti, A. and A. Dell’Aquila, 2012: Decadal climate variability in the Mediterraneanregion: roles of large-scale forcings and regional processes. CLIMATE DYNAMICS ,38, 1129–1145, doi:10.1007/s00382-011-1056-7.
Mbengue, C. and T. Schneider, 2013: Storm Track Shifts under Climate Change: WhatCan Be Learned from Large-Scale Dry Dynamics. JOURNAL OF CLIMATE , 26, 9923–9930, doi:10.1175/JCLI-D-13-00404.1.
McDonald, R. E., 2011: Understanding the impact of climate change on North-ern Hemisphere extra-tropical cyclones. CLIMATE DYNAMICS , 37, 1399–1425,doi:10.1007/s00382-010-0916-x.
McGregor, J. L. and M. R. Dix, 2008: An Updated Description of the Conformal-CubicAtmospheric Model , Springer Science+Business Media, chapter 4 in High ResolutionNumerical Modelling of the Atmosphere and Ocean.
Meehl, G. A., 1990: Development of global coupled ocean-atmosphere general circulationmodels. CLIMATE DYNAMICS , 5, 19–33, doi:10.1007/BF00195851.URL http://dx.doi.org/10.1007/BF00195851
Meehl, G. A., C. Covey, T. Delworth, M. Latif, B. McAvaney, J. F. B. Mitchell, R. J.Stouffer, and K. E. Taylor, 2007: The WCRP CMIP3 multimodel dataset - A new erain climate change research. BULLETIN OF THE AMERICAN METEOROLOGICALSOCIETY , 88, 1383+, doi:10.1175/BAMS-88-9-1383.
Meehl, G. A. and T. F. Stocker, 2007: Global Climate Projections. CLIMATE CHANGE2007: THE PHYSICAL SCIENCE BASIS , Solomon, S and Qin, D and Manning, Mand Marquis, M and Averyt, K and Tignor, MMB and Miller, HL and Chen, ZL, ed.,747–845.
Miralles, D. G., A. J. Teuling, C. C. van Heerwaarden, and J. V.-G. de Arellano, 2014:Mega-heatwave temperatures due to combined soil desiccation and atmospheric heataccumulation. NATURE GEOSCIENCE , 7, 345–349, doi:10.1038/NGEO2141.
Mitas, C. and A. Clement, 2006: Recent behavior of the Hadley cell and tropical thermo-dynamics in climate models and reanalyses. GEOPHYSICAL RESEARCH LETTERS ,33, doi:10.1029/2005GL024406.
Bibliography 97
Montesarchio, M., A. L. Zollo, E. Bucchignani, P. Mercogliano, and S. Castellari, 2014:Performance evaluation of high-resolution regional climate simulations in the Alpinespace and analysis of extreme events. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 119, 3222–3237, doi:10.1002/2013JD021105.
Moss, R., M. Babiker, S. Brinkman, E. Calvo, T. Carter, J. Edmonds, I. Elgizoul,S. Emori, L. Erda, K. Hibbard, R. Jones, M. Kainuma, J. Kelleher, J. Lamarque,M. Manning, B. Matthews, J. Meehl, L. Meyer, J. Mitchell, N. Nakicenovic, B. ONeill,R. Pichs, K. Riahi, S. Rose, P. Runci, R. Stouffer, D. van Vuuren, J. Weyant,T. Wilbanks, J. van Ypersele, and M. Zurek, 2008: Towards new scenarios for anal-ysis of emissions, climate change, impacts, and response strategies. Technical Report978-92-9169-124-1, Intergovernmental Panel on Climate Change, Geneva.
Moss, R. H., J. A. Edmonds, K. A. Hibbard, M. R. Manning, S. K. Rose, D. P. vanVuuren, T. R. Carter, S. Emori, M. Kainuma, T. Kram, G. A. Meehl, J. F. B. Mitchell,N. Nakicenovic, K. Riahi, S. J. Smith, R. J. Stouffer, A. M. Thomson, J. P. Weyant,and T. J. Wilbanks, 2010: The next generation of scenarios for climate change researchand assessment. NATURE , 463, 747–756, doi:10.1038/nature08823.
Muller, C. J., P. A. O’Gorman, and L. E. Back, 2011: Intensification of PrecipitationExtremes with Warming in a Cloud-Resolving Model. JOURNAL OF CLIMATE , 24,2784–2800, doi:10.1175/2011JCLI3876.1.
Nikulin, G., E. Kjellstrom, U. Hansson, G. Strandberg, and A. Ullerstig, 2011:Evaluation and future projections of temperature, precipitation and wind ex-tremes over Europe in an ensemble of regional climate simulations. TELLUSSERIES A-DYNAMIC METEOROLOGY AND OCEANOGRAPHY , 63, 41–55,doi:10.1111/j.1600-0870.2010.00466.x.
O’Gorman, P. A., R. P. Allan, M. P. Byrne, and M. Previdi, 2012: Energetic Constraintson Precipitation Under Climate Change. SURVEYS IN GEOPHYSICS , 33, 585–608,doi:10.1007/s10712-011-9159-6.
O’Gorman, P. A. and T. Schneider, 2009: The physical basis for increases in precipitationextremes in simulations of 21st-century climate change. PROCEEDINGS OF THENATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA,106, 14773–14777, doi:10.1073/pnas.0907610106.
Orlowsky, B. and S. I. Seneviratne, 2012: Global changes in extreme events: regionaland seasonal dimension. CLIMATIC CHANGE , 110, 669–696, doi:10.1007/s10584-011-0122-9.
Pal, J. and E. Eltahir, 2001: Pathways relating soil moisture conditions to future summerrainfall within a model of the land-atmosphere system. JOURNAL OF CLIMATE , 14,1227–1242, doi:10.1175/1520-0442(2001)014¡1227:PRSMCT¿2.0.CO;2.
Pall, P., M. R. Allen, and D. A. Stone, 2007: Testing the Clausius-Clapeyron constrainton changes in extreme precipitation under CO2 warming. CLIMATE DYNAMICS , 28,351–363, doi:10.1007/s00382-006-0180-2.
98 Bibliography
Parey, S., D. Dacunha-Castelle, and T. T. H. Hoang, 2010: Mean and variance evolutionsof the hot and cold temperatures in Europe. CLIMATE DYNAMICS , 34, 345–359,doi:10.1007/s00382-009-0557-0.
Petterss, S. and S. Smebye, 1971: Development of Extratropical Cyclones. QUAR-TERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY , 97, 457–&,doi:10.1002/qj.49709741407.
Plavcova, E. and J. Kysely, 2013: Projected evolution of circulation types and theirtemperatures over Central Europe in climate models. THEORETICAL AND APPLIEDCLIMATOLOGY , 114, 625–634, doi:10.1007/s00704-013-0874-4.
Quesada, B., R. Vautard, P. Yiou, M. Hirschi, and S. I. Seneviratne, 2012: AsymmetricEuropean summer heat predictability from wet and dry southern winters and springs.NATURE CLIMATE CHANGE , 2, 736–741, doi:10.1038/NCLIMATE1536.
Radermacher, C. and L. Tomassini, 2012: Thermodynamic Causes for Future Trends inHeavy Precipitation over Europe Based on an Ensemble of Regional Climate Model Sim-ulations. JOURNAL OF CLIMATE , 25, 7669–7689, doi:10.1175/JCLI-D-11-00304.1.
Rajczak, J., P. Pall, and C. Schar, 2013: Projections of extreme precipitation events inregional climate simulations for Europe and the Alpine Region. JOURNAL OF GEO-PHYSICAL RESEARCH-ATMOSPHERES , 118, 3610–3626, doi:10.1002/jgrd.50297.
Rasmussen, R., C. Liu, K. Ikeda, D. Gochis, D. Yates, F. Chen, M. Tewari, M. Bar-lage, J. Dudhia, W. Yu, K. Miller, K. Arsenault, V. Grubisic, G. Thompson, andE. Gutmann, 2011: High-Resolution Coupled Climate Runoff Simulations of SeasonalSnowfall over Colorado: A Process Study of Current and Warmer Climate. JOURNALOF CLIMATE , 24, 3015–3048, doi:10.1175/2010JCLI3985.1.
Raval, A. and V. Ramanthan, 1989: Observational determination of the greenhouse-effect.NATURE , 342, 758–761, doi:10.1038/342758a0.
Rebetez, M., O. Dupont, and M. Giroud, 2009: An analysis of the July 2006 heatwave ex-tent in Europe compared to the record year of 2003. THEORETICAL AND APPLIEDCLIMATOLOGY , 95, 1–7, doi:10.1007/s00704-007-0370-9.
Reinhardt, T. and A. Seifert, 2006: A three-category ice scheme for the LMK. COSMONewsletter , 6, 115 120.URL http://www.cosmomodel.org
Riediger, U. and A. Gratzki, 2014: Future weather types and their influence on mean andextreme climate indices for precipitation and temperature in Central Europe. METE-OROLOGISCHE ZEITSCHRIFT , 23, 231–252, doi:10.1127/0941-2948/2014/0519.
Rind, D., R. Goldberg, J. Hansen, C. Rosenzweig, and R. Ruedy, 1990: Potential evap-otranspiration and the likelihood of future drought. JOURNAL OF GEOPHYSICALRESEARCH-ATMOSPHERES , 95, 9983–10004, doi:10.1029/JD095iD07p09983.
Bibliography 99
Ritter, B. and J. Geleyn, 1992: A comprehensive radiation scheme for nu-merical weather prediction models with potential applications in climate sim-ulations. MONTHLY WEATHER REVIEW , 120, 303–325, doi:10.1175/1520-0493(1992)120¡0303:ACRSFN¿2.0.CO;2.
Rockel, B., A. Will, and A. Hense, 2008: The Regional Climate ModelCOSMO-CLM(CCLM). METEOROLOGISCHE ZEITSCHRIFT , 17, 347–348,doi:10.1127/0941-2948/2008/0309.
Ross, R., W. Elliott, D. Seidel, and P. A.-I. M. Grp, 2002: Lower-tropospheric humidity-temperature relationships in radiosonde observations and atmospheric general circula-tion models. JOURNAL OF HYDROMETEOROLOGY , 3, 26–38, doi:10.1175/1525-7541(2002)003¡0026:LTHTRI¿2.0.CO;2.
Rowell, D. P., 2012: Sources of uncertainty in future changes in local precipitation. CLI-MATE DYNAMICS , 39, 1929–1950, doi:10.1007/s00382-011-1210-2.
Rowell, D. P. and R. G. Jones, 2006: Causes and uncertainty of future summer dryingover Europe. CLIMATE DYNAMICS , 27, 281–299, doi:10.1007/s00382-006-0125-9.
Rummukainen, M., 2010: State-of-the-art with regional climate models. WILEY INTER-DISCIPLINARY REVIEWS-CLIMATE CHANGE , 1, 82–96, doi:10.1002/wcc.8.
Ruscica, R. C., A. A. Srensson, and C. G. Menndez, 2015: Pathways between soil moistureand precipitation in southeastern south america. Atmospheric Science Letters , 16, 267–272, doi:10.1002/asl2.552.URL http://dx.doi.org/10.1002/asl2.552
Sanchez-Gomez, E., S. Somot, and A. Mariotti, 2009: Future changes in the Mediter-ranean water budget projected by an ensemble of regional climate models. GEOPHYS-ICAL RESEARCH LETTERS , 36, doi:10.1029/2009GL040120.
Sanderson, B. M. and R. Knutti, 2012: On the interpretation of constrained climate modelensembles. GEOPHYSICAL RESEARCH LETTERS , 39, doi:10.1029/2012GL052665.
Sanderson, B. M., R. Knutti, and P. Caldwell, 2015: A Representative Democracy toReduce Interdependency in a Multimodel Ensemble. JOURNAL OF CLIMATE , 28,5171–5194, doi:10.1175/JCLI-D-14-00362.1.
Santer, B. D., P. W. Thorne, L. Haimberger, K. E. Taylor, T. M. L. Wigley, J. R. Lanzante,S. Solomon, M. Free, P. J. Gleckler, P. D. Jones, T. R. Karl, S. A. Klein, C. Mears,D. Nychka, G. A. Schmidt, S. C. Sherwood, and F. J. Wentz, 2008: Consistency of mod-elled and observed temperature trends in the tropical troposphere. INTERNATIONALJOURNAL OF CLIMATOLOGY , 28, 1703–1722, doi:10.1002/joc.1756.
Santos, J. A., M. Belo-Pereira, H. Fraga, and J. G. Pinto, 2016: Understand-ing climate change projections for precipitation over western europe with aweather typing approach. Journal of Geophysical Research: Atmospheres , n/a–n/a,doi:10.1002/2015JD024399, 2015JD024399.
100 Bibliography
Schar, C., C. Frei, D. Luthi, and H. Davies, 1996: Surrogate climate-change scenariosfor regional climate models. GEOPHYSICAL RESEARCH LETTERS , 23, 669–672,doi:10.1029/96GL00265.
Schar, C. and N. Kroner, 2016: Sequential factor separation for the analysis of weather andclimate model results. JOURNAL OF THE ATMOSPHERIC SCIENCES , submitted.
Schar, C., D. Luthi, U. Beyerle, and E. Heise, 1999: The soil-precipitation feedback: Aprocess study with a regional climate model. JOURNAL OF CLIMATE , 12, 722–741,doi:10.1175/1520-0442(1999)012¡0722:TSPFAP¿2.0.CO;2.
Schar, C., P. Vidale, D. Luthi, C. Frei, C. Haberli, M. Liniger, and C. Appenzeller,2004: The role of increasing temperature variability in European summer heatwaves.NATURE , 427, 332–336, doi:10.1038/nature02300.
Scheff, J. and D. M. W. Frierson, 2012: Robust future precipitation declines in CMIP5largely reflect the poleward expansion of model subtropical dry zones. GEOPHYSICALRESEARCH LETTERS , 39, doi:10.1029/2012GL052910.
Schlemmer, L., C. Hohenegger, J. Schmidli, C. S. Bretherton, and C. Schaer, 2011: AnIdealized Cloud-Resolving Framework for the Study of Midlatitude Diurnal Convec-tion over Land. JOURNAL OF THE ATMOSPHERIC SCIENCES , 68, 1041–1057,doi:10.1175/2010JAS3640.1.
Schmidli, J., C. Frei, and P. Vidale, 2006: Downscaling from GC precipitation: A bench-mark for dynamical and statistical downscaling methods. INTERNATIONAL JOUR-NAL OF CLIMATOLOGY , 26, 679–689, doi:10.1002/joc.1287.
Schneider, T., T. Bischoff, and H. Plotka, 2015: Physics of Changes in Synop-tic Midlatitude Temperature Variability. JOURNAL OF CLIMATE , 28, 2312–2331,doi:10.1175/JCLI-D-14-00632.1.
Schneider, T., P. A. O’Gorman, and X. J. Levine, 2010: Water vapor and the dynamicsof climate changes. Reviews of Geophysics , 48, n/a–n/a, doi:10.1029/2009RG000302,rG3001.URL http://dx.doi.org/10.1029/2009RG000302
Schrum, C., U. Hubner, D. Jacob, and R. Podzun, 2003: A coupled atmosphere/ice/oceanmodel for the North Sea and the Baltic Sea. CLIMATE DYNAMICS , 21, 131–151,doi:10.1007/s00382-003-0322-8.
Seager, R., H. Liu, N. Henderson, I. Simpson, C. Kelley, T. Shaw, Y. Kushnir, andM. Ting, 2014: Causes of Increasing Aridification of the Mediterranean Region inResponse to Rising Greenhouse Gases. JOURNAL OF CLIMATE , 27, 4655–4676,doi:10.1175/JCLI-D-13-00446.1.
Seager, R., N. Naik, and G. A. Vecchi, 2010: Thermodynamic and Dynamic Mechanismsfor Large-Scale Changes in the Hydrological Cycle in Response to Global Warming.JOURNAL OF CLIMATE , 23, 4651–4668, doi:10.1175/2010JCLI3655.1.
Bibliography 101
Seidel, D. J., Q. Fu, W. J. Randel, and T. J. Reichler, 2008: Widening of the tropical beltin a changing climate. NATURE GEOSCIENCE , 1, 21–24, doi:10.1038/ngeo.2007.38.
Sejas, S. A., O. S. Albert, M. Cai, and Y. Deng, 2014: Feedback attribution of the land-seawarming contrast in a global warming simulation of the ncar ccsm4. ENVIRONMEN-TAL RESEARCH LETTERS , 9, 124005.
Seneviratne, S., N. Nicholls, D. Easterling, C. Goodess, S. Kanae, J. Kossin, Y. Luo,J. Marengo, K. McInnes, M. Rahimi, M. Reichstein, A. Sorteberg, C. Vera, andX. Zhang, 2012: Changes in climate extremes and their impacts on the natural physicalenvironment. A Special Report of Working Groups I and II of the IntergovernmentalPanel on Climate Change (IPCC), Cambridge University Press.
Seneviratne, S., J. Pal, E. Eltahir, and C. Schar, 2002: Summer dryness in a warmerclimate: a process study with a regional climate model. CLIMATE DYNAMICS , 20,69–85, doi:10.1007/s00382-002-0258-4.
Seneviratne, S. I., T. Corti, E. L. Davin, M. Hirschi, E. B. Jaeger, I. Lehner,B. Orlowsky, and A. J. Teuling, 2010: Investigating soil moisture-climate interac-tions in a changing climate: A review. EARTH-SCIENCE REVIEWS , 99, 125–161,doi:10.1016/j.earscirev.2010.02.004.
Seneviratne, S. I., R. D. Koster, Z. Guo, P. A. Dirmeyer, E. Kowalczyk, D. Lawrence,P. Liu, C.-H. Lu, D. Mocko, K. W. Oleson, and D. Verseghy, 2006a: Soil mois-ture memory in AGCM simulations: Analysis of global land-atmosphere coupling ex-periment (GLACE) data. JOURNAL OF HYDROMETEOROLOGY , 7, 1090–1112,doi:10.1175/JHM533.1.
Seneviratne, S. I., D. Luethi, M. Litschi, and C. Schaer, 2006b: Land-atmosphere couplingand climate change in Europe. NATURE , 443, 205–209, doi:10.1038/nature05095.
Sevault, F., S. Somot, A. Alias, C. Dubois, C. Lebeaupin-Brossier, P. Nabat, F. Adloff,M. Dqu, and B. Decharme, 2014: A fully coupled mediterranean regional climate systemmodel: design and evaluation of the ocean component for the 1980-2012 period. TellusA, 66.
Shepherd, T. G., 2014: Atmospheric circulation as a source of uncertainty in climatechange projections. NATURE GEOSCIENCE , 7, 703–708, doi:10.1038/NGEO2253.
Sherwood, S. and Q. Fu, 2014: A Drier Future? SCIENCE , 343, 737–739,doi:10.1126/science.1247620.
Sillmann, J., V. V. Kharin, F. W. Zwiers, X. Zhang, and D. Bronaugh, 2013: Climateextremes indices in the CMIP5 multimodel ensemble: Part 2. Future climate pro-jections. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 118, 2473–2493, doi:10.1002/jgrd.50188.
Soden, B., D. Jackson, V. Ramaswamy, M. Schwarzkopf, and X. Huang, 2005: Theradiative signature of upper tropospheric moistening. SCIENCE , 310, 841–844,doi:10.1126/science.1115602.
102 Bibliography
Solomon, S. a. D. Q., M. Manning, Z. Chen, M. Marquis, K. Averyt, M. Tignor, andH. Miller, 2007: Climate Change 2007: The Physical Science Basis. Contribution ofWorking Group I to the Fourth Assessment Report of the Intergovernmental Panel onClimate Change. Cambridge University Press.
Sorland, S. and A. Sorteberg, 2015: Low-pressure systems and extreme precipitationin central india: sensitivity to temperature changes. CLIMATE DYNAMICS , 1–16,doi:10.1007/s00382-015-2850-4.
Stauffer, R., G. J. Mayr, M. Dabernig, and A. Zeileis, 2015: Somewhere overthe rainbow how to make effective use of colors in meteorological visualizations.BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY , 96, 203–215,doi:10.1175/BAMS-D-13-00155.1.
Stein, U. and P. Alpert, 1993: Factor separation in numerical simulations. JOUR-NAL OF THE ATMOSPHERIC SCIENCES , 50, 2107–2115, doi:10.1175/1520-0469(1993)050¡2107:FSINS¿2.0.CO;2.
Stephens, G. L. and T. D. Ellis, 2008: Controls of Global-Mean Precipitation Increasesin Global Warming GCM Experiments. JOURNAL OF CLIMATE , 21, 6141–6155,doi:10.1175/2008JCLI2144.1.
Stephenson, D. B., V. Pavan, M. Collins, M. M. Junge, R. Quadrelli, and P. C. M. Grou,2006: North Atlantic Oscillation response to transient greenhouse gas forcing and theimpact on European winter climate: a CMIP2 multi-model assessment. CLIMATEDYNAMICS , 27, 401–420, doi:10.1007/s00382-006-0140-x.
Stocker, T., D. Qin, G.-K. Plattner, L. Alexander, S. Allen, N. Bindoff, F.-M. Breon,J. Church, U. Cubasch, S. Emori, P. Forster, P. Friedlingstein, N. Gillett, J. Gre-gory, D. Hartmann, E. Jansen, B. Kirtman, R. Knutti, K. Krishna Kumar, P. Lemke,J. Marotzke, V. Masson-Delmotte, G. Meehl, I. Mokhov, S. Piao, V. Ramaswamy,D. Randall, M. Rhein, M. Rojas, C. Sabine, D. Shindell, L. Talley, D. Vaughan, andS.-P. Xie, 2013: Technical Summary , Cambridge University Press, Cambridge, UnitedKingdom and New York, NY, USA, book section TS. 33115.
Sutton, R. T., B. Dong, and J. M. Gregory, 2007: Land/sea warming ratio in responseto climate change: IPCC AR4 model results and comparison with observations. GEO-PHYSICAL RESEARCH LETTERS , 34, doi:10.1029/2006GL028164.
Tanaka, H., N. Ishizaki, and A. Kitoh, 2004: Trend and interannual variability of Walker,monsoon and Hadley circulations defined by velocity potential in the upper troposphere.TELLUS SERIES A-DYNAMIC METEOROLOGY AND OCEANOGRAPHY , 56,250–269, doi:10.1111/j.1600-0870.2004.00049.x.
Tang, Q., G. Leng, and P. Y. Groisman, 2012: European Hot Summers Associated with aReduction of Cloudiness. JOURNAL OF CLIMATE , 25, 3637–3644, doi:10.1175/JCLI-D-12-00040.1.
Tanre, D., J. Geleyn, and J. Slingo, 1984: First results of the introduction of an advancedaerosolradiation interaction in ECMWF low resolution global model, in Aerosols andTheir Climatic Effects, edited by H, Gerber and Deepak, A, 133–177.
Bibliography 103
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of cmip5 and the exper-iment design. BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY ,93, 485–498, doi:10.1175/BAMS-D-11-00094.1.
Tett, S., J. Mitchell, D. Parker, and M. Allen, 1996: Human influence on the atmosphericvertical temperature structure: Detection and observations. SCIENCE , 274, 1170–1173, doi:10.1126/science.274.5290.1170.
Tiedtke, M., 1989: A Comprehensive Mass Flux Scheme for Cumulus Parameteri-zation in Large-Scale Models. MONTHLY WEATHER REVIEW , 117, 1779–1800,doi:10.1175/1520-0493(1989)117¡1779:ACMFSF¿2.0.CO;2.
Torma, C., F. Giorgi, and E. Coppola, 2015: Added value of regional climatemodeling over areas characterized by complex terrainPrecipitation over the Alps.JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES , 120, 3957–3972,doi:10.1002/2014JD022781.
Trenberth, K., A. Dai, R. Rasmussen, and D. Parsons, 2003: The changing character ofprecipitation. BULLETIN OF THE AMERICAN METEOROLOGICAL SOCIETY ,84, 1205+, doi:10.1175/BAMS-84-9-1205.
Ulbrich, U., J. G. Pinto, H. Kupfer, G. C. Leckebusch, T. Spangehl, and M. Reyers, 2008:Changing northern hemisphere storm tracks in an ensemble of IPCC climate changesimulations. JOURNAL OF CLIMATE , 21, 1669–1679, doi:10.1175/2007JCLI1992.1.
Vallis, G. K., P. Zurita-Gotor, C. Cairns, and J. Kidston, 2015: Response of the large-scale structure of the atmosphere to global warming. Quarterly Journal of the RoyalMeteorological Society , 141, 1479–1501, doi:10.1002/qj.2456.
van der Linden, P. and J. Mitchell, 2009: Ensembles: Climate change and its impacts:Summary of research and results from the ensembles project. Technical report, MetOffice Hadley Centre, FitzRoy Road, Exeter EX1 3PB, UK.
van Ulden, A., G. Lenderink, B. van den Hurk, and E. van Meijgaard, 2007: Circula-tion statistics and climate change in Central Europe: PRUDENCE simulations andobservations. CLIMATIC CHANGE , 81, 179–192, doi:10.1007/s10584-006-9212-5.
van Vuuren, D. P., J. Edmonds, M. Kainuma, K. Riahi, A. Thomson, K. Hibbard, G. C.Hurtt, T. Kram, V. Krey, J.-F. Lamarque, T. Masui, M. Meinshausen, N. Nakicenovic,S. J. Smith, and S. K. Rose, 2011: The representative concentration pathways: anoverview. CLIMATIC CHANGE , 109, 5–31, doi:10.1007/s10584-011-0148-z.
Vautard, R., 1990: Multiple weather regimes over the north-atlantic - analysisof precursors and successors. MONTHLY WEATHER REVIEW , 118, 2056–2081,doi:10.1175/1520-0493(1990)118¡2056:MWROTN¿2.0.CO;2.
Vautard, R., A. Gobiet, D. Jacob, M. Belda, A. Colette, M. Deque, J. Fernan-dez, M. Garcia-Diez, K. Goergen, I. Guettler, T. Halenka, T. Karacostas, E. Ka-tragkou, K. Keuler, S. Kotlarski, S. Mayer, E. van Meijgaard, G. Nikulin, M. Patar-cic, J. Scinocca, S. Sobolowski, M. Suklitsch, C. Teichmann, K. Warrach-Sagi,
104 Bibliography
V. Wulfmeyer, and P. Yiou, 2013: The simulation of European heat waves from anensemble of regional climate models within the EURO-CORDEX project. CLIMATEDYNAMICS , 41, 2555–2575, doi:10.1007/s00382-013-1714-z.
Vautard, R. and P. Yiou, 2009: Control of recent European surface cli-mate change by atmospheric flow. GEOPHYSICAL RESEARCH LETTERS , 36,doi:10.1029/2009GL040480.
Vecchi, G. A. and B. J. Soden, 2007: Global warming and the weakening of the tropicalcirculation. JOURNAL OF CLIMATE , 20, 4316–4340, doi:10.1175/JCLI4258.1.
Vidale, P. L., D. Luethi, R. Wegmann, and C. Schar, 2007: European summer climatevariability in a heterogeneous multi-model ensemble. CLIMATIC CHANGE , 81, 209–232, doi:10.1007/s10584-006-9218-z.
Wallace, J. and P. Hobbs, 2006: Atmospheric Science An Introductory Survey . Elsevier,London, United Kingdom, 492 pp.
Woollings, T., 2008: Vertical structure of anthropogenic zonal-mean atmo-spheric circulation change. GEOPHYSICAL RESEARCH LETTERS , 35,doi:10.1029/2008GL034883.
— 2010: Dynamical influences on European climate: an uncertain future. PHILOSOPH-ICAL TRANSACTIONS OF THE ROYAL SOCIETY A-MATHEMATICAL PHYSI-CAL AND ENGINEERING SCIENCES , 368, 3733–3756, doi:10.1098/rsta.2010.0040.
Woollings, T., J. M. Gregory, J. G. Pinto, M. Reyers, and D. J. Brayshaw, 2012: Responseof the North Atlantic storm track to climate change shaped by ocean-atmosphere cou-pling. NATURE GEOSCIENCE , 5, 313–317, doi:10.1038/NGEO1438.
Zaengl, G., D. Reinert, P. Ripodas, and M. Baldauf, 2015: The ICON (ICOsahedralNon-hydrostatic) modelling framework of DWD and MPI-M: Description of the non-hydrostatic dynamical core. QUARTERLY JOURNAL OF THE ROYAL METEORO-LOGICAL SOCIETY , 141, 563–579, doi:10.1002/qj.2378.
Zappa, G., L. C. Shaffrey, and K. I. Hodges, 2013: The Ability of CMIP5 Models toSimulate North Atlantic Extratropical Cyclones. JOURNAL OF CLIMATE , 26, 5379–5396, doi:10.1175/JCLI-D-12-00501.1.
Zhang, X., L. Alexander, G. C. Hegerl, P. Jones, A. K. Tank, T. C. Peterson, B. Trewin,and F. W. Zwiers, 2011: Indices for monitoring changes in extremes based on dailytemperature and precipitation data. Wiley Interdisciplinary Reviews: Climate Change,2, 851–870, doi:10.1002/wcc.147.
Acknowledgements
This PhD Thesis would not exist without the help and encouragement of many differentpeople. I would like to express my gratitude to them.Christoph Schar, my doctoral supervisor, who gave me the opportunity to join his groupand take on the challenge of attaining a PhD. He supported me during this journey withhis expert knowledge and guidance. His expertise in the dynamics of the atmosphereoften helped improve my work. He also gave me the opportunity to present my work atdifferent interesting conferences and workshops.My team of three supervisors each of whom has contributed greatly, supporting me withtheir knowledge, experience and enthusiasm.
Sven Kotlarski, who was my primary supervisor and the crucial factor for the successof this PhD. He strongly supported me in times of doubts and encouraged me to go on.I appreciate the time he spent discussing with me, giving me valuable new input as wellas the (hard) time he spent proof reading my work. I enjoyed the time we spend togetherat conferences in Hamburg or Sweden.
Daniel Luthi’s door was always open for me. Not only was he able to solve all thetechnical difficulties (with the COSMO-Model or Linux) which surfaced during this lastthree and a half years, he also spent a lot of time with me going over figures trying tounderstand what they actually meant.
Erich Fischer connected me with Christoph and made all of this start. He continuedto provide excellent scientific advice during my PhD, often being the one to remind meto keep the focus and not wander too far from the topic and questions from which westarted.I also want to thank Roy Rasmussen for reviewing my PhD and making the long tripfrom the USA to Switzerland.The people at IAC and especially the L-Floor Gang with whom I spent many happyevenings, for example visiting Greek restaurants, enjoying a beer together after work atBQM or doing sports. Many of them became friends.Finally, I want to thank my family and friends who supported me and made sure that Ihad a life beside the PhD. One person I want to thank in particular is Fatma. She was theone cheering me up when I had a bad time at work and always believed in me, especiallythe last stressful month; I would not have survived without her.
105
Curriculum Vitae
Nico Robert KronerRegensbergstr. 81, 8050 ZurichBorn on January 6th, 1985 in Frankfurt am Main (Germany)Citizen of Germany
Education
2012–2016 PhD student in the group of Prof. C. ScharInstitute for Atmosperic and Climate Science, ETH Zurich,Zurich, CH
2009–2012 Master in Environmental Sciences ETH, Major in Atmosphericand Climate SciencesETH Zurich, Zurich, CH
09.2011–12.2011 Exchange program at the University of EdinburghEdinburgh, UK
2006–2009 Bachelor in Environmental Sciences ETHETH Zurich, Zurich, CH
2004 University-entrance Diploma, AbiturBettinaschule, Frankfurt am Main, DE
Professional Experience
10.2012–06.2016 Research Assistant and PhD StudentInstitute for Atmospheric and Climate Science, ETH Zurich,Zurich, CH
01.2012–03.2012 Ethics AnalystINVERA Ethics Research and Advisery AG, Zurich, CH
05.2011–09.2011 Intern (Junior Ethics Analyst)INVERA Ethics Research and Advisery AG, Zurich, CH
09.2006–05.2011 Assistant for the lecture Chemistry I. and II.Institute of Biogeochemistry and Pollutant Dynamics, ETHZurich, Zurich, CH
05.2005–08.2005 Temporary EmployeeMedico International (NGO), Frankfurt am Main, DE
08.2004–05.2005 Civil ServiceMedico International (NGO), Frankfurt am Main, DE
108 Bibliography
Awards
Best Poster Award of the 10th CLM-Community Assembly 2015
Publications
[4] Kroner, N., D. Luthi, E. Fischer, S. Kotlarski and C. Schar (2016): Quantifyingthe drivers of European precipitation changes: Large-scale thermodynamics, lapse-rateand circulation changes, in preparation
[3] Schar, C. and N. Kroner (2016): Sequential Factor Separation for the Analysisof Weather and Climate Model Results, Submitted to Journal of Atmospheric Science
[2] Kroner, N., S. Kotlarski, E. Fischer, D. Luthi, E. Zubler and C. Schar (2016):Separating climate change signals into thermodynamic, lapse-rate and circulation ef-fects: Theory and application to the European summer climate, Climate Dynamics,doi:10.1007/s00382-016-3276-3
[1] Jacob, D., and J. Petersen, B. Eggert, A. Alias, O.B. Christensen, L.M. Bouwer, A.Braun, A. Colette, M. Dqu, G. Georgievski, E. Georgopoulou, A. Gobiet, L. Menut, G.Nikulin, A. Haensler, N. Hempelmann, C. Jones, K. Keuler, S. Kovats, N. Kroner,S. Kotlarski, A. Kriegsmann, E. Martin, E. Meijgaard, C. Moseley, S. Pfeifer, S.Preuschmann, C. Radermacher, K. Radtke, D. Rechid, M. Rounsevell, P. Samuelsson, S.Somot, J.-F. Soussana, C. Teichmann, R. Valentini, R. Vautard, B. Weber, and P. Yiou(2014): EURO-CORDEX: new high-resolution climate change projections for Europeanimpact research, Regional Environmental Change, 14, 563-578, doi:10.1007/s10113-013-0499-2
Conferences, Workshops and Summerschools2016: European Geoscience Union General Assembly 2016 (EGU2016), Vienna, Austria2015: 10th CLM-Community Assembly 2015, Belvaux, Luxenbourg2015: European Geoscience Union General Assembly 2015 (EGU2015), Vienna, Austria2014: 3th International Lund Regional-Scale Climate Modelling Workshop: 21st CenturyChallenges in Regional Climate Modelling, Lund, Sweden2014: 9th CLM-Community Assembly 2014 Frankfurt am Main, Germany2014: 4th EURO-CORDEX meeting, Hamburg, Germany2013: 8th CLM-Community Assembly 2013 ETH Zurich, Zurich, Switzerland2013: NCCR Summer School, From Climate Reconstructions to Climate Predictions,Grindelwald, Switzerland2013: COSMO-CLM Trainings course, Deutscher Wetterdienst, Offenbach, Germany2013: 2nd EURO-CORDEX meeting, Hamburg, Germany
