Nat. Hazards Earth Syst. Sci.

, 21, 171–186, 2021

https://doi.org/10.5194/nhess-21-171-2021
© Author(s) 2021. This work is distributed under
the Creative Commons Attribution 4.0 License.

A methodology for attributing the role of climate change

in extreme events: a global spectrally nudged storyline
Linda van Garderen1 , Frauke Feser1 , and Theodore G. Shepherd2
1 Institute
for Coastal Research – Analysis and Modelling, Helmholtz-Zentrum Geesthacht,
Max-Planck-Straße 1, 21502 Geesthacht, Germany
2 Department of Meteorology, University of Reading, Reading RG6 6BB, United Kingdom

Correspondence: Linda van Garderen ([email protected])

Received: 5 June 2020 – Discussion started: 26 June 2020

Revised: 25 September 2020 – Accepted: 26 November 2020 – Published: 18 January 2021

Abstract. Extreme weather events are generally associated 1 Introduction

with unusual dynamical conditions, yet the signal-to-noise
ratio of the dynamical aspects of climate change that are rel- There is increasing interest in understanding and quantifying
evant to extremes appears to be small, and the nature of the the impact of climate change on individual extreme weather
change can be highly uncertain. On the other hand, the ther- and climate events. This is to be distinguished from detect-
modynamic aspects of climate change are already largely ing the effect of climate change on the statistics of extreme
apparent from observations and are far more certain since events (IPCC, 2012). In the most commonly used approach,
they are anchored in agreed-upon physical understanding. changes in the probability distribution of an event class,
The storyline method of extreme-event attribution, which whose definition is motivated by a historical event, are cal-
has been gaining traction in recent years, quantitatively es- culated by simulating large ensembles with an atmosphere-
timates the magnitude of thermodynamic aspects of climate only climate model (Watanabe et al., 2013). The changes are
change, given the dynamical conditions. There are different computed between the “factual” ensemble, corresponding to
ways of imposing the dynamical conditions. Here we present observed forcings (e.g. sea-surface temperatures (SSTs) and
and evaluate a method where the dynamical conditions are greenhouse-gas (GHG) concentrations), and a “counterfac-
enforced through global spectral nudging towards reanalysis tual” ensemble, corresponding to an imagined world without
data of the large-scale vorticity and divergence in the free at- climate change. The latter is usually constructed by remov-
mosphere, leaving the lower atmosphere free to respond. We ing an estimate of the forced changes in SSTs and imposing
simulate the historical extreme weather event twice: first in pre-industrial GHG concentrations. As discussed by Shep-
the world as we know it, with the events occurring on a back- herd (2016), this probabilistic approach has two prominent
ground of a changing climate, and second in a “counterfac- limitations. The first is that every extreme event is unique,
tual” world, where the background is held fixed over the past but the construction of a general event class blurs the connec-
century. We describe the methodology in detail and present tion to the actual event and makes it difficult to link the event
results for the European 2003 heatwave and the Russian 2010 attribution to climate impacts. This is important because ex-
heatwave as a proof of concept. These show that the con- treme impacts are not always associated with extreme me-
ditional attribution can be performed with a high signal-to- teorology (van der Wiel et al., 2020). The second limitation
noise ratio on daily timescales and at local spatial scales. is that extreme events are generally associated with extreme
Our methodology is thus potentially highly useful for real- dynamical conditions, and there is little understanding, let
istic stress testing of resilience strategies for climate impacts alone agreement, on how those dynamical conditions might
when coupled to an impact model. respond to climate change (Hoskins and Woollings, 2015;
Shepherd, 2014). This represents an uncertainty in the prob-
abilistic estimates that is difficult to quantify.

Published by Copernicus Publications on behalf of the European Geosciences Union.

172 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

On the other hand, thermodynamic aspects of climate whatsoever, seeks to avoid “Type 1” errors or false alarms
change such as warming and increasing specific humidity (Lloyd and Oreskes, 2018; Trenberth et al., 2015). A collo-
are robust in sign, anchored in agreed-upon physical un- quial way of putting this is that rather than asking what ex-
derstanding, and clearly emerging in observations (IPCC, treme events can tell us about climate change, we ask what
2018). Moreover in many cases the signal-to-noise ratio of known aspects of climate change can tell us about partic-
the forced dynamical changes appears likely to be small ular extreme events. Although its results are not expressed
(Deser et al., 2016; Schneider et al., 2012). Thus, although probabilistically, the storyline approach enables a quantita-
dynamical and thermodynamic processes are interwoven in tive estimate of climate change with a clear causal interpreta-
the real climate system, it can be useful to regard the uncer- tion (Pearl and Mackenzie, 2018). Notwithstanding the need
tainties in their forced response to climate change as being for asking both kinds of questions as they provide different
separable, at least to a first approximation. This has been a kinds of information (Lloyd and Shepherd, 2020), the story-
growing theme in climate change attribution over the past line approach is a new development, and there are as yet not
few decades. The distinction between thermodynamic and so many studies employing this approach.
dynamical changes is not precise, and various ways of im- In previous applications of the storyline approach, indi-
plementing the separation diagnostically have been used in vidual extreme weather events have been dynamically con-
different contexts. For extratropical regional climate, it has strained through boundary conditions applied to a regional
been common to regard the component of change congruent model (Meredith et al., 2015) or by controlling the ini-
with large-scale internal variability (e.g. as defined by em- tial conditions in a weather forecast model (Patricola and
pirical orthogonal functions or by self-organizing maps) as Wehner, 2018). More recently, nudging the free atmosphere
“dynamical” (Deser et al., 2016; Horton et al., 2015) and the to reanalysis data (leaving the boundary layer free to re-
residual as “thermodynamic”. For tropical climate or for ex- spond) has been applied in a global medium-resolution at-
tratropical storms, dynamical changes are instead commonly mospheric model to constrain the dynamical conditions lead-
identified with changes in vertical velocity (Bony et al., 2013; ing to heatwaves, first to determine the effect of soil mois-
Pfahl et al., 2017). In the absence of evidence to the contrary, ture changes on selected recent heatwaves (Wehrli et al.,
a reasonable hypothesis is that the forced dynamical changes 2019) and subsequently to determine the effect of past and
are undetectable; this hypothesis is implemented explicitly projected future warming on the 2018 Northern Hemisphere
in the “pseudo global warming” methodology used for re- heatwave (Wehrli et al., 2020). The concept of nudging the
gional climate studies (Schär et al., 1996) and in the “dy- atmospheric circulation in order to impose the dynamical
namical adjustment” methodology used to study observed conditions has a long history. In particular, spectral nudg-
climate trends (Wallace et al., 2012). ing (von Storch et al., 2000; Waldron et al., 1996) allows for
Trenberth et al. (2015) suggested that the same think- scale-selective nudging so that only the large spatial scales
ing could be usefully applied to the attribution of individ- of the model are constrained, while the smaller scales, in-
ual extreme events. Specifically, the extreme dynamical cir- cluding those relevant to extreme events, are free to be sim-
cumstances leading to the event could be regarded as given, ulated by the high-resolution model. The climate model can
i.e. arising by chance, and the question posed of how the thus potentially add value and regional detail to the coarser-
event was modified by the known thermodynamic aspects resolution forcing dataset. Spectral nudging has been used in
of climate change. This conditional framing of the attribu- regional climate modelling (Feser and Barcikowska, 2012;
tion question was subsequently dubbed the “storyline” ap- Scinocca et al., 2015) and in boundary-layer sensitivity stud-
proach (Shepherd, 2016) and has a precedent in the appli- ies (van Niekerk et al., 2016). Note that in all these modelling
cation of dynamical adjustment to extreme seasonal climate approaches, the dynamical constraint is imposed “remotely”
anomalies (Cattiaux et al., 2010). As emphasized by Shep- from the phenomenon of interest (in space, time, and/or spa-
herd (2016) and NAS (2016), there is actually a continuum tial scale) in contrast to the diagnostic approaches mentioned
between the storyline and probabilistic approaches: story- earlier and thus preserves the physical interplay between dy-
lines are highly conditioned probabilities, and probabilistic namics and thermodynamics within the extreme event itself.
approaches generally involve some form of dynamical con- The purpose of this paper is to provide a methodolog-
ditioning too, through the imposed SST patterns. However, ical underpinning for the application of large-scale spec-
the extent of conditioning imposed by constraining the at- tral nudging of divergence and vorticity in a global high-
mospheric state is so severe that in practice the storyline ap- resolution atmospheric model for the purpose of attributing
proach can be regarded as deterministic, just as weather fore- the role of thermodynamic aspects of climate change (or
casts, whilst probabilistic in principle, are interpreted deter- other conditional perturbations) in extreme events of vari-
ministically when the ensemble spread is sufficiently narrow. ous types and timescales. A key question is to determine
By focusing on the known effects of climate change, the what level of refinement of the attribution, in both space and
storyline approach seeks to avoid “Type 2” errors or missed time, is possible. The outline of the paper is as follows. In
warnings, in contrast to the probabilistic approach, which, Sect. 2, we elaborate on the technicalities of spectral nudg-
by needing to reject the null hypothesis of no climate change ing within the ECHAM6 model and its parameter sensitiv-

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 173

ities as well as the construction of the counterfactual simu- only the large-scale fields are nudged. We chose NCEP1 due
lations. In Sect. 3, we exemplify the method by applying it to its starting date in 1948, which is earlier than any of the
to two well-studied heatwaves: the European 2003 heatwave other reanalysis data, enabling application of our method
and the Russian 2010 heatwave. As well as identifying some over a longer period of time. It is conceivable that for certain
important differences between the two events, we examine kinds of extreme events involving a tight coupling between
the signal-to-noise ratio of our attribution. A concluding dis- resolved and parameterized processes, ensuring consistency
cussion follows in Sect. 4. between the reanalysis and the model would be beneficial. In
a previous application nudging was applied for pressure, tem-
perature, vorticity, and divergence (Jeuken et al., 1996) with
2 Method a constant height profile throughout the entire atmosphere.
However, we want to reproduce only the large-scale atmo-
2.1 Spectral nudging
spheric circulation and in particular leave the thermodynamic
The spectral-nudging technique is well established within the fields (temperature and moisture) free to respond; hence we
context of regional climate modelling (Miguez-Macho et al., only nudge vorticity and divergence in the free atmosphere.
2004; von Storch et al., 2000, 2018; Waldron et al., 1996). The aim is to constrain the model as little as possible so that it
In this approach, so-called “nudging terms” are added to can freely develop small-scale meteorological processes and
the large-scale part of the climate model trajectory, which extreme events while still achieving an effective control of
draws the model towards reanalysis data. Global spectral the large-scale weather situation.
nudging (Kim and Hong, 2012; Schubert-Frisius et al., 2017; The nudging of variable X over time is applied in the spec-
Yoshimura and Kanamitsu, 2008) works in a similar way. It tral domain as follows (adapted from Jeuken et al., 1996):
constrains large-scale weather patterns of the climate model, ∂X

FX + G (XNCEP − X) for n ≤ 20, p < 750 hPa
such as high- and low-pressure systems or fronts, to stay = FX otherwise , (1)
∂t
close to reanalysis data in order to derive a global high-
resolution weather reconstruction. The general idea is that where X is the variable to be nudged (either vorticity or di-
the realistic large-scale state of the reanalysis data is fol- vergence), FX is the model tendency for variable X, and
lowed by the global climate model (GCM), while at smaller XNCEP is the state of that variable in NCEP1. The thresh-
scales the model provides additional detail to improve high- olds p and n need to be met for nudging to happen, namely
resolution weather patterns. Another merit of the approach is pressure p must be below 750 hPa, and the spherical har-
the potential to reduce inhomogeneities in the dataset by us- monic index n must not exceed 20. G is the relaxation coeffi-
ing only a very limited number of variables from the reanal- cient in units of 10−5 s−1 determining the nudging strength.
ysis data, although this is less of an issue for our application Nudging is performed at every time step.
because we compare factual and counterfactual simulations We applied most settings according to Schubert-Frisius
for the same large-scale conditions, so any inhomogeneity in et al. (2017), including the usage of spectral nudging in
the reanalysis would apply equally to both. For the same rea- both meridional and zonal directions. We use a plateau
son, our approach can be expected to be robust to any differ- nudging-strength height profile (see Fig. 1a), which starts at
ences between reanalyses. In order to define a noise level for 750 hPa, then quickly increases up to its maximum nudging
our analysis, we construct small ensembles of three factual strength, and stays there for higher tropospheric and lower
and three counterfactual simulations. Although such small and medium stratospheric levels until it again quickly tapers
ensembles are clearly inadequate for quantifying conditional back to 0 at a height corresponding to 5 hPa. The reason for
probabilities, they have been successfully used in the past the latter choice is that above 5 hPa there is no NCEP1 re-
(e.g. Shepherd, 2008) to identify robust differences between analysis data available.
the two ensembles from a deterministic perspective, which is The strength of nudging is determined by the relaxation
our interest here. coefficient (G; in 10−5 s−1 ); see Eq. (1). The relaxation co-
efficient is often described using the e-folding time (G−1 ;
2.2 ECHAM6 application in 105 s) which represents the simulated time necessary for
nudging to dampen out a model-introduced disturbance. For
For this study, we use the high-resolution T255L95 GCM example, if the e-folding time is 10 h then the nudged model
ECHAM6 (Stevens et al., 2013) with the JSBACH land com- will dampen out that disturbance (with an assumed amplitude
ponent sub-model (Reick et al., 2013); however the method of 1) to a value of 1/e and thus greatly reduce it within 10 h.
is applicable to any atmospheric GCM. SSTs and sea ice A larger relaxation coefficient implies a stronger nudging and
concentrations (SICs) are prescribed from NCEP1 reanalysis translates into a shorter e-folding time or dampening time
data (Kalnay et al., 1996). ECHAM6 is globally spectrally (von Storch et al., 2000). We have tested several e-folding
nudged towards the NCEP1 reanalysis data to achieve real- times to see if the settings could be further relaxed and still
istic weather patterns and extreme events of the past. How- reproduce the large-scale weather conditions. In Fig. 1b the
ever, any other reanalysis should provide similar results since impact of the tested e-folding time settings on the temporal

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

174 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

Figure 1. (a) Nudging strength G [10−5 s−1 ] as a function of model level for different choices of minimum e-folding time as indicated.
(b) Daily mean temperatures at 2 m height [◦ C] of ECHAM6 in November 2013 averaged over the European domain (35–60◦ N, 10◦ W–
35◦ E) using the different e-folding times shown in (a) in comparison to ERA-Interim. (c) Daily mean temperatures as in (b) but with a
50 min nudging timescale at different truncations again in comparison to ERA-Interim.

evolution of the 2 m temperature averaged over Europe (35–

60◦ N, 10◦ W–30◦ E) in comparison to ERA-Interim is shown
through November 2013. There is little difference visible be-
tween the 50 min and 5 h e-folding times. The 10 h results
start to show small deviations, whilst the 20 h results deviate
even more noticeably. On the basis of this sensitivity study,
we conclude that the e-folding time can safely be relaxed
from 50 min to 5 h without losing the accuracy of the results.
We similarly aim to limit the range of spatial scales be-
ing nudged as much as possible. In Fig. 1c we show the 2 m
temperature results for the different nudging wavelengths Figure 2. Geopotential height (z500 ) June–July–August (JJA)
in comparison to ERA-Interim. The original T30 settings anomalies [m] for the Northern Hemisphere showing the averaged
used by Schubert-Frisius et al. (2017), which translate to a spectrally nudged dynamic situation over (a) factual members and
minimum wavelength of approximately 1300 km (360◦ /30× (b) counterfactual members of the summer 2010 blocking. Anoma-
lies were calculated relative to the ECHAM_SN 1980–2014 JJA
111 km), show comparable results to the T25 and T20 resolu-
climatology.
tions. The nudging was therefore relaxed to the T20 resolu-
tion, which translates to a minimum wavelength of approx-
imately 2000 km (360◦ /20 × 111 km). This should be suf-
ficient to resolve the large-scale circulation while allowing 2.3 Simulating the counterfactual
smaller-scale processes related to local weather events to de-
velop freely. In Fig. 2 the geopotential height anomalies for
summer 2010 in the factual and counterfactual simulations In this study, as in probabilistic event attribution, counterfac-
show a strong resemblance. Even though the background tual and factual climate simulations are used to assess the
conditions of the two simulations are different (which is fur- effect of climate change on extreme events. Factual is de-
ther explained in Sect. 2.3), the blocking pattern formed over fined as the world as we know it or a historical simulation.
Russia in 2010 is clearly present in both simulations, demon- Counterfactual is defined as an imagined modern world with-
strating the capability of our nudging method to reproduce out climate change. In our simulations, land use and volcanic
the complex dynamical situation. activity as well as aerosol forcing and sea ice concentration
We used ECHAM_SN throughout this paper to calculate are unchanged between factual and counterfactual. The dif-
climatological data for comparison to our own findings. The ferences between the two worlds are created by altering two
ECHAM_SN dataset is a spectrally nudged global historical important aspects of the simulation: (a) sea-surface temper-
simulation from 1948–2015 (Schubert-Frisius et al., 2017). It ature (SST) and (b) greenhouse gases (GHG). Both worlds
nudged vorticity and divergence towards NCEP1 in a vertical are spectrally nudged in the same way. A potential way to
plateau-shaped profile, equal to the profile we use, at spatial check the results of the counterfactual simulation, especially
scales corresponding to T30 or larger, with an e-folding time for simulations over a longer time span, is to study the con-
of 50 min. sistency between the inferred signals of climate change for
smaller climate forcings (e.g. since mid-century) and the at-
tributed changes in the observational record. Our simulations
are 5 years each and therefore cannot be tested in this way.

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 175

However for longer simulations such a test would be benefi- Table 1. Greenhouse-gas concentrations for the ECHAM6 counter-
cial. factual simulations.
SST patterns such as the Atlantic Multidecadal Oscilla-
tion or El Niño greatly influence weather extremes. There- Greenhouse gas Concentration
fore, as with probabilistic event attribution, we impose the Carbon dioxide (CO2 ) 285 ppmv
same SST variability for both the factual and counterfactual Methane (CH4 ) 790 ppbv
simulation, based on the observed SST pattern. (However, Nitrous oxide (N2 O) 275 ppbv
this is expected to be less critical in our case since we are Chlorofluorocarbons (CFCs) 0
imposing the large-scale atmospheric circulation.) We create
the counterfactual SST conditions by subtracting a climato-
logical warming pattern from the observed pattern, which is
a standard procedure in probabilistic event attribution studies
(Otto, 2017; Vautard et al., 2016; Stott et al., 2016). Although SIC. This shows that the sea ice edge is well away from the
it is common to consider different climatological warming European and western Russian domains. Moreover, even un-
patterns as a means of exploring uncertainty, this is not so rel- der counterfactual conditions the SST remains almost com-
evant in our case since the large-scale circulation is imposed. pletely physically self-consistent with the SIC according to
The climatological warming pattern is computed using the the constraints of Hurrell et al. (2008); in particular, there are
ECHAM6 CMIP6 (MPI-ESM1.2-HR) control and historical only a very few isolated regions where the SST falls below
simulations at an atmospheric resolution of T127 (Müller et −2 ◦ C. Nevertheless, we tested the impact of altering SIC
al., 2018). The procedure is shown in Eq. (2): in a counterfactual simulation of the Russian heatwave based
  on the counterfactual SSTs, using the linear relation found by
SSTt,c = SSTNCEP1
t − SST CMIP6
t,h − SST CMIP6
t,pi , (2) Hurrell et al. (2008). Specifically, SIC was set to 100 % for
SSTs below −1.7 ◦ C and to 0 % for SSTs above 3 ◦ C, with
where SSTt,c is the counterfactual SST at time t, SSTNCEP1
t is a linear interpolation in between. The results show no dif-
the NCEP1 SST at time t, SSTCMIP6 t,h is the CMIP6 historical ferences compared to the unaltered SIC counterfactual mem-
SST at time t, and SSTCMIP6
t,pi is the CMIP6 pre-industrial SST bers (see Fig. 5b). However, to apply our method to other
at time t (for the latter, the only relevant time dependence seasons or regions in close proximity to areas of sea ice loss,
would be seasonal). In our present implementation, which the counterfactual simulations would benefit from including
targets boreal summer only and concerns only a fairly short SIC changes in the same way as was done with SST.
time period, the seasonal time dependence is suppressed, and In the factual simulation the GHGs change according to
the historical CMIP SSTs are taken to be the 2000–2009 av- observed values (Meinshausen et al., 2011). In the coun-
erage. For a simulation covering a full year the warming terfactual simulation, GHGs remain at their 1890 values as
pattern should be made seasonal, and for one covering sev- listed in Table 1. This means that, strictly speaking, our at-
eral decades it would furthermore need to be weighted over tribution is to the combined effects of anthropogenic climate
time. In Fig. 3 the CMIP6 SST warming pattern shows a change (including aerosol forcing) recorded in the SSTs as
good resemblance to the observed HadSST3 warming pat- well as the direct radiative effects of GHG forcing.
tern. The HadSST3 pattern is obtained by subtracting the The default initial atmospheric state of the ECHAM6
1880–1890 average from the 1980–1990 average SST val- model is a random state during the simulated mid-1990s.
ues. The general warming and cooling patches in the Pacific Changing that initial state to a counterfactual initial state re-
Ocean and Atlantic Ocean south of Greenland agree well. quires a spin-up time to allow the atmosphere and land sur-
Also, the warming north of Scandinavia is clearly visible in face enough time to reach a new equilibrium state with their
both warming patterns. Despite the observational data-void new boundary conditions. To accomplish this we run a non-
region east of Greenland and north of Iceland, there is a good nudged counterfactual spin-up ensemble for 3 model years
resemblance of our modelled warming pattern with observa- with three members. We chose a 3-year spin-up after con-
tions. Note that pre-industrial SST observations were depen- firming the soil moisture was adapted to the new counterfac-
dent upon ship records, which in the polar region were very tual situation (not shown). Each member was initiated at a
few (Rayner et al., 2006), causing this part of the observa- different starting date (January–March 1995). The results of
tional dataset to be incomplete. these spin-ups are three random atmospheric counterfactual
For technical reasons, we did not alter the SIC in the states, which are used as initial conditions for the counter-
counterfactual simulations. Given that the atmospheric cir- factual experiments. Although in principle both the factual
culation is nudged, changes in SIC are not expected to be and counterfactual conditions define conditional probabili-
relevant for summertime heatwaves as Arctic amplification ties, our three-member ensembles are certainly not sufficient
from sea ice loss is a wintertime phenomenon (Screen and to estimate those probabilities. As noted earlier, our goal here
Simmonds, 2010). In Fig. 4 the counterfactual SSTs for is simply to determine the robustness of the deterministic dif-
July 2003 and July 2010 are shown together with the factual ferences between the factual and counterfactual ensembles.

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

176 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

Figure 3. Sea-surface temperature (SST) warming pattern [◦ C] calculated (a) from ECHAM6 CMIP6 modelled data and (b) from HadSST3
observed data.

Figure 4. Counterfactual SST [◦ C] in shaded colours and factual SIC [%] in greyscale for (a) July 2003 and (b) July 2010. The SST 5 ◦ C
(dashed green), 0 ◦ C (orange), and −2 ◦ C (red) contours are marked for reference.

The ECHAM_SN simulation and the altered SIC simulation observations we know that the earth has experienced a global
provide out-of-sample tests of robustness for the factual and warming of approximately 0.7–0.8 ◦ C between preindustrial
counterfactual ensembles, respectively. Figure 5 shows that times and 2010 (IPCC, 2018). Our modelled global warming,
in both cases, these simulations fall largely within the range found through the difference between the factual and coun-
of the three-member ensembles. terfactual simulations, thus represents this difference well,
For the European 2003 heatwave the three counterfac- albeit with a slight underestimation.
tual members run from 1 March and are initialized with
the spin-up counterfactual atmospheric state members (year
3, March). The three factual members are started 1 month 3 Results
apart from each other (in January–March 2003), and ini-
tialized with the corresponding atmospheric state from the To illustrate our method, we provide two examples, namely
ECHAM_SN dataset. For the Russian 2010 heatwave the the European heatwave of 2003 and the Russian heatwave
three counterfactual members run instead from 1 January be- of 2010. These events are considered the two strongest Eu-
cause of the known importance of soil preconditioning for ropean heatwaves on record (Russo et al., 2014, 2015). In
this event (Wehrli et al., 2019). The three factual members Sect. 3.3 we look deeper into the signal-to-noise ratio of each
again run with 1-month differences in their starting dates, of the examples and how they compare to each other.
but here from November 2009, December 2009, and Jan-
uary 2010, again initialized with the corresponding state 3.1 European heatwave 2003
from the ECHAM_SN dataset. For analysis regions we se-
lect 35–50◦ N, 10◦ W–25◦ E as the domain for the European The European summer of 2003 was exceptionally hot and
heatwave 2003 and 50–60◦ N, 35–55◦ E for the Russian heat- exceptionally dry (Black et al., 2004; Schär et al., 2004;
wave 2010, in line with previous literature (Dole et al., 2011; Stott et al., 2004). Two heatwaves occurred, a milder one in
García-Herrera et al., 2010; Otto et al., 2012; Rasmijn et al., June and an extreme heatwave in August, with peak temper-
2018; Wehrli et al., 2019). atures in France and Switzerland (Black et al., 2004; Schär
For the summer of 2003, the global temperature difference et al., 2004; Trigo et al., 2005) but also affecting Portugal,
between factual and counterfactual simulations is 0.64 ◦ C, northern Italy, western Germany, and the UK (Feudale and
while for the summer of 2010 the difference is 0.66 ◦ C. From Shukla, 2011a; Muthers et al., 2017). Temperatures exceeded
the 1961–1990 average by 2.3–12.5 ◦ C, depending on loca-

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 177

Figure 5. Daily mean temperature at 2 m height [◦ C] averaged over (a) Europe (35–50◦ N, 10◦ W–25◦ E) for summer 2003 and over (b) Rus-
sia (50–60◦ N, 35–55◦ E) for summer 2010 for the factual (blue), counterfactual (red), and ECHAM_SN (green) simulations and ERA-
Interim (dashed black) reanalysis data. The climatology (green shaded area) is the 5th–95th ranked percentile range between 1985–2015
calculated with ECHAM_SN (Schubert-Frisius et al., 2017). The dashed red line in (b) shows the simulation with SIC changed in one of the
counterfactual simulations (see text for details).

tion, without much cooling during the night (García-Herrera out climate change (Hannart et al., 2016; Schär et al., 2004;
et al., 2010; Schär et al., 2004; Stott et al., 2004; Muthers et Stott et al., 2004). The probabilistic event attribution stud-
al., 2017). The 2003 summer was at that point in time not just ies show an increased likelihood of the extreme tempera-
the hottest on record (Bastos et al., 2014; Fink et al., 2004), it tures from increased GHGs (Hannart et al., 2016; Schär et
was the hottest summer in the past 500 years (Luterbacher et al., 2004; Stott et al., 2004). Other studies focused on the ex-
al., 2004). The consequences were devastating. Estimates ac- ceptionally high SSTs in the Mediterranean Sea and North
count for 22 000–40 000 heat-related deaths, USD 12–14 bil- Sea as a cause of reduced baroclinicity, providing an envi-
lion in economic losses, 20 %–30 % decrease in net primary ronment conducive to blocking (Black et al., 2004; Feudale
productivity (NPP), 5 %–10 % of Alpine glacier loss, and and Shukla, 2011a, b). By applying the storyline approach,
many more human-health-related issues due to increased sur- we can consider both causal factors together and shed some
face ozone concentrations (Ciais et al., 2005; Fischer et al., additional insight on this event. The dry spring leading up to
2007; García-Herrera et al., 2010). the warm summer conditions was captured by initializing the
Both the June and August heatwaves were caused by sta- simulations by 1 March at the latest.
tionary anticyclonic circulations or blocking (Black et al., In Fig. 5a, the daily evolution of the domain-averaged
2004). The first block formed in June, then broke and quickly temperature at 2 m height for June–August for each of
reformed in July, which then caused the second heatwave in the ensemble members is plotted in comparison to the
August (García-Herrera et al., 2010). However, the extreme ECHAM_SN 5th–95th-percentile (1985–2005) climatology
temperatures cannot be explained by atmospheric blocking and ERA-Interim (Dee et al., 2011). The ECHAM_SN
alone. Due to large precipitation deficits in spring that year, 2003 temperature is also plotted for reference and shows a
the heatwaves happened in very dry conditions. The lack of strong coherence with the factual ensemble, confirming the
clouds and soil moisture caused latent heat transfer to turn appropriateness of using the ECHAM_SN climatology as a
into sensible heat transfer, which dramatically increased sur- reference for our factual simulations. The first thing to note
face temperatures (Bastos et al., 2014; Ciais et al., 2005; Fis- is that the factual and counterfactual ensembles evolve very
cher et al., 2007; Fink et al., 2004; Miralles et al., 2014). It similarly in time but (except for the third week of June) are
is considered highly unlikely that the 2003 European heat- well separated, by approximately 0.6 ◦ C, indicating a high
waves would have reached the temperatures they did with- signal-to-noise ratio at daily resolution for the domain av-

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

178 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

Figure 6. July and August divided into four half-month periods. Columns (a) and (b) show the European heatwave 2003, while columns (c)
and (d) show the Russian heatwave 2010. In columns (a) and (c), the factual geopotential height at z500 [m] is shown as black contour lines,
while temperatures at 2 m height [◦ C] are shown as shaded fields. In columns (b) and (d), the differences in 2 m temperature [◦ C] between
the factual and counterfactual simulations are shown as shaded fields. Stippling shows where all the factual members are > 0.1 ◦ C above all
the counterfactual members for that grid point. Note that the Russian domain is smaller; therefore the stippling has a different spacing than
in the European domain.

erage. This value of 0.6 ◦ C is in line with the global-mean and third half-month periods (16–31 July, 1–15 August), the
warming. ERA-Interim and the factual members show a temperatures in the factual simulations can locally be up to
strong correlation in time, although the ERA-Interim temper- 2–2.5 ◦ C higher than in the counterfactual simulations, with
atures are higher especially in June and during the heatwave the differences spread over a large area including Spain, Por-
in the first half of August. The factual temperatures exceed tugal, France, Germany, Hungary, and Romania. During the
the 95th percentile several times in June–August. In August, period 1–15 August, which according to Fig. 5a was the
the exceedance lasts for almost 2 weeks, whereas in June it peak of the heatwave, the hottest area in Europe (Fig. 6a)
does so for approximately 1 week. The counterfactual tem- is located in south-west France and southern Iberia. However
peratures are not quite so extreme; they exceed the 95th per- the largest differences between the factual and counterfactual
centile only for a few days at a time in June and August. simulations (up to 2.5 ◦ C) are found to the north of both of
Nevertheless, it is clear that there would have been a Euro- these regions, suggesting a shift in the peak temperature. In
pean heatwave in 2003 even without climate change, albeit the second half of August, there are still some strong temper-
with less extreme temperatures. This analysis thus supports ature differences visible over most of these regions, although
both of the perspectives on the event discussed earlier whilst the differences over western France have dampened.
providing a daily resolution of the climate change attribution. As noted earlier, the dryness of the soil has been identified
The temperature differences between the factual and coun- as an important contributing factor to the 2003 heatwave. Our
terfactual ensembles are spatially nonuniform over Europe. interest here, however, is on whether the soil wetness differed
In Fig. 6a the factual members’ average of the 2 m tempera- between factual and counterfactual conditions. In Fig. 7a we
ture and of the geopotential height (z500 ) show the meteoro- see a very similar decline in soil wetness for both the fac-
logical situation averaged over half-month periods following tual and counterfactual ensemble members from May until
García-Herrera et al. (2010). Figure 6b shows the local dif- the end of August. The counterfactual simulations start out
ferences in 2 m temperatures between the counterfactual and with somewhat higher soil wetness than the factual simula-
factual ensemble averages. Stippling is added to each grid tions, but over the course of the summer the values of both
point where all the three factual members are at least 0.1 ◦ C sets of simulations move closer towards each other so that
warmer than all the counterfactual members. There is strong by August the ensembles are close together. Thus it does not
local variance, especially during the heatwave in the first half appear that climate change had a first-order impact on soil
of August, with differences of up to 2.5 ◦ C. In the first period wetness in this case.
(1–15 July) the local differences are generally modest, except
in northern Spain, where they reach 1.5–2 ◦ C. In the second

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 179

Figure 7. Average soil wetness in the root zone [m] averaged over Europe in 2003 and Russia in 2010 during July and August of each year.
The factual simulations are shown in blue and the counterfactual simulations in red.

3.2 Russian heatwave 2010 atures correlate highly with the counterfactual members,
though are somewhat higher at the end of June and begin-
In August 2010 western Russia was hit by an unprecedented ning of July, and decline much more rapidly following the
heatwave caused by a large quasi-stationary anticyclonic cir- heatwave halfway through August. Starting after the second
culation, or blocking (Galarneau et al., 2012; Grumm, 2011; half of July, both the factual and counterfactual temperatures
Matsueda, 2011). It was a heatwave that broke all records exceed the 95th-percentile climatological temperature, peak
such as temperature anomalies during both day and night, around 8 August, and return to climatological temperatures
temporal duration, and spatial extent. The effect of soil wet- around 17 August. This analysis shows that this would have
ness, or rather the lack thereof, on the magnitude of the been an unprecedented event even without climate change.
temperatures was profound (Lau and Kim, 2012; Rasmijn The differences between the factual and counterfactual tem-
et al., 2018; Wehrli et al., 2019; Bastos et al., 2014). The peratures during the core of the heatwave are noticeably
2010 Russian heatwave is considered the most extreme heat- higher (about 2 ◦ C) than in the European heatwave 2003, as is
wave in Europe on record (Russo et al., 2015). Approxi- the spread between the ensemble members. In contrast to the
mately 50 000 lives were lost, 5000 km2 forest burned, 25 % European case, the anthropogenic warming during the core
of the crop failed and over USD 15 billion worth of economic of the heatwave is considerably higher than the global-mean
damage was recorded due to this heatwave (Barriopedro et warming. We attribute both aspects – the greatly enhanced
al., 2011; Lau and Kim, 2012; Otto et al., 2012; Rasmijn et anthropogenic warming and the larger internal variability –
al., 2018). In some of the attribution studies, the heatwave to the fact that the Russian domain is much farther inland
was primarily attributed to internal variability as the dynami- than the European domain, and thus the blocking conditions
cal situation strongly depended on the El Niño–Southern Os- cut off the influence of the SST forcing and allow a direct
cillation (ENSO) being in a La Niña state (Dole et al., 2011; radiative effect of GHG increases (Wehrli et al., 2019). Note
Russo et al., 2014; Schneidereit et al., 2012). However, the that western Russia is known for having large internal vari-
likelihood of the temperatures reaching such extreme values ability (Dole et al., 2011; Russo et al., 2014; Schneidereit et
has also been assessed as being significantly exacerbated by al., 2012), which is clearly apparent in our results. It is also
climate change (Otto et al., 2012; Rahmstorf and Coumou, the case that the Russian domain is smaller than the European
2011). As with the previous example, the storyline approach domain by a factor of 3.4, which would furthermore tend to
can represent both of these perspectives. Moreover, it over- increase the variability in the domain-averaged temperature
comes the limitation that the climate models used to perform shown in Fig. 5.
probabilistic event attribution generally have trouble repro- The range of temperature differences between factual and
ducing a blocking situation correctly (Trenberth and Fasullo, counterfactual simulations reach values up to 4 ◦ C locally,
2012; Watanabe et al., 2013). as seen in Fig. 6d. Note that the scale for the Russian heat-
In Fig. 5b, the daily evolution of the domain-averaged wave reaches up to 4.5 ◦ C, whereas the scale for the Euro-
temperature at 2 m height for each of the ensemble mem- pean heatwave reaches only 3 ◦ C. In the first half-month pe-
bers is shown in comparison to ECHAM_SN 2010, the riod (1–15 July), when the heatwave had not yet started, the
ECHAM_SN 5th–95th-percentile climatological tempera- local temperature differences are between 0.5–2.5 ◦ C, with
tures (1985–2015), and ERA-Interim. ERA-Interim temper- the maximum differences in the south-east of the domain.

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

180 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

Figure 8. Distributions across grid points of differences between ensemble members in temperature at 2 m height [◦ C], separated into the
four half-monthly periods. FF: differences between pairs of factual members; CC: differences between pairs of counterfactual members; FC:
differences between pairs of factual and counterfactual members. The boxes represent the 25th-to-75th-percentile range of the distributions,
the red lines represent the 50th percentiles (the median), and the blue bars indicate the 5th-to-95th-percentile range. The dashed horizontal
line indicates 1 ◦ C for reference. Columns (a) and (b) are for the European 2010 heatwave, and columns (c) and (d) for the Russian 2010
heatwave. Note the different vertical scales for the two events. Columns (a) and (c) show the differences in daily averages, and columns (b)
and (d) show the differences in half-monthly averages.

The temperature differences are largest in the core of the lapping and even cross each other in the beginning of Au-
block region, reaching up to 3.5 ◦ C in the south-east in the gust. These findings are in agreement with those of Hauser
second period (16–31 July) and up to 4 ◦ C in the south, be- et al. (2016), who reproduced the Russian heatwave under
low Moscow, in the third period (1–15 August). The block 1960 conditions and found that the dry conditions occurred
broke in the fourth period (16–31 August) and resulted in a there too, thus concluding they found no direct link between
virtual elimination of the temperature difference. In contrast the drought conditions and climate change. It must be em-
to the European heatwave 2003, here the biggest temperature phasized that this is not to downplay in any way the impact
differences between factual and counterfactual are found in of soil wetness on the event itself, which has been well estab-
the regions with the highest temperatures. lished in the literature. It is only to indicate that the impact
As with the European heatwave 2003, the differences in would have been there even without climate change.
soil wetness do not appear to be of first-order importance
to explain the temperature differences between the factual 3.3 Signal-to-noise analysis
and counterfactual simulations. In Fig. 7b the soil wetness in
the factual simulations is seen to decrease somewhat more The temperature differences found between the factual and
rapidly than in the counterfactual, which could be due to counterfactual simulations are meaningful if they are outside
the higher surface temperature and thus greater evaporation of the internal variability within each ensemble. A different
of soil moisture. However, the soil wetness values are over- way of saying this is that the differences are meaningful if the
two ensembles are distinguishable from each other. To assess

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 181

this in a statistical manner, temperature differences between approaches to extreme-event attribution, the effect of climate
pairs of factual members (FF), pairs of counterfactual mem- change on the occurrence likelihood of those dynamical con-
bers (CC), and factual–counterfactual pairs (FC) are plotted ditions is not assessed. In that respect, this approach is com-
for each half-month period in Fig. 8. The FF and CC pairs plementary to the more widely used probabilistic event at-
have a median close to 0 and represent the noise level; in tribution. However, since most results of probabilistic event
both cases there are three pairs (F1–F2, F2–F3, F3–F1; C1– attribution appeal in any case to the known thermodynamic
C2, C2–C3, C3–C1). The FC pairs contain the signal; here aspects of climate change, it can be argued that not much
there are nine pairs (F1–C1, F2–C2, F3–C3, F2–C1, F3–C2, is lost in the storyline approach, yet much is gained by the
F1–C3, F3–C1, F1–C2, F2–C3). Each box plot represents the specificity. This is especially the case for extreme events
distribution of 2 m temperature differences across the pairs whose dynamical conditions are not well represented in cli-
and across all grid points. The half-monthly panels represent mate models, e.g. blocking. Spectral nudging enables the re-
distributions of half-month-averaged values, and the daily production of extreme events with their particular dynami-
panels represent distributions of daily values within the half- cal details, allowing the same dynamical events to be repro-
month period. duced in simulations with different boundary conditions and
The daily differences for the European heatwave (Fig. 8a) thereby achieving a high signal-to-noise ratio of the climate
show a median value of about 0.6 ◦ C, irrespective of whether change effect. The combination of both methods – global
the time frame is during the heatwave itself or directly before spectral nudging and the storyline method – thus presents a
or directly after it, consistent with Fig. 5a. Although these are way to quantify, in great detail, the role of known thermody-
not really probability distributions (since they include contri- namic aspects of climate change, together with the specific
butions from different locations within the domain), we can dynamical conditions, in selected extreme events which hap-
use the inter-quartile ranges as measures of signal and noise. pened in the recent past. This can help reconcile the some-
The median difference for FC is above the 75th percentile of times different perspectives on those events that appear in
both CC and FF for daily values, giving confidence that our the literature (some emphasizing climate change, others em-
results are clearly above the noise level. Half-monthly time phasizing internal variability).
averages (Fig. 8b) produce nearly identical median values, We illustrated the method by applying it to two extreme
but we see that the spread is much smaller, as expected. The events that have been the subject of much study: the Euro-
25th percentile of FC now lies above the 75th percentile of pean heatwave of 2003 and the Russian heatwave of 2010.
the CC and FF boxes. By using a small ensemble of both factual and counterfac-
The differences between CF and either FF or CC for the tual simulations, we were able to determine a noise level for
Russian heatwave (Fig. 8c and d) are clearly larger than for our analysis. This revealed that the (conditional) signal of
the European heatwave and in contrast to the European case climate change is determinable at both daily timescales and
vary substantially between the different periods. Consistent local spatial scales. It follows that our methodology could
with Fig. 5b, in the periods outside of the core of the heat- be used to drive climate impact models and thus perform re-
wave (1–15 July, 16–31 August) the median difference be- alistic stress-testing of resilience strategies. With regard to
tween FC is about 1 ◦ C. Inside the core heatwave period the two heatwave examples, our analysis revealed a striking
(16–31 July, 1–15 August), however, the median difference contrast between the two events. In the European heatwave
is more like 2 ◦ C, reaching 2.2 ◦ C for 1–15 August. During of 2003, the effect of climate change was to increase tem-
this latter period the 5th-percentile whisker of half-monthly peratures across Europe by about the global-mean warming
FC differences is above the 75th percentile of FF and CC, level throughout the summer, and the heatwave was simply
which is a very strong signal indeed. When looking at the the dynamical event riding on top of that. In the Russian heat-
results for individual members, the larger internal variabil- wave of 2010, in contrast, the effect of climate change was
ity within the Russian domain (apparent also in Fig. 5b) is much higher than the global warming level and was particu-
clearly visible (not shown), as compared with the European larly enhanced, approximately threefold, during the peak of
case. the heatwave. We interpret this difference as reflecting the
role of direct GHG radiative forcing, which can become ap-
parent when air masses are cut off from marine influence.
4 Discussion and conclusion However, further analysis would be required to confirm this
hypothesis.
We have presented a detailed description and assessment of It is not possible to make a direct comparison between our
a global spectrally nudged storyline methodology to quantify results and probabilistic attribution of these heatwaves be-
the role of known thermodynamic aspects of climate change cause they are answering different questions, and the condi-
in specific extreme weather events. In this methodology, the tionalities are quite different. However, from a methodologi-
particular dynamical conditions leading to the event are taken cal perspective it is useful to contrast the nature of the attri-
as given, i.e. are regarded as random, and the attribution is bution statements that can be made using the different meth-
therefore highly conditional. Thus, as with all such storyline ods. We do this in Table 2 for the case of the Russian 2010

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

182 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

Table 2. Example of attribution statements that are possible using the probabilistic and storyline approaches for the case of the 2010 Russian
heatwave.

Probabilistic attribution Averaged over the Russian domain and over the month of July, temperatures in
(based on results from 2010 were 5 ◦ C above the 1960s climatology, of which 4 ◦ C was due to internal
Otto et al., 2012) variability, and 1 ◦ C was due to anthropogenic climate change.

The heatwave represented a 1-in-33-year event, which was 3 times more

likely than it would have been in the 1960s.
Storyline attribution Averaged over the Russian domain, temperatures in 2010 steadily increased from
(based on present results) the 1985–2015 climatology through the month of July until about 10 August, then
rapidly returned to climatology.

The domain-averaged heatwave reached 10 ◦ C above the 1985–2015 climatology

in early August, of which 8 ◦ C was due to internal variability, and 2 ◦ C was due to
anthropogenic climate change.

The anthropogenic component of the warming reached 4 ◦ C in the region to the

south of Moscow during the first half of August, where it exacerbated the already
warm temperatures there.

heatwave. Having said that, there is a continuum between RCM_CMIP6_Historical-HR (last access: 15 January 2021)
storyline and probabilistic approaches (Shepherd, 2016), and (Schupfner, 2021). For analysis we have used the open-access
it is possible to imagine intermediate set-ups which would Python packages.
provide a seamless connection between event attribution and
probabilistic weather prediction (NAS, 2016). These would
need to involve large ensembles (to calculate conditional Author contributions. LvG wrote the article, ran the simulations,
probabilities) and pay more attention to the self-consistency and analysed their results. FF and TGS conceived the study and
contributed to the writing and the interpretation of the results.
of how the counterfactual conditions are imposed. An ex-
ample is the recent use of an operational subseasonal-to-
seasonal prediction system, which involves modifying the at-
Competing interests. The authors declare that they have no conflict
mospheric state and land surface conditions as well as the of interest.
SSTs in generating the counterfactual (Wang et al., 2020).
The nudged global storyline method is an important step
towards a holistic approach within the attribution of individ- Acknowledgements. We would like to thank Sebastian Rast from
ual extreme events, which can quantify the role of both dy- the Max Planck Institute for Meteorology (MPI-M) in Hamburg for
namical variability and known thermodynamic aspects of cli- his technical support in applying spectral nudging in the ECHAM6
mate change and the interplay between them in great spatio- model and Matthias Bittner and Wolfgang Müller from MPI-M
temporal detail. As shown by Wehrli et al. (2020), the method for providing ECHAM6 CMIP6 data. Simulations were carried out
can easily be expanded to a larger number of storylines for on the MISTRAL supercomputer at the German Climate Comput-
both past and future. The method could also be applied to ing Centre (DKRZ), with technical support from Irina Fast. This
other extreme events affected thermodynamically by climate work is a contribution to the “Helmholtz Climate Initiative REK-
change such as tropical cyclones (Feser and Barcikowska, LIM” (Regional Climate Change), a joint research project of the
Helmholtz Association of German Research Centres (HGF), as well
2012). Our future applications are, therefore, intended to
as to the European Research Council advanced grant “Understand-
cover a wide variety of extreme events over the historical ing the atmospheric circulation response to climate change” (grant
record. no. 339390). The authors would like to thank Francis Zwiers and
an anonymous reviewer for their thoughtful and constructive com-
ments which helped improve the manuscript.
Code and data availability. The ECHAM6.1 global atmo-
spheric model is available from the Max Planck Institute for
Meteorology (MPI-M) website: https://code.mpimet.mpg.de/ Financial support. This research has been supported by the Euro-
projects/mpi-esm-users/files (last access: 15 January 2021) pean Research Council (grant no. 339390).
(Giorgetta et al., 2021). The CMIP6 historical simulation data
are archived at the World Data Centre for Climate (WDCC):
https://cera-www.dkrz.de/WDCC/ui/cerasearch/entry?acronym=

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 183

The article processing charges for this open-access anticipating the 2010 Russian heat wave?, Geophys. Res. Lett.,
publication were covered by a Research 38, L06702, https://doi.org/10.1029/2010gl046582, 2011.
Centre of the Helmholtz Association. Feser, F. and Barcikowska, M.: The influence of spectral nudging
on typhoon formation in regional climate models, Environ. Res.
Lett., 7, 014024, https://doi.org/10.1088/1748-9326/7/1/014024,
Review statement. This paper was edited by Joaquim G. Pinto and 2012.
reviewed by Francis Zwiers and one anonymous referee. Feudale, L. and Shukla, J.: Influence of sea surface tem-
perature on the European heat wave of 2003 summer.
Part I: an observational study, Clim. Dynam., 36, 1691–1703,
https://doi.org/10.1007/s00382-010-0788-0, 2011a.
References Feudale, L. and Shukla, J.: Influence of sea surface tem-
perature on the European heat wave of 2003 summer.
Barriopedro, D., Fischer, E. M., Luterbacher, J., Trigo, R., and Part II: a modeling study, Clim. Dynam., 36, 1705–1715,
Garcia-Herrera, R.: The Hot Summer of 2010: Redrawing the https://doi.org/10.1007/s00382-010-0789-z, 2011b.
Temperature Record Map of Europe, Science, 332, 220–224, Fink, A. H., Brücher, T., Krüger, A., Leckebusch, G. C., Pinto, J.
https://doi.org/10.1126/science.1201224, 2011. G., and Ulbrich, U.: The 2003 European summer heatwaves and
Bastos, A., Gouveia, C. M., Trigo, R. M., and Running, S. W.: drought–synoptic diagnosis and impacts, Weather, 59, 209–216,
Analysing the spatio-temporal impacts of the 2003 and 2010 ex- https://doi.org/10.1256/wea.73.04, 2004.
treme heatwaves on plant productivity in Europe, Biogeo- Fischer, E. M., Seneviratne, S. I., Vidale, P. L., Lüthi, D., and
sciences, 11, 3421–3435, https://doi.org/10.5194/bg-11-3421- Schär, C.: Soil moisture – Atmosphere interactions during the
2014, 2014. 2003 European summer heat wave, J. Climate, 20, 5081–5099,
Black, E., Blackburn, M., Harrison, G., Hoskins, B., and Methven, https://doi.org/10.1175/jcli4288.1, 2007.
J.: Factors contributing to the summer 2003 European heatwave, Galarneau, T. J., Hamill, T. M., Dole, R. M., and Perlwitz, J.:
Weather, 59, 217–223, https://doi.org/10.1256/wea.74.04, 2004. A Multiscale Analysis of the Extreme Weather Events over
Bony, S., Bellon, G., Klocke, D., Sherwood, S., Fermepin, S., and Western Russia and Northern Pakistan during July 2010, Mon.
Denvil, S.: Robust direct effect of carbon dioxide on tropical Weather Rev., 140, 1639–1664, https://doi.org/10.1175/mwr-d-
circulation and regional precipitation, Nat. Geosci., 6, 447–451, 11-00191.1, 2012.
https://doi.org/10.1038/ngeo1799, 2013. García-Herrera, R., Díaz, J., Trigo, R. M., Luterbacher, J., and
Cattiaux, J., Vautard, R., Cassou, C., Yiou, P., Masson-Delmotte, Fischer, E. M.: A review of the European summer heat wave
V., and Codron, F.: Winter 2010 in Europe: A cold ex- of 2003, Crit. Rev. Environ. Sci. Technol., 40, 267–306,
treme in a warming climate, Geophys. Res. Lett., 37, L20704, https://doi.org/10.1080/10643380802238137, 2010.
https://doi.org/10.1029/2010GL044613, 2010. Giorgetta, M. A., Roeckner, E., Mauritsen, T., Bader, J., Crueger,
Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogee, J., Al- T., Esch, M., Rast, S., Kornblueh, L., Schmidt, H., Kinne,
lard, V., Aubinet, M., Buchmann, N., Bernhofer, C., Carrara, S., Hohenegger, C., Möbis, B., Krismer, T., Wieners, K.-
A., Chevallier, F., De Noblet, N., Friend, A. D., Friedlingstein, H., and Stevens, B.: ECHAM 6.1.00, available at: https://
P., Grunwald, T., Heinesch, B., Keronen, P., Knohl, A., Krin- code.mpimet.mpg.de/projects/mpi-esm-users/files, last access:
ner, G., Loustau, D., Manca, G., Matteucci, G., Miglietta, F., 15 January 2021.
Ourcival, J. M., Papale, D., Pilegaard, K., Rambal, S., Seufert, Grumm, R. H.: The central European and Russian heat event
G., Soussana, J. F., Sanz, M. J., Schulze, E. D., Vesala, T., of July–August 2010, B. Am. Meteorol. Soc., 92, 1285–1296,
and Valentini, R.: Europe-wide reduction in primary productivity https://doi.org/10.1175/2011BAMS3174.1, 2011.
caused by the heat and drought in 2003, Nature, 437, 529–533, Hannart, A., Pearl, J., Otto, F. E. L., Naveau, P., and Ghil, M.:
https://doi.org/10.1038/nature03972, 2005. Causal Counterfactual Theory for the Attribution of Weather and
Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, Climate-Related Events, B. Am. Meteorol. Soc., 97, 99–110,
P., Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., https://doi.org/10.1175/bams-d-14-00034.1, 2016.
Bauer, P., Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bid- Hauser, M., Orth, R., and Seneviratne, S. I.: Role of soil
lot, J., Bormann, N., Delsol, C., Dragani, R., Fuentes, M., Geer, moisture versus recent climate change for the 2010 heat
A. J., Haimberger, L., Healy, S. B., Hersbach, H., Holm, E. V., wave in western Russia, Geophys. Res. Lett., 43, 2819–2826,
Isaksen, L., Kallberg, P., Kohler, M., Matricardi, M., McNally, https://doi.org/10.1002/2016gl068036, 2016.
A. P., Monge-Sanz, B. M., Morcrette, J. J., Park, B. K., Peubey, Horton, D. E., Johnson, N. C., Singh, D., Swain, D. L., Rajaratnam,
C., de Rosnay, P., Tavolato, C., Thepaut, J. N., and Vitart, F.: The B., and Diffenbaugh, N. S.: Contribution of changes in atmo-
ERA-Interim reanalysis: configuration and performance of the spheric circulation patterns to extreme temperature trends, Na-
data assimilation system, Q. J. Roy. Meteorol. Soc., 137, 553– ture, 522, 465–469, https://doi.org/10.1038/nature14550, 2015.
597, https://doi.org/10.1002/qj.828, 2011. Hoskins, B. and Woollings, T.: Persistent extratropical regimes
Deser, C., Terray, L., and Phillips, A. S.: Forced and Internal Com- and climate extremes, Curr. Clim. Change Rep., 1, 115–124,
ponents of Winter Air Temperature Trends over North America https://doi.org/10.1007/s40641-015-0020-8, 2015.
during the past 50 Years: Mechanisms and Implications, J. Cli- Hurrell, J. W., Hack, J. J., Shea, D., Caron, J. M., and Rosinski, J.:
mate, 29, 2237–2258, https://doi.org/10.1175/jcli-d-15-0304.1, A new sea surface temperature and sea ice boundary dataset for
2016. the Community Atmosphere Model, J. Climate, 21, 5145–5153,
Dole, R., Hoerling, M., Perlwitz, J., Eischeid, J., Pegion, P., Zhang, https://doi.org/10.1175/2008JCLI2292.1, 2008.
T., Quan, X. W., Xu, T. Y., and Murray, D.: Was there a basis for

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

184 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

IPCC: Managing the Risks of Extreme Events and Disasters to Ad- Meredith, E. P., Semenov, V. A., Maraun, D., Park, W., and Cher-
vance Climate Change Adaptation, in: A Special Report of Work- nokulsky, A. V.: Crucial role of Black Sea warming in ampli-
ing Groups I and II of the Intergovernmental Panel on Climate fying the 2012 Krymsk precipitation extreme, Nat. Geosci., 8,
Change, edited by: Field, C. B., Barros, V., Stocker, T. F., Qin, 615–619, https://doi.org/10.1038/ngeo2483, 2015.
D., Dokken, D. J., Ebi, K. L., Mastrandrea, M. D., Mach, K. J., Miguez-Macho, G., Stenchikov, G. L., and Robock, A.: Spectral
Plattner, G.-K., Allen, S. K., Tignor, M., and Midgley, P. M., nudging to eliminate the effects of domain position and geometry
Cambridge University Press, Cambridge, UK, and New York, in regional climate model simulations, J. Geophys. Res.-Atmos.,
NY, USA, 582 pp., 2012. 109, D13104, https://doi.org/10.1029/2003jd004495, 2004.
IPCC – Inter-governmental Panel on Climate Change: Summary for Miralles, D. G., Teuling, A. J., van Heerwaarden, C. C., and Vilà-
Policymakers, Global Warming of 1.5 ◦ C: An IPCC Special Re- Guerau de Arellano, J.: Mega-heatwave temperatures due to
port on the Impacts of Global Warming of 1.5 ◦ C Above Pre- combined soil desiccation and atmospheric heat accumulation,
industrial Levels and Related Global Greenhouse Gas Emission Nat. Geosci., 7, 345–349, https://doi.org/10.1038/ngeo2141,
Pathways, in: the Context of Strengthening the Global Response 2014.
to the Threat of Climate Change, Sustainable Development, and Müller, W. A., Jungclaus, J. H., Mauritsen, T., Baehr, J., Bit-
Efforts to Eradicate Poverty, edited by: Masson-Delmotte, V., tner, M., Budich, R., Bunzel, F., Esch, M., Ghosh, R.,
Zhai, P., Pörtner, H.-O., Roberts, D., Skea, J., Shukla, P., Pi- Haak, H., Ilyina, T., Kleine, T., Kornblueh, L., Li, H.,
rani, A., Moufouma-Okia, W., Péan, C., Pidcock, R., Connors, Modali, K., Notz, D., Pohlmann, H., Roeckner, E., Stemm-
S., Matthews, J., Chen, Y., Zhou, X., Gomis, M., Lonnoy, E., ler, I., Tian, F., and Marotzke, J.: A Higher-resolution Ver-
Maycock, T., Tignor, M., and Watefield, T., World Meteorologi- sion of the Max Planck Institute Earth System Model (MPI-
cal Organization, Geneva, Switzerland, 32 pp., 2018. ESM1.2-HR), J. Adv. Model. Earth Syst., 10, 1383–1413,
Jeuken, A. B. M., Siegmund, P. C., Heijboer, L. C., Feichter, J., https://doi.org/10.1029/2017ms001217, 2018.
and Bengtsson, L.: On the potential of assimilating meteoro- Muthers, S., Laschewski, G., and Matzarakis, A.: The Sum-
logical analyses in a global climate model for the purpose of mers 2003 and 2015 in South-West Germany: Heat Waves and
model validation, J. Geophys. Res.-Atmos., 101, 16939–16950, Heat-Related Mortality in the Context of Climate Change, At-
https://doi.org/10.1029/96jd01218, 1996. mosphere, 8, 224, https://doi.org/10.3390/atmos8110224, 2017.
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, NAS – National Academies of Sciences Engineering and Medicine:
D., Gandin, L., Iredell, M., Saha, S., White, G., Woollen, Attribution of Extreme Weather Events in the Context of Cli-
J., Zhu, Y., Chelliah, M., Ebisuzaki, W., Higgins, W., mate Change, The National Academies Press, Washington, D.C.,
Janowiak, J., Mo, K. C., Ropelewski, C., Wang, J., Leet- https://doi.org/10.17226/21852, 2016.
maa, A., Reynolds, R., Jenne, R., and Joseph, D.: The Otto, F. E. L.: Attribution of Weather and Climate Events, Annu.
NCEP/NCAR 40-year reanalysis project, B. Am. Me- Rev. Environ. Resour., 42, https://doi.org/10.1146/annurev-
teorol. Soc., 77, 437–471, https://doi.org/10.1175/1520- environ-102016-060847, 627–646, 2017.
0477(1996)077<0437:Tnyrp>2.0.Co;2, 1996. Otto, F. E. L., Massey, N., van Oldenborgh, G. J., Jones, R. G.,
Kim, J. E. and Hong, S. Y.: A Global Atmospheric Analysis Dataset and Allen, M. R.: Reconciling two approaches to attribution of
Downscaled from the NCEP-DOE Reanalysis, J. Climate, 25, the 2010 Russian heat wave, Geophys. Res. Lett., 39, L04702,
2527–2534, https://doi.org/10.1175/jcli-d-11-00534.1, 2012. https://doi.org/10.1029/2011gl050422, 2012.
Lau, W. K. M. and Kim, K. M.: The 2010 Pakistan Flood Patricola, C. M. and Wehner, M. F.: Anthropogenic influences
and Russian Heat Wave: Teleconnection of Hydrome- on major tropical cyclone events, Nature, 563, 339–346,
teorological Extremes, J. Hydrometeorol., 13, 392–403, https://doi.org/10.1038/s41586-018-0673-2, 2018.
https://doi.org/10.1175/jhm-d-11-016.1, 2012. Pearl, J. and Mackenzie, D.: The book of Why: The new science of
Lloyd, E. A. and Oreskes, N.: Climate Change Attribution: When Is cause and effect, Basic Books, New York, 2018.
It Appropriate to Accept New Methods?, Earths Future, 6, 311– Pfahl, S., O’Gorman, P. A., and Fischer, E. M.: Under-
325, https://doi.org/10.1002/2017ef000665, 2018. standing the regional pattern of projected future changes
Lloyd, E. A. and Shepherd, T. G.: Environmental catastrophes, cli- in extreme precipitation, Nat. Clim. Change, 7, 423–427,
mate change, and attribution, Ann. NY Acad. Sci., 1469, 105– https://doi.org/10.1038/nclimate3287, 2017.
124, https://doi.org/10.1111/nyas.14308, 2020. Rahmstorf, S. and Coumou, D.: Increase of extreme events in a
Luterbacher, J., Dietrich, D., Xoplaki, E., Grosjean, M., and Wan- warming world, P. Natl. Acad. Sci. USA, 108, 17905–17909,
ner, H.: European seasonal and annual temperature variabil- https://doi.org/10.1073/pnas.1101766108, 2011.
ity, trends, and extremes since 1500, Science, 303, 1499–1503, Rasmijn, L. M., van der Schrier, G., Bintanja, R., Barkmeijer, J.,
https://doi.org/10.1126/science.1093877, 2004. Sterl, A., and Hazeleger, W.: Future equivalent of 2010 Rus-
Matsueda, M.: Predictability of Euro-Russian blocking in sian heatwave intensified by weakening soil moisture constraints,
summer of 2010, Geophys. Res. Lett., 38, L06801, Nat. Clim. Change, 8, 381–385, https://doi.org/10.1038/s41558-
https://doi.org/10.1029/2010gl046557, 2011. 018-0114-0, 2018.
Meinshausen, M., Smith, S. J., Calvin, K., Daniel, J. S., Kainuma, Rayner, N. A., Brohan, P., Parker, D. E., Folland, C. K., Kennedy,
M. L. T., Lamarque, J. F., Matsumoto, K., Montzka, S. A., Raper, J. J., Vanicek, M., Ansell, T. J., and Tett, S. F. B.: Im-
S. C. B., Riahi, K., Thomson, A., Velders, G. J. M., and van Vu- proved Analyses of Changes and Uncertainties in Sea Sur-
uren, D. P. P.: The RCP greenhouse gas concentrations and their face Temperature Measured In Situ since the Mid-Nineteenth
extensions from 1765 to 2300, Climatic Change, 109, 213–241, Century: The HadSST2 Dataset, J. Climate, 19, 446–469,
https://doi.org/10.1007/s10584-011-0156-z, 2011. https://doi.org/10.1175/jcli3637.1, 2006.

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021

L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events 185

Reick, C. H., Raddatz, T., Brovkin, V., and Gayler, V.: Rep- cus, R., Reichler, T., and Roeckner, E.: Atmospheric component
resentation of natural and anthropogenic land cover change of the MPI-M Earth System Model: ECHAM6, J. Adv. Model.
in MPI-ESM, J. Adv. Model. Earth Syst., 5, 459–482, Earth Syst., 5, 146–172, https://doi.org/10.1002/jame.20015,
https://doi.org/10.1002/jame.20022, 2013. 2013.
Russo, S., Dosio, A., Graversen, R. G., Sillmann, J., Car- Stott, P. A., Stone, D. A., and Allen, M. R.: Human contribu-
rao, H., Dunbar, M. B., Singleton, A., Montagna, P., Bar- tion to the European heatwave of 2003, Nature, 432, 610–614,
bola, P., and Vogt, J. V.: Magnitude of extreme heat https://doi.org/10.1038/nature03089, 2004.
waves in present climate and their projection in a warm- Stott, P. A., Christidis, N., Otto, F. E. L., Sun, Y., Vanderlinden, J.
ing world, J. Geophys. Res.-Atmos., 119, 12500–12512, P., van Oldenborgh, G. J., Vautard, R., von Storch, H., Walton,
https://doi.org/10.1002/2014jd022098, 2014. P., Yiou, P., and Zwiers, F. W.: Attribution of extreme weather
Russo, S., Sillmann, J., and Fischer, E. M.: Top ten European heat- and climate-related events, Wiley Interdisciplin. Rev. – Clim.
waves since 1950 and their occurrence in the coming decades, Change, 7, 23–41, https://doi.org/10.1002/wcc.380, 2016.
Environ. Res. Lett., 10, 124003, https://doi.org/10.1088/1748- Trenberth, K. E. and Fasullo, J. T.: Climate extremes and
9326/10/12/124003, 2015. climate change: The Russian heat wave and other climate
Schär, C., Frei, C., Lüthi, D., and Davies, H. C.: Surrogate climate- extremes of 2010, J. Geophys. Res.-Atmos., 117, D17103,
change scenarios for regional climate models, Geophys. Res. https://doi.org/10.1029/2012jd018020, 2012.
Lett., 23, 669–672, https://doi.org/10.1029/96GL00265, 1996. Trenberth, K. E., Fasullo, J. T., and Shepherd, T. G.: Attribu-
Schär, C., Vidale, P. L., Lüthi, D., Frei, C., Häberli, C., Liniger, tion of climate extreme events, Nat. Clim. Change, 5, 725–730,
M. A., and Appenzeller, C.: The role of increasing temperature https://doi.org/10.1038/nclimate2657, 2015.
variability in European summer heatwaves, Nature, 427, 332– Trigo, R. M., García-Herrera, R., Díaz, J., Trigo, I. F., and Va-
336, https://doi.org/10.1038/nature02300, 2004. lente, M. A.: How exceptional was the early August 2003
Schneider, D. P., Deser, C., and Okumura, Y.: An assess- heatwave in France?, Geophys. Res. Lett., 32, L10701,
ment and interpretation of the observed warming of West https://doi.org/10.1029/2005gl022410, 2005.
Antarctica in the austral spring, Clim. Dynam., 38, 323–347, van der Wiel, K., Selten, F. M., Bintanja, R., Blackport, R., and
https://doi.org/10.1007/s00382-010-0985-x, 2012. Screen, J. A.: Ensemble climate-impact modelling: extreme im-
Schneidereit, A., Schubert, S., Vargin, P., Lunkeit, F., Zhu, X. H., pacts from moderate meteorological conditions, Environ. Res.
Peters, D. H. W., and Fraedrich, K.: Large-Scale Flow and the Lett., 15, 034050, https://doi.org/10.1088/1748-9326/ab7668,
Long-Lasting Blocking High over Russia: Summer 2010, Mon. 2020.
Weather Rev., 140, 2967–2981, https://doi.org/10.1175/mwr-d- van Niekerk, A., Shepherd, T. G., Vosper, S. B., and Webster, S.:
11-00249.1, 2012. Sensitivity of resolved and parametrized surface drag to changes
Schubert-Frisius, M., Feser, F., von Storch, H., and Rast, S.: Opti- in resolution and parametrization, Q. J. Roy. Meteorol. Soc., 142,
mal Spectral Nudging for Global Dynamic Downscaling, Mon. 2300–2313, https://doi.org/10.1002/qj.2821, 2016.
Weather Rev., 145, 909–927, https://doi.org/10.1175/mwr-d-16- Vautard, R., Yiou, P., Otto, F., Stott, P., Christidis, N., Van
0036.1, 2017. Oldenborgh, G. J., and Schaller, N.: Attribution of human-
Schupfner, M.: CMIP6 Scenario MIP DKRZ MPI-ESM1- induced dynamical and thermodynamical contributions in
2-HR historical – RCM-forcing data, available at: extreme weather events, Environ. Res. Lett., 11, 114009,
https://cera-www.dkrz.de/WDCC/ui/cerasearch/entry? https://doi.org/10.1088/1748-9326/11/11/114009, 2016.
acronym=RCM_CMIP6_Historical-HR, last access: 15 Jan- von Storch, H., Langenberg, H., and Feser, F.: A spectral
uary 2021. nudging technique for dynamical downscaling purposes, Mon.
Scinocca, J. F., Kharin, V. V., Jiao, Y., Qian, M. W., Lazare, M., Sol- Weather Rev., 128, 3664–3673, https://doi.org/10.1175/1520-
heim, L., Flato, G. M., Biner, S., Desgagne, M., and Dugas, B.: 0493(2000)128<3664:Asntfd>2.0.Co;2, 2000.
Coordinated Global and Regional Climate Modeling, J. Climate, von Storch, H., Cavicchia, L., Feser, F., and Li, D. L.: The Concept
29, 17–35, https://doi.org/10.1175/jcli-d-15-0161.1, 2015. of Large-Scale Conditioning of Climate Model Simulations of
Screen, J. A. and Simmonds, I.: Increasing fall-winter en- Atmospheric Coastal Dynamics: Current State and Perspectives,
ergy loss from the Arctic Ocean and its role in Arctic Atmosphere, 9, 337, https://doi.org/10.3390/atmos9090337,
temperature amplification, Geophys. Res. Lett., 37, L16707, 2018.
https://doi.org/10.1029/2010GL044136, 2010. Waldron, K. M., Paegle, J., and Horel, J. D.: Sensi-
Shepherd, T. G.: Dynamics, stratospheric ozone, tivity of a spectrally filtered and nudged limited-
and climate change, Atmos.-Ocean, 46, 117–138, area model to outer model options, Mon. Weather
https://doi.org/10.3137/ao.460106, 2008. Rev., 124, 529–547, https://doi.org/10.1175/1520-
Shepherd, T. G.: Atmospheric circulation as a source of uncer- 0493(1996)124<0529:Soasfa>2.0.Co;2, 1996.
tainty in climate change projections, Nat. Geosci., 7, 703–708, Wallace, J. M., Fu, Q., Smoliak, B. V., Lin, P., and Johan-
https://doi.org/10.1038/ngeo2253, 2014. son, C. M.: Simulated versus observed patterns of warming
Shepherd, T. G.: A Common Framework for Approaches to Ex- over the extratropical Northern Hemisphere continents during
treme Event Attribution, Curr. Clim. Change Rep., 2, 28–38, the cold season, P. Natl. Acad. Sci. USA, 109, 14337–14342,
https://doi.org/10.1007/s40641-016-0033-y, 2016. https://doi.org/10.1073/pnas.1204875109, 2012.
Stevens, B., Giorgetta, M., Esch, M., Mauritsen, T., Crueger, T., Wang, G., Hope, P., Lim, E. P., Hendon, H. H., and Ar-
Rast, S., Salzmann, M., Schmidt, H., Bader, J., Block, K., blaster, J. M.: An initialized attribution method of extreme
Brokopf, R., Fast, I., Kinne, S., Kornblueh, L., Lohmann, U., Pin-

https://doi.org/10.5194/nhess-21-171-2021 Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021

186 L. van Garderen et al.: A methodology for attributing the role of climate change in extreme events

events on subseasonal to seasonal time-scales, J. Climate, Wehrli, K., Hauser, M., and Seneviratne, S. I.: Storylines of
https://doi.org/10.1175/JCLI-D-19-1021.1, in press, 2020. the 2018 Northern Hemisphere heatwave at pre-industrial and
Watanabe, M., Shiogama, H., Imada, Y., Mori, M., Ishii, M., higher global warming levels, Earth Syst. Dynam., 11, 855–873,
and Kimoto, M.: Event attribution of the August 2010 Rus- https://doi.org/10.5194/esd-11-855-2020, 2020.
sian heat wave, Scient. Online Lett. Atmos., 9, 65–68, Yoshimura, K. and Kanamitsu, M.: Dynamical global downscal-
https://doi.org/10.2151/sola.2013-015, 2013. ing of global reanalysis, Mon. Weather Rev., 136, 2983–2998,
Wehrli, K., Guillod, B. P., Hauser, M., Leclair, M., and Senevi- https://doi.org/10.1175/2008mwr2281.1, 2008.
ratne, S. I.: Identifying Key Driving Processes of Major Re-
cent Heat Waves, J. Geophys. Res.-Atmos., 124, 11746–11765,
https://doi.org/10.1029/2019JD030635, 2019.

Nat. Hazards Earth Syst. Sci., 21, 171–186, 2021 https://doi.org/10.5194/nhess-21-171-2021