The 2018–2020 Multi‐Year Drought Sets a New Benchmark in Europe

Abstract

During the period 2018–20, Europe experienced a series of hot and dry weather conditions with significant socioeconomic and environmental consequences. Yet, the extremity of these multi-year dry conditions is not recognized. Here, we provide a comprehensive spatio-temporal assessment of the drought hazard over Europe by benchmarking past exceptional events during the period from 1766–2020. We identified the 2018–20 drought event as a new benchmark having an unprecedented intensity that persisted for more than two years, exhibiting a mean areal coverage of 35.6% and an average duration of 12.2 months. What makes this event truly exceptional compared with past events is its near-surface air temperature anomaly reaching +2.8 K, which constitutes a further evidence that the ongoing global warming is exacerbating present drought events. Furthermore, future events based on climate model simulations (CMIP5) suggest that Europe should be prepared for events of comparable intensity as the 2018–20 event but with durations longer than any of those experienced in the last 250 years. Our study thus emphasizes the urgent need for adaption and mitigation strategies to cope with such multi-year drought events across Europe.

Full text

1. Introduction
The observed increase in the frequency of droughts and heatwaves over the Northern Hemisphere in the 21st
century poses immediate socio-economic threats affecting the well-being of the people by triggering negative
health effects. These adverse hydro-meteorological conditions can lead to agricultural and ecological impacts
such as crop losses, poor water quality conditions in water bodies, and wildfires. The reduction of the streamflow
resulting from a drought event combined with high air temperatures also creates a threat to existing infrastruc-
ture. Several authors have reported cases of reduction of the cooling capacity in power plants, the reduction of
tonnage in fluvial transportation, and the drop in reservoir storage leading to drinking water shortages (Naumann
etal.,2021; Peichl etal.,2018; Stanke etal.,2013; Vogel etal.,2019). Europe, in particular, has experienced a
series of dry summers with substantial socioeconomic and environmental impacts in 2003 (Fischer etal.,2007),
2010 (Flach etal.,2018), 2015 (Van Lanen etal.,2016) and 2018–2020 (Hari etal.,2020; Peters etal., 2020).
The latest report of the European Commission estimates that the current annual monetary losses across Europe
due to droughts were 9 billion EUR every year. Depending on the region, between 39% and 60% of the losses are
related to agriculture, 22%–48% to the energy sector, while 9%–20% of the total damages correspond to public
water supply systems (Naumann etal.,2021). Besides direct financial losses, the natural net ecosystem carbon
uptake can get further significantly reduced by drought conditions (Ciais etal.,2005).
Abstract During the period 2018–2020, Europe experienced a series of hot and dry weather conditions
with significant socioeconomic and environmental consequences. Yet, the extremity of these multi-year dry
conditions is not recognized. Here, we provide a comprehensive spatio-temporal assessment of the drought
hazard over Europe by benchmarking past exceptional events during the period from 1766 to 2020. We
identified the 2018–2020 drought event as a new benchmark having an unprecedented intensity that persisted
for more than 2years, exhibiting a mean areal coverage of 35.6% and an average duration of 12.2months.
What makes this event truly exceptional compared with past events is its near-surface air temperature anomaly
reaching +2.8K, which constitutes a further evidence that the ongoing global warming is exacerbating
present drought events. Furthermore, future events based on climate model simulations Coupled Model
Intercomparison Project v5 suggest that Europe should be prepared for events of comparable intensity as the
2018–2020 event but with durations longer than any of those experienced in the last 250years. Our study thus
emphasizes the urgent need for adaption and mitigation strategies to cope with such multi-year drought events
across Europe.
Plain Language Summary This manuscript demonstrates that the 2018–2020 multi-year drought
event constitutes a new benchmark in Europe, with an unprecedented level of intensity over the past 250years.
What makes this event truly exceptional compared with past events is its temperature anomaly reaching
+2.8K. This finding provides new evidence that the ongoing global warming exacerbates current drought
events. The key message of this study is that the projected future events across the European continent will
have a comparable intensity as the 2018–2020 drought but exhibit considerably longer durations than any of
those observed during the last 250years. Our analysis also shows that these exceptional temperature-enhanced
droughts significantly negatively impact commodity crops across Europe.
RAKOVEC ET AL.
© 2022 The Authors. Earth's Future
published by Wiley Periodicals LLC on
behalf of American Geophysical Union.
This is an open access article under
the terms of the Creative Commons
Attribution License, which permits use,
distribution and reproduction in any
medium, provided the original work is
properly cited.
The 2018–2020 Multi-Year Drought Sets a New Benchmark in
Europe
Oldrich Rakovec1,2 , Luis Samaniego1 , Vittal Hari1,3 , Yannis Markonis2 ,
Vojtěch Moravec2,4 , Stephan Thober1 , Martin Hanel2,4 , and Rohini Kumar1
1UFZ-Helmholtz Centre for Environmental Research, Leipzig, Germany, 2Faculty of Environmental Sciences, Czech
University of Life Sciences Prague, Praha-Suchdol, Czech Republic, 3Department of Environmental Science and Engineering,
Indian Institute of Technology (ISM) Dhanbad, Dhanbad, India, 4T. G. Masaryk Water Research Institute, Praha 6, Czech
Republic
Key Points:
• The 2018–2020 multi-year drought
shows unprecedented level of intensity
during the past 250years
• The 2018–2020 event reached record-
breaking +2.8K temperature anomaly
and negatively impacted major crops
• Future drought events reach
comparable intensity of 2018–2020
but with considerably longer durations
Supporting Information:
Supporting Information may be found in
the online version of this article.
Correspondence to:
O. Rakovec, L. Samaniego, and R.
Kumar,
oldrich.rakovec@ufz.de;
luis.samaniego@ufz.de;
rohini.kumar@ufz.de
Citation:
Rakovec, O., Samaniego, L., Hari, V.,
Markonis, Y., Moravec, V., Thober, S.,
etal. (2022). The 2018–2020 multi-year
drought sets a new benchmark in Europe.
Earth's Future, 10, e2021EF002394.
https://doi.org/10.1029/2021EF002394
Received 25 AUG 2021
Accepted 11 MAR 2022
10.1029/2021EF002394
RESEARCH ARTICLE
1 of 11

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
2 of 11
The reconstructions of hydro-climatic data suggest that the recent sequence of European summer droughts
is unprecedented during the past centuries (Büntgen et al., 2021)—with reported several multi-year droughts
(2018–2019 [Hari etal.,2020] and 2014–2018 [Moravec etal.,2021]). While the characteristics and impacts
of individual drought years are well described in the literature (Moravec etal.,2019; Peters etal., 2020; Van
Lanen etal.,2016), the simultaneous evolution of droughts in space and time is rarely explored (Herrera-Estrada
etal.,2017). Furthermore, the severity of the recent droughts highlights the concern that global warming may
be significantly contributing to their evolution (Williams etal.,2020) and that its effects will continue to exac-
erbate them in the future (Hari etal.,2020; Samaniego etal.,2018). A comprehensive long-span benchmarking
of droughts, considering their space-time evolution, is therefore urgently needed to be able to place the recent
multi-year drought within the variability in the observational period and the near and long-term projections. By
doing so, we substantiate how exceptional this recent drought event is in the historical record and the likelihood
of such events under different future climate scenarios.
In this study, we benchmark the recent droughts from a long-term perspective taking into account the spatio-tem-
poral evolution of drought propagation across Europe. Our drought analysis is based on characterizing anom-
alous conditions of root-zone soil moisture that reflect the antecedent and contemporary hydro-meteorologic
conditions and constitutes the primary source of water for plant growth (Andreadis etal., 2005; Seneviratne
et al., 2010). The soil moisture is estimated with the well-established mesoscale Hydrologic Model (mHM)
(Kumar etal.,2013; Samaniego etal.,2010) (Section2.3). The monthly soil moisture estimates are transformed
into a percentile-based monthly soil moisture index (SMI (Samaniego etal.,2013))—which is then taken as a
basis for clustering the space-time evolution of soil moisture droughts (Samaniego etal., 2013,2018). Finally,
we quantify the drought characteristics (i.e., areal extent, duration, total drought magnitude, and intensity, see
Section2.4) from 1766 to 2100; the period 1766–2020 corresponds to observation-based simulation (reference),
and further, we quantify these characteristics in the climate models and also analyze the changes of these under
moderate and high emission scenarios (1950–2100; see Section2.1).
2. Data and Methods
2.1. Meteorological Model Forcing Data
2.1.1. Observation-Based Historical Forcing Data
This study is mainly based on the precipitation and near-surface air temperature observation-based data sets for
the period 1766–2015 (Casty etal.,2007) and E-OBS v21 (Hofstra etal.,2009) (daily gridded observational data
set for precipitation, and temperature in Europe) available for the period 1950–2020. The overlapping period
allowed for correction of biases (Gudmundsson etal., 2012) in the data set of Casty with respect to E-OBS
(Hanel etal.,2018; Moravec etal.,2021), and the merged meteorologic product is used as an input into the mHM
(Kumar etal.,2013; Samaniego etal.,2010) to obtain the soil moisture simulations. The estimates of potential
evapotranspiration required by mHM are based on mean near-surface air temperature and approximations for
extraterrestrial solar radiation (Oudin etal.,2005) during the historical period.
2.1.2. Climate Model Forcing Data
Besides the observation-based model forcings, we utilized the bias corrected daily precipitation, average, maxi-
mum and minimum temperature fields at 0.5° resolution were made available by the ISI-MIP project (Warszawski
etal., 2014). This suite is based on five Coupled Model Intercomparison Project v5 (CMIP5) Global Climate
Models (HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, GFDL-ESM2M and NorESM1-M) in
the historical mode and under two future representative concentration pathways (RCP4.5 and RCP8.5). The
trend-preserving bias correction method (Hempel et al., 2013) was used in the ISI-MIP project to match the
climatology of the CMIP5 GCMs with observations. We refer to Frieler etal.(2017) and Warszawski etal.(2014)
for a full description of the ISIMIP experiment, as detailed descriptions of the climate model projection used in
this study are beyond the scope of the present study. The five models selected for the ISI-MIP project follow the
strategy to represent warm and wet and cold and dry climates, which are most relevant for impact studies (Frieler
etal.,2017). Each of the GCMs is available at a single realization. Thus internal climate variability and uncer-
tainty due to initial conditions are not considered in the ISIMIP-2b experiment. The Penman-Monteith potential
evapotranspiration method (Allen etal.,1998) also needed as forcing for mHM is based on the surface energy
budget (i.e., Rn-G=SH+LH), where Rn, G, SH, and LH are the net radiation, ground heat flux, and sensible

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
3 of 11
and latent heat flux (Scheff & Frierson,2014), respectively. This approach implicitly considers the ambient CO2
effects on plant transpiration, vegetation growth (Trnka etal.,2019).
2.2. Crop Yield Data
We utilized the data of agriculture yields 1961–2019 from Food and Agriculture Organization of the United
Nations (FAO Global Statistical Yearbook,2021) and harvested production and areas under cultivation 2010–
2020 from EUROSTAT(2021). The overlapping period 2010–2019 was used for data quality checks between
both data sets and supplementing missing data. We considered three dominant kinds of cereals: wheat, grain
maize and barley as representative set of agricultural production. To account for technological advances (e.g.,
improvements in plant genetics, fertilizer, pesticides) (Lu etal.,2017) the systematic linear trend was removed
from data of agriculture yields of given crops. Three-year rolling mean was then applied on these detrended data
to smooth out the year-to-year variability. Finally, the percentage difference of 3year rolling mean from the base
linear trend was calculated to express crop yield deviation of a given 3year period for each country.
2.3. Mesoscale Hydrologic Model
The mHM (Kumar etal., 2013; Samaniego et al., 2010) is a spatially explicit, grid-based hydrologic model
developed at the UFZ-Helmholtz Centre for Environmental Research(Samaniego, Kaluza, etal.,2019) aiming
at providing seamless prediction of hydrological fluxes and storages at multiple spatial resolutions and locations
across the globe and it has reached Technology readiness levels of 9. The model uses the grid cell as a primary
hydrologic unit, and accounts for the following hydrological processes: canopy interception, snow accumula-
tion and melt, soil moisture and evapotranspiration, surface and subsurface runoff generations, deep percolation
and baseflow, and flood routing along with a river network. mHM uses a novel Multiscale Parameterization
Regionalization technique which includes the regionalization and spatial scaling approaches to generate a set of
regionalized model parameter fields at required modeling resolutions, while explicitly accounting for the sub-grid
variability of the fine-scale information on terrain, soil, vegetation, and other landscape properties (Samaniego
etal.,2017).
The calibration of the mHM parameters was conducted and demonstrated earlier (Samaniego, Thober, etal.,2019),
and was based on the multi-basin optimizations performed on a wide range of hydrologic regimes. The 48 transfer
function parameters used in mHM were estimated simultaneously at nine (N=9) geographically and hydro-cli-
matically diverse basins across the European domain. This procedure was repeated 30 times. Thus, 30 different
parameter sets were obtained tailor-made to different groups of nine “donor” basins. Each optimization run was
carried out with the Dynamically Dimensioned Search algorithm (Tolson & Shoemaker,2007) using 1,500 iter-
ations. The Kling-Gupta Efficiency (KGE) (Gupta etal.,2009) obtained between simulated and observed daily
streamflow for each selected basin was used as an objective function. Each of the 30 optimized parameter sets was
then cross-validated in 1,266 European river basins having at least 5years of streamflow records. The parameter
set that exhibited the best performance using the median daily KGE in the cross-validation test was used for the
final runs (KGE=0.55) for obtaining seamless soil moisture predictions across Europe and is depicted by a black
line in Figure S1 in Supporting InformationS1.
2.4. Drought Analysis
Understanding of continental-wide drought propagation and its changes over time require methods which quan-
tify drought areal extent and duration together with drought severity in a combined manner. The spatio-temporal
drought cluster analysis (Andreadis etal.,2005; Diaz etal.,2020; Lloyd-Hughes,2012; Samaniego etal.,2013;
Zhou etal.,2019; Zink etal.,2016) allow to track events in space and time and thus to quantify their character-
istics across domain and over the entire period. To quantify agricultural droughts we use the deficit of the mHM
simulated soil moisture with respect to its seasonal climatology at a given grid cell. The index is called (SMI)
(Sheffield etal.,2004) and its implementation has been described in the past (Samaniego etal.,2013) as follows.
The SMI varies between two limits 0 and 1. The lower bound indicates the driest condition with respect to the
reference period, whereas the upper bound indicates the wettest. The SMIi,m is estimated such that it represents
the conditional cumulative distribution function of the soil water content in the root zone at a given grid cell i
at the calendar month m. Given a time series x1, x2, …, xn that corresponds to the soil moisture fractions of a

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
4 of 11
given cell at calendar month m (e.g., January), the kernel density estimate at a given value x can be obtained
by



()= 1
𝑛
∑

=1 
(
−
𝑛
)
, wit h K(x) denoting a Gaussian smoothing kernel, n the sampling size, and h the
bandwidth. The SMI for a soil moisture fraction value x is estimated with the quantile function by numerically
integrating the following expression
SMI = ∫
0
()

.
To estimate spatio-temporal drought clusters we follow this procedure (Samaniego etal.,2013). First, we select
regions under drought by masking the SMI fields: SMIt<τ, with τ=0.2, according to Andreadis etal. (2005)
and Vidal etal.(2010). Second, consolidate spatial drought clusters at every time step. Clusters with an area of
less than η×Ac are neglected, with Ac the area of a grid cell. In this study η≈10 and η×Ac>25,000km
2 was
selected. This subjective choice is necessary to eliminate small isolated areas that are suffering a drought but
are too small to be considered as a regional event, and it does not have an impact on large-scale cluster charac-
teristics. On the final step, independent spatial drought clusters over successive time steps are consolidated into
regional, multi-temporal clusters. The only condition to join spatial clusters over consecutive time steps is that
their overlapping area is larger than ι×Ac, with ι≈60. It represents the connectivity between the clusters of two
consecutive time periods. Events having an intersection area at any sequential time step of less than this threshold
are considered independent events. The SMI code that executes this algorithm is available at https://git.ufz.de/
chs/progs/SMI (Samaniego etal.,2022). The drought severity (Sd) for a grid cell over a duration d (in months) is
estimated by integrating the masked SMI fields over time:
𝑆
𝑑
= 1− 1
𝑑∑𝑡∈𝑑SMI𝑡
. This indicator aims to measure
the duration and intensity of drought event. It ranges from zero to one, with one denoting locations with the
strongest impact during interval d. The total magnitude of a drought event is defined as the spatio-temporal inte-
gral of the SMI under the deficit threshold τ, explicitly:

=∑

1
=
0∫


(− SMI())+

. Here, t0 and t1 denote
the onset and the ending months of a drought event. At is the area under drought at time point t expressed as the
percentage of total drain area (in this case Europe); and (⋅)+ is the positive part function. Hence, M is expressed
in [months×%Total Area]. The intensity of a drought event (Id) is obtained by normalizing M by the arbitrary
drought duration d [months] from the onset:

=
1

∑

0
+
=
0
∫

(− SMI())+

expressed in [%Total Area]. Note
that drought intensity changes over time from the drought onset, while the drought magnitude is here estimated
for the entire cluster.
The values of M and Id are accompanied by “mean drought characteristics” of an identified drought cluster/event
in terms of “mean drought area over time” as a fraction of European domain [in %] and “mean duration over
space” of identified drought cluster [in months]. Here, consider that a drought cluster dynamically evolves over
space and time. First, it gets initiated, then it evolves/increases, and finally, it gets terminated. For example, the
drought cluster/event of 2018–2020 was initiated in April 2018 and lasted until December 2020. Although it took
33months from t0 to t1, its “mean duration,” averaged over entire cluster domain, yields 12.2months.
Historical and future extreme soil moisture drought events are estimated by forcing mHM with the respective
meteorological outputs of the GCMs (Section2.1.2). To maintain the consistency with the historical drought
analysis, the GCM specific soil moisture distribution functions estimated during the period from 1950 to 2020
were taken as a reference for estimating the soil moisture droughts in future. The drought characteristics are
calculated for each model realization separately, to avoid a smoothing effect of averaging, in other words, to be
able to generate extremes (rare events) realistically.
3. Results and Discussion
The spatio-temporal analysis of the largest European soil moisture droughts since 1766, summarized in Figure1a,
reveals three outstanding and well-comparable soil moisture drought events in terms of areal coverage, duration
from the onset, and drought intensity: 1857–1860, 1920–1922, and 2018–2020. The 2018–2020 event exhibits
the largest mean drought area over time, covering approximately on average 36% of Europe with a mean dura-
tion over space of 12.2months. Note that these characteristics reflect the average behavior accounting for the
spatio-temporal development of the drought events (see Section2.4). Concerning duration, the 2018–2020 event
ranks second after the event of 1857–1860. Additionally, we highlight other four well-documented and signif-
icant European drought events (Masante etal., 2019): 1947–1948, 1975–1977, 2003–2004, and 2015–2016.
Figure S2 in Supporting InformationS1 shows the negligible effect of the varying cluster parameter ι on the soil
moisture cluster identification. The 2018–2020 drought event is identified as the largest across a range of ι=40,

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
5 of 11
50, 60, 70, 80 values, which yield areal coverage range between 35.1% and 36.5% and average duration range of
12.0–12.5months.
In this context, what makes the 2018–2020 event outstanding with respect to the other events is the strong tempo-
ral development of its mean intensity since the onset of the drought event (Figure1b). We refer to intensity as the
total drought magnitude normalized by its duration (see Section2.4). The 2018–2020 event has the steepest and
continuous rise in intensity reaching a historical maximum after only 10months from the onset, while reaching
80% of their maximum intensity within 4months only. Furthermore, we notice that the four exceptional summer
droughts with the steepest rise in intensity (1947–1948, 2015–2016, 2003–2004, 2015–2016, and 2018–2020)
have been always initiated in spring (April–May) primarily as a result of compound effects of low precipitation
and high air temperatures leading to severe soil water deficits (Ionita et al.,2020). After the highest intensity
is reached, it usually takes between 6 and 12months until a drought event terminates (Figure1b). This point is
reached in most cases during the following winter and spring seasons due to the significant contribution of the
snowmelt. For example, the 2003 and 2015 warm-season events, which quickly built their peak intensity during
late spring and summer, and slowly vanished in the following year leading to a recovery of the vegetation health
status (Hari etal.,2020). There has been generally a continuous rise in the intensity and its peak since the begin-
ning of the 21st century (Figure1b, progression in peak intensity from 2003 to 2004, 2015–2016, 2018–2020).
Figure 1. Main characteristics of the three largest soil moisture drought clusters identified in Europe since 1766. (a) Scatter plot of the mean area and duration of
European droughts from 1766 to 2020 based on the mesoscale Hydrologic Model simulations forced with reconstructed and observation-based historical forcing data.
The cluster labels define period (start year–end year) of the well-known drought events. The bubble size corresponds to the total drought magnitude [-] (b) Temporal
evolution of the drought intensity from the onset of the drought. The 2018–2020 event exhibits the largest drought intensity in comparison to all other events overall
time (c–e) Spatial map depicting the mean drought duration [months] of the 1857–1860, 1920–1922, and 2018–2020 events. The inset plot shows the temporal
evolution of the areal coverage from the onset of the corresponding event.

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
6 of 11
The spatial distribution of the mean drought duration for the three largest events (Figures1c–1e) suggest that the
2018–2020 event “extended” across entire Europe, which has so-far never happened for any other major events
in the past. Our simulations reveal that 20% (40%) of the European domain was under drought for more than 18
(12) months during 2018–2020. The inset of Figure1e depicts the temporal evolution of the area under drought,
which exhibits a lot of variability reaching a maximum of 50% of Europe's area was affected by a drought. The
2018–2020 event, exhibited several peaks in the evolution of drought coverage (snapshots of four states are
provided in Figure S3 in Supporting InformationS1: The centroid of this drought event is located in the central
and eastern Europe, although during 2018 the event covered parts of Scandinavia). In 2019, it expanded to the
Mediterranean region, which corresponds well with the entry of the European Drought Observatory database
(Masante etal.,2019).
Büntgen etal. (2021) showed that the recent sequences of European seasonal droughts are unprecedented at a
millennial time scale, suggesting that amplification in anthropogenic warming may have played a significant role
in exacerbating its evolution. To analyze the role of near-surface air temperature and precipitation conditions on
the soil moisture droughts, we quantify the mean precipitation and temperature anomalies over the area affected
by corresponding drought events, as depicted in Figure2. It reveals that the precipitation deficit during the
2018–2020 event is around 20% with respect to the long term mean and therefore comparable to previous major
drought events. This event, however, is exceptional considering the record-breaking high-temperature anomaly
that reached up to +2.8K with respect to the long term mean. This finding indicates the amplifying effect of
temperature on drought evolution as depicted by Chiang etal. (2018). We also noticed a rather localized event
over the Mediterranean region, where precipitation deficit is one of the largest in relative terms (up to 35%), but
that did not induce considerable changes in the soil moisture deficit with a large areal coverage and/or multi-year
Figure 2. Meteorological conditions during the soil moisture droughts from 1766 to the present. Average relative
precipitation anomaly and average absolute near-surface air temperature anomaly based on the monthly climatology during
the period 1766–2020. The bubble size corresponds to the total drought magnitude. The colored bubbles correspond to the
three major events depicted in Figure1a.

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
7 of 11
duration. Additionally, the corresponding temperature anomaly was close to zero. These investigations were
focused on providing a glimpse on the anomalous meteorological conditions associated with large-scale soil
moisture drought events occurred across Europe. However, a proper attribution of disentangling the explicit
control of different drivers including meteorological and other (land-surface feedback) conditions (Seneviratne
etal.,2010) require careful considerations, which is beyond the scope of current work.
The aforementioned analysis establishes the 2018–2020 drought as a record breaking event over the past 250years.
Considering the high impacts of events like 2018–2020, the dominant role of the rain-fed agriculture in Europe
(Trnka etal., 2019), and the negligible mitigation effect of irrigation systems on drought stress at continental
scales (especially in Central Europe), next we scrutinize how soil moisture droughts have affected cropland in
Europe during the last seven decades, during which consistent crop yield data are available (EUROSTAT,2021;
FAO Global Statistical Yearbook, 2021). We analyze the possible impact of soil moisture drought on the loss
in agriculture productivity across Europe. Figure3 shows a substantial drop in major crop yields (wheat, grain
maize, and barley) during the 2018–2020 drought event across most of the European countries. The 3-year average
crop yield anomaly for three dominant cereals is considered after removing the systematic linear trend accounting
for technological advances (e.g., improvements in plant genetics, fertilizer, pesticides). All three cereal products
exhibit sharp decline from the expected (linear trend) yields across western, central and northern Europe: losses
of up to 17.5% for wheat in Germany, 20%–40% for grain maize in western Europe (including Benelux, Germany,
and France) and around 10% losses for barley in most countries in Europe except for the Iberian Peninsula and
several south-eastern European countries. The extremity of the expected crop yield losses is further depicted in
Figure S4 in Supporting InformationS1, which shows the 3year exceedance probability of the averaged ranks
(Benard & Bos-Levenbach,1953).
Considering the extremity of the 2018–2020 event, it is imperative to understand how the characteristics of this
event compare with those of future alike events. To this end, we compare the areal extent, the duration, and the
total drought magnitude of this 2018–2020 event against those of potential events resulting from climate projec-
tions. Figures4a and4b shows that the 2018–2020 event ranks as one of the most extreme when compared against
GCM simulated events during the 1950–2020 period, in terms of both areal coverage and duration. It is worth
noting that Figures4a and4b integrates all three drought aspects: severity, duration and area from all five GCMs
at once. Ensemble averaging in this case due to the fact that under different climate forcing's data, droughts evolve
during different times; they do not happen at the exact location.
Figure 3. Crop yield anomaly. Average percent crop yield anomaly during 2018–2020 with respect to 1961–2021 period (after removal of linear trend representing
technological advances) for three major cereals: (a) wheat, (b) grain maize, and (c) barley during period 1961–2021 (EUROSTAT,2021; FAO Global Statistical
Yearbook,2021).

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
8 of 11
The spatio-temporal evolution of future soil moisture droughts during the period from 2021 to 2099 suggests
that the most extreme droughts are projected to be significantly longer (RCP4.5: up to 100months; RCP8.5:
up to 300 months) than that of the 2018–2020 drought event. This result is in line with the previous study
(Samaniego etal.,2018), and the longer duration of RCP8.5 is projected independently by all five GCMs (see
Figure S5 in Supporting InformationS1). While the moderate RCP4.5 emission scenario projects the most signif-
icant drought clusters to cover up to 50% of the entire domain, this areal extent reaches up to 65% based on the
high-emission scenarios. Additionally, the 2018–2020 event dominates all GCM-based events in terms of drought
intensity during the historical period, similar to what was seen with the observation-based events (see Figure
S6 in Supporting InformationS1). Only a few future RCP's realizations show greater intensity than that of the
2018–2020 event. Here, the climatological conditions defining the soil anomalies are based on the contemporary
conditions (1950–2020), assuming there is no adaption to persistent drying patterns considered. This allows the
drought to develop for a longer duration to develop into a multiple-year drought. Our analysis cannot confirm
higher occurrence frequencies for a 2018–2020-like event in the future. The reason for that is that our GCM
Figure 4. Main characteristics of large soil moisture drought events based on five Global Climate Models (GCMs). Bubble
plot depicting areal extent, duration and normalized magnitude (intensity) of the drought events in Europe during the
1950–2020 and 2021–2099 periods, based on the (a) representative concentration pathways (RCP 4.5) and (b) RCP 8.5,
respectively. The drought intensity is proportional to the circle size. The benchmarking 2018–2020 drought event, derived
from observation-based simulations from Figure1, is included for comparison. The boundary drawn by the dashed gray lines
denotes events whose areal extent or duration are greater than the 2018–2020 event. The sampling distributions of the mean
duration and mean areal drought extent across five GCMs corresponding to the RCPs 4.5 and 8.5 during 2021–2099, for
events larger than the benchmark historical 2018–2020 event, are depicted in panels (c) and (d), respectively.

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
9 of 11
simulations are limited until 2100. However, a future event can evolve for a long time given the drought threshold
set in our analysis.
Figures4c and4d depicts two individual drought characteristics in terms of the sampling distribution of the mean
duration and the mean areal extent of the GCM-derived drought events under both RCPs to illustrate their differ-
ences in projected future drought characteristics of all events greater than historical 2018–2020 benchmark. On
average, future drought events whose duration is greater than the 2018–2020 event are projected to last approx-
imately 40months (range: 25–60months) under the moderate emission scenario, while they will nearly double
their duration (range: 25–180months) under a high-emission scenario (Figure4c). This implies that future events
could last, on average, three to six times longer than the 2018–2020 event, respectively. With respect to the areal
extent, we observe a slight increase from 41% to 43% (Figure4d), between RCP 4.5 and 8.5, which corresponds
to an increase of 14%–20% with respect to the 2018–2020 event, respectively.
4. Summary and Conclusions
We synthesized long-term simulations showing the spatio-temporal evolution of soil moisture droughts in Europe
during the period from 1766 to 2020. Our analysis helped to better understand the development of multi-year
droughts by taking into account long-term historical changes in hydroclimatic variability. We concluded that the
recent 2018–2020 drought event is exceptional because of its significantly higher intensity and fast development
from its onset compared to past events. Our study highlights that the 2018–2020 drought event, compared to all
historical events, emerges as a new benchmark that can be used to gauge the potential impact of future drought
events in terms of socioeconomic and ecological damages in Europe. From our recent analysis, and in accordance
with previous study (Samaniego etal., 2018), we conclude that Europe should prepare adaptation and mitiga-
tion plans for future events whose intensity may be comparable to the previous event, but whose duration (and
partly their spatial extent) will be much greater than any event observed in the last 250years. Our analysis shows
that exceptional agricultural droughts enhanced by record-breaking near-surface air temperature anomalies have
significant impact (decline) on crop yields across the European countries. Soil moisture drought projections
synthesized in this study, even under a moderate emission scenario, indicate that decision-makers in Europe
should be prepared for drought events of comparable intensity in future. Thus, the 2018–2020 drought event
could be considered as a wake-up call on agricultural policies. In this study, we compared and contrasted this
event with earlier events of similar magnitudes and showed the role of increasing temperature rises. This study
has focused on detecting the soil moisture droughts and their driving meteorological conditions. Future studies
should aim at disentangling the roles of precipitation and temperature drivers, including climate model runs.
Finally, we emphasize the need for new technological developments to mitigate the effects of extreme droughts
and heatwaves and further research to understand how this new kind of fast intensified droughts will impact
human health, ecosystems, and, ultimately, our living conditions.
Data Availability Statement
Data analysis was conducted at the High-Performance Computing (HPC) Cluster EVE, a joint effort of both the
Helmholtz Centre for Environmental Research-UFZ and the German Centre for Integrative Biodiversity Research
(iDiv) Halle-Jena-Leipzig. Model simulations used in this study can be obtained from https://dx.doi.org/10.5281/
zenodo.5082089. The mHM (v5.10) code can be found at Samaniego, Kaluza, etal.(2019) and the SMI code can
be found under Samaniego etal.(2022). OR and RK conceptualized and designed the study with inputs from LS.
OR conducted the hydrologic simulations and analysis with inputs from LS, VH, YM, MH, and RK. LS and ST
provided the SMI code. VM processed the crop yield data. OR wrote the initial draft with inputs from LS, VH
and RK. All authors contributed to discussion and edited the manuscript.
References
Allen, R. G., Pereira, L. S., Raes, D., & Smith, M. (1998). Crop evapotranspiration-Guidelines for computing crop water requirements-FAO
Irrigation and drainage paper 56. Fao, Rome, 300(9), D05109.
Andreadis, K. M., Clark, E. A., Wood, A. W., Hamlet, A. F., & Lettenmaier, D. P. (2005). Twentieth-century drought in the conterminous United
States. Journal of Hydrometeorology, 6(6), 985–1001. https://doi.org/10.1175/jhm450.1
Benard, A., & Bos-Levenbach, E. (1953). The plotting of observations on probability paper. Statistica, 7, 163–173. https://doi.
org/10.1111/j.1467-9574.1953.tb00821.x
Acknowledgments
This work was carried out within the
bilateral project eXtreme EuRopean
drOughtS (multimodel synthesis of past,
present and future events), funded by the
Deutsche Forschungsgemeinschaft (grant
RA 3235/1-1) and Czech Science Foun-
dation (grant 19-24089J). The authors
would also like to thank the partial fund-
ing from the Helmholtz Climate Initiative
project. We acknowledge Song Feng (the
University of Nebraska–Lincoln) for
sharing and processing potential evap-
otranspiration from the Coupled Model
Intercomparison Project v5 archive. Open
access funding enabled and organized by
Projekt DEAL.

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
10 of 11
Büntgen, U., Urban, O., Krusic, P. J., Rybníček, M., Kolář, T., Kyncl, T., etal. (2021). Recent European drought extremes beyond Common Era
background variability. Nature Geoscience, 14(4), 190–196. https://doi.org/10.1038/s41561-021-00698-0
Casty, C., Raible, C. C., Stocker, T. F., Wanner, H., & Luterbacher, J. (2007). A European pattern climatology 1766–2000. Climate Dynamics,
29(7–8), 791–805. https://doi.org/10.1007/s00382-007-0257-6
Chiang, F., Mazdiyasni, O., & AghaKouchak, A. (2018). Amplified warming of droughts in southern United States in observations and model
simulations. Science Advances, 4(8), eaat2380. https://doi.org/10.1126/sciadv.aat2380
Ciais, P., Reichstein, M., Viovy, N., Granier, A., Ogée, J., Allard, V., etal. (2005). Europe-wide reduction in primary productivity caused by the
heat and drought in 2003. Nature, 437(7058), 529–533. https://doi.org/10.1038/nature03972
Diaz, V., Perez, G. A. C., Van Lanen, H. A. J., Solomatine, D., & Varouchakis, E. A. (2020). An approach to characterise spatio-temporal drought
dynamics. Advances in Water Resources, 137. https://doi.org/10.1016/j.advwatres.2020.103512
EUROSTAT.(2021). Agricultural production – crops Retrieved from https://ec.europa.eu/eurostat/en/web/main/data/database. Accessed: 1 June
2021.
FAO Global Statistical Yearbook. (2021). FAO Regional Statistical Yearbooks. Retrieved from http://www.fao.org/faostat/en/#data/QC. Accessed:
1 June 2021.
Fischer, E. M., Seneviratne, S. I., Vidale, P. L., Lüthi, D., & Schär, C. (2007). Soil moisture–atmosphere interactions during the 2003 European
summer heat wave. Journal of Climate, 20(20), 5081–5099. https://doi.org/10.1175/jcli4288.1
Flach, M., Sippel, S., Gans, F., Bastos, A., Brenning, A., Reichstein, M., & Mahecha, M. D. (2018). Contrasting biosphere responses to hydromete-
orological extremes: Revisiting the 2010 western Russian heatwave. Biogeosciences, 16, 6067–6085. https://doi.org/10.5194/bg-15-6067-2018
Frieler, K., Lange, S., Piontek, F., Reyer, C. P. O., Schewe, J., Warszawski, L., & Yamagata, Y. (2017). Assessing the impacts of 1.5 c global
warming—Simulation protocol of the inter-sectoral impact model intercomparison project (ISIMIP2b). Geoscientific Model Development,
10(12), 4321–4345. https://doi.org/10.5194/gmd-10-4321-2017
Gudmundsson, L., Bremnes, J. B., Haugen, J. E., & Engen-Skaugen, T. (2012). Downscaling RCM precipitation to the station scale using
statistical transformations–a comparison of methods. Hydrology and Earth System Sciences, 16(9), 3383–3390. https://doi.org/10.5194/
hess-16-3383-2012
Gupta, H. V., Kling, H., Yilmaz, K. K., & Martinez, G. F. (2009). Decomposition of the mean squared error and NSE performance criteria:
Implications for improving hydrological modelling. Journal of hydrology, 377(1–2), 80–91. https://doi.org/10.1016/j.jhydrol.2009.08.003
Hanel, M., Rakovec, O., Markonis, Y., Máca, P., Samaniego, L., Kyselý, J., & Kumar, R. (2018). Revisiting the recent European droughts from a
long-term perspective. Scientific Reports, 8(1), 9499. https://doi.org/10.1038/s41598-018-27464-4
Hari, V., Rakovec, O., Markonis, Y., Hanel, M., & Kumar, R. (2020). Increased future occurrences of the exceptional 2018–2019 Central Euro-
pean drought under global warming. Scientific Reports, 10(1), 1–10. https://doi.org/10.1038/s41598-020-68872-9
Hempel, S., Frieler, K., Warszawski, L., Schewe, J., & Piontek, F. (2013). A trend-preserving bias correction - The ISI-MIP approach. Earth
System Dynamics, 4(2), 219–236. https://doi.org/10.5194/esd-4-219-2013
Herrera-Estrada, J. E., Satoh, Y., & Sheffield, J. (2017). Spatiotemporal dynamics of global drought. Geophysical Research Letters, 44(5),
2254–2263.
Hofstra, N., Haylock, M., New, M., & Jones, P. D. (2009). Testing E-OBS European high-resolution gridded data set of daily precipitation and
surface temperature. Journal of Geophysical Research: Atmospheres, 114(D21). https://doi.org/10.1029/2009jd011799
Ionita, M., Nagavciuc, V., Kumar, R., & Rakovec, O. (2020). On the curious case of the recent decade, mid-spring precipitation deficit in central
Europe. NPJ Climate and Atmospheric Science, 3(49), 1–10. https://doi.org/10.1038/s41612-020-00153-8
Kumar, R., Samaniego, L., & Attinger, S. (2013). Implications of distributed hydrologic model parameterization on water fluxes at multiple scales
and locations. Water Resources Research, 49(1), 360–379. https://doi.org/10.1029/2012wr012195
Lloyd-Hughes, B. (2012). A spatio-temporal structure-based approach to drought characterisation. International Journal of Climatology, 32(3),
406–418. https://doi.org/10.1002/joc.2280
Lu, J., Carbone, G. J., & Gao, P. (2017). Detrending crop yield data for spatial visualization of drought impacts in the United States, 1895–2014.
Agricultural and Forest Meteorology, 237–238, 196–208. https://doi.org/10.1016/j.agrformet.2017.02.001
Masante, D., Barbosa, P., & Magni, D. (2019). European drought observatory. Retrieved from https://edo.jrc.ec.europa.eu/documents/news/
EDODroughtNews201908_Europe.pdf. Accessed: 5 May 2021.
Moravec, V., Markonis, Y., Rakovec, O., Kumar, R., & Hanel, M. (2019). A 250-year European drought inventory derived from ensemble hydro-
logic modeling. Geophysical Research Letters, 46(11), 5909–5917. https://doi.org/10.1029/2019GL082783
Moravec, V., Markonis, Y., Rakovec, O., Svoboda, M., Trnka, M., Kumar, R., & Hanel, M. (2021). Europe under multi-year droughts: How severe
was the 2014–2018 drought period? Environmental Research Letters, 16(3), 034062. https://doi.org/10.1088/1748-9326/abe828
Naumann, G., Cammalleri, C., Mentaschi, L., & Feyen, L. (2021). Increased economic drought impacts in Europe with anthropogenic warming.
Nature Climate Change, 11, 485–491. https://doi.org/10.1038/s41558-021-01044-3
Oudin, L., Hervieu, F., Michel, C., Perrin, C., Andréassian, V., Anctil, F., & Loumagne, C. (2005). Which potential evapotranspiration input for
a lumped rainfall–runoff model?: Part 2—towards a simple and efficient potential evapotranspiration model for rainfall–runoff modelling.
Journal of Hydrology, 303(1), 290–306. https://doi.org/10.1016/j.jhydrol.2004.08.026
Peichl, M., Thober, S., Meyer, V., & Samaniego, L. (2018). The effect of soil moisture anomalies on maize yield in Germany. Natural Hazards
and Earth System Sciences, 18(3), 889–906. https://doi.org/10.5194/nhess-18-889-2018
Peters, W., Bastos, A., Ciais, P., & Ver meulen, A. (2020). A historical, geographical and ecological perspective on the 2018 European summer
drought. Philosophical transactions of the royal society B, 375(1810), 20190505. https://doi.org/10.1098/rstb.2019.0505
Samaniego, L., Kaluza, M., Kumar, R., Rakovec, O., Schüler, L., Schweppe, R., etal. (2019). Mesoscale Hydrologic Model (v5.10). [Computer
software]. Zenodo. https://doi.org/10.5281/ZENODO.3239055
Samaniego, L., Kumar, R., & Attinger, S. (2010). Multiscale parameter regionalization of a grid-based hydrologic model at the mesoscale. Water
Resources Research, 46(W05523). https://doi.org/10.1029/2008wr007327
Samaniego, L., Kumar, R., Thober, S., Rakovec, O., Zink, M., Wanders, N., & Attinger, S. (2017). Toward seamless hydrologic predictions across
spatial scales. Hydrology and Earth System Sciences, 21(9), 4323–4346. https://doi.org/10.5194/hess-21-4323-2017
Samaniego, L., Kumar, R., & Zink, M. (2013). Implications of parameter uncertainty on soil moisture drought analysis in Germany. Journal of
Hydrometeorology, 14(1), 47–68. https://doi.org/10.1175/jhm-d-12-075.1
Samaniego, L., Kumar, R., Zink, M., Mai, J., Boeing, F., Shrestha, P.-K., etal. (2022). The Soil Moisture Index - SMI program (Version 2.0.5)
[Computer software]. Zenodo. https://doi.org/10.5281/ZENODO.5842486
Samaniego, L., Thober, S., Kumar, R., Wanders, N., Rakovec, O., Pan, M., & Marx, A. (2018). Anthropogenic warming exacerbates European
soil moisture droughts. Nature Climate Change, 8(5), 421. https://doi.org/10.1038/s41558-018-0138-5

Earth’s Future
RAKOVEC ET AL.
10.1029/2021EF002394
11 of 11
Samaniego, L., Thober, S., Wanders, N., Pan, M., Rakovec, O., Sheffield, J., & Kumar, R. (2019). Hydrological forecasts and projections for
improved decision-making in the water sector in Europe. Bulletin of the American Meteorological Society, 100(12), 2451–2471. https://doi.
org/10.1175/bams-d-17-0274.1
Scheff, J., & Frierson, D. M. (2014). Scaling potential evapotranspiration with greenhouse warming. Journal of Climate, 27(4), 1539–1558.
https://doi.org/10.1175/jcli-d-13-00233.1
Seneviratne, S. I., Corti, T., Davin, E. L., Hirschi, M., Jaeger, E. B., Lehner, I., & Teuling, A. J. (2010). Investigating soil moisture–climate inter-
actions in a changing climate: A review. Earth-Science Reviews, 99(3–4), 125–161. https://doi.org/10.1016/j.earscirev.2010.02.004
Sheffield, J., Goteti, G., Wen, F., & Wood, E. F. (2004). A simulated soil moisture based drought analysis for the United States. Journal of
Geophysical Research: Atmospheres, 109(D24). https://doi.org/10.1029/2004jd005182
Stanke, C., Kerac, M., Prudhomme, C., Medlock, J., & Murray, V. (2013). Health effects of drought: A systematic review of the evidence. PLoS
Currents, 5. https://doi.org/10.1371/currents.dis.7a2cee9e980f91ad7697b570bcc4b004
Tolson, B. A., & Shoemaker, C. A. (2007). Dynamically dimensioned search algorithm for computationally efficient watershed model calibration.
Water Resources Research, 43(1). https://doi.org/10.1029/2005WR004723
Trnka, M., Feng, S., Semenov, M. A., Olesen, J. E., Kersebaum, K. C., Rötter, R. P., etal. (2019). Mitigation efforts will not fully alleviate the
increase in water scarcity occurrence probability in wheat-producing areas. Science Advances, 5(9), eaau2406. https://doi.org/10.1126/sciadv.
aau2406
Van Lanen, H. A., Laaha, G., Kingston, D. G., Gauster, T., Ionita, M., Vidal, J.-P., etal. (2016). Hydrology needed to manage droughts: The 2015
European case. Hydrological Processes, 30(17), 3097–3104. https://doi.org/10.1002/hyp.10838
Vidal, J.-P., Martin, E., Franchistéguy, L., Habets, F., Soubeyroux, J.-M., Blanchard, M., & Baillon, M. (2010). Multilevel and multiscale drought
reanalysis over France with the Safran-Isba-Modcou hydrometeorological suite. Hydrology and Earth System Sciences, 14(3), 459–478.
https://doi.org/10.5194/hess-14-459-2010
Vogel, M. M., Zscheischler, J., Wartenburger, R., Dee, D., & Seneviratne, S. I. (2019). Concurrent 2018 hot extremes across Northern Hemisphere
due to human-induced climate change. Earth's Future, 7(7), 692–703. https://doi.org/10.1029/2019ef001189
Warszawski, L., Frieler, K., Huber, V., Piontek, F., Serdeczny, O., & Schewe, J. (2014). The inter-sectoral impact model intercomparison
project (ISI–MIP): Project framework. Proceedings of the National Academy of Sciences, 111(9), 3228–3232. https://doi.org/10.1073/
pnas.1312330110
Williams, A. P., Cook, E. R., Smerdon, J. E., Cook, B. I., Abatzoglou, J. T., Bolles, K., & Livneh, B. (2020). Large contribution from anthropo-
genic warming to an emerging North American megadrought. Science, 368(6488), 314–318. https://doi.org/10.1126/science.aaz9600
Zhou, H., Liu, Y., & Liu, Y. (2019). An approach to tracking meteorological drought migration. Water Resources Research, 55(4), 3266–3284.
https://doi.org/10.1029/2018wr023311
Zink, M., Samaniego, L., Kumar, R., Thober, S., Mai, J., Schäfer, D., & Marx, A. (2016). The German drought monitor. Environmental Research
Letters, 11(7), 074002. https://doi.org/10.1088/1748-9326/11/7/074002