Geophysical Research Letters,
Год журнала:
2025,
Номер
52(6)
Опубликована: Март 23, 2025
Abstract
Extreme
climate
events
(ECEs)
like
heavy
rainfall
and
heatwaves
significantly
impact
society,
change
is
altering
their
magnitude
frequency.
Generalized
Value
(GEV)
distributions
help
quantify
these
ECEs
guide
human
system
design.
We
train
a
machine
learning
(ML)
model
using
set
of
arbitrary
GEV
to
estimate
the
sample
size
required
determine
return
value
with
specific
uncertainty.
For
negative
shape
parameter
maximum
extreme
temperatures
are
bounded
fewer
samples
needed
given
uncertainty
than
extremes
which
have
positive
unbounded
values.
example,
if
1‐in‐20‐year
heatwave
event
requires
400
1%
uncertainty,
one
would
need
20
different
20‐year
simulations.
Achieving
such
quantities
will
require
extensive
downscaling
simulations,
potentially
aided
by
ML‐based
methods
increase
ensemble
size.
Meteorological Applications,
Год журнала:
2022,
Номер
29(3)
Опубликована: Май 1, 2022
Abstract
Low‐likelihood
weather
events
can
cause
dramatic
impacts,
especially
when
they
are
unprecedented.
In
2020,
amongst
other
high‐impact
events,
UK
floods
caused
more
than
£300
million
damage,
prolonged
heat
over
Siberia
led
to
infrastructure
failure
and
permafrost
thawing,
while
wildfires
ravaged
California.
Such
rare
phenomena
cannot
be
studied
well
from
historical
records
or
reanalysis
data.
One
way
improve
our
awareness
is
exploit
ensemble
prediction
systems,
which
represent
large
samples
of
simulated
events.
This
‘UNSEEN’
method
has
been
successfully
applied
in
several
scientific
studies,
but
uptake
hindered
by
data
processing
requirements,
uncertainty
regarding
the
credibility
simulations.
Here,
we
provide
a
protocol
apply
ensure
UNSEEN
for
studying
low‐likelihood
globally,
including
an
open
workflow
based
on
Copernicus
Climate
Change
Services
(C3S)
seasonal
predictions.
Demonstrating
using
European
Centre
Medium‐Range
Weather
Forecasts
(ECMWF)
SEAS5,
find
that
2020
March–May
Siberian
heatwave
was
predicted
one
members;
record‐shattering
August
California‐Mexico
temperatures
were
part
strong
increasing
trend.
However,
each
case
studies
exposes
challenges
with
respect
sensitivity
outcomes
user
decisions.
We
conclude
new
insights
about
decisions
transparent,
sensitivities
acknowledged.
Anticipating
plausible
extreme
uncovering
unforeseen
hazards
under
changing
climate
warrants
further
research
at
science‐policy
interface
manage
high
impacts.
Environmental Research Letters,
Год журнала:
2022,
Номер
18(1), С. 015004 - 015004
Опубликована: Дек. 14, 2022
Abstract
Elucidating
the
statistical
properties
of
extreme
meteo-climatic
events
and
capturing
physical
processes
responsible
for
their
occurrence
are
key
steps
improving
our
understanding
climate
variability
change
better
evaluating
associated
hazards.
It
has
recently
become
apparent
that
large
deviation
theory
(LDT)
is
very
useful
investigating
persistent
events,
specifically,
flexibly
estimating
long
return
periods
introducing
a
notion
dynamical
typicality.
Using
methodological
framework
based
on
LDT
taking
advantage
simulations
by
state-of-the-art
Earth
system
model,
we
investigate
2021
Western
North
America
summer
heatwave.
Indeed,
analysis
shows
event
can
be
seen
as
an
unlikely
but
possible
manifestation
variability,
whilst
its
probability
greatly
amplified
ongoing
change.
We
also
clarify
spatial
coherence
heatwave
elucidate
role
played
Rocky
Mountains
in
modulating
hot,
dry,
Pacific
region
America.
Abstract
Compound
climate
extremes
(here
referred
to
compound
dry–hot
events
and
pluvial–hot
events)
result
in
devastating
disasters
which
threaten
water‐food‐energy
security.
However,
a
warming
scenario,
the
risk
of
occurrence,
quantification
uncertainty,
associated
drivers
extremes—particularly
events—have
not
been
fully
explored.
By
leveraging
model
large
ensembles,
it
is
revealed
that
projected
increase
2–3
times
over
most
global
land
masses
future
Representative
Concentration
Pathway
(RCP)
8.5
forcing
compared
with
historical
forcing.
Increased
risks
are
mainly
attributed
changes
temperature
dependence
between
precipitation
temperature,
while
change
contributing
these
two
exhibits
approximately
spatial
complementary.
In
world,
hot
spots
lie
Europe,
South
Africa,
Amazon,
those
mostly
eastern
USA,
southern
Asia,
Australia,
central
Africa.
These
findings
help
stakeholders
decision
makers
develop
package
adaptation
strategies
manage
mitigate
extremes.
The
observed
increase
in
extreme
weather
has
prompted
recent
methodological
advances
event
attribution.
We
propose
a
machine
learning–based
approach
that
uses
convolutional
neural
networks
to
create
dynamically
consistent
counterfactual
versions
of
historical
events
under
different
levels
global
mean
temperature
(GMT).
apply
this
technique
one
heat
(southcentral
North
America
2023)
and
several
have
been
previously
analyzed
using
established
attribution
methods.
estimate
temperatures
during
the
southcentral
were
1.18°
1.42°C
warmer
because
warming
similar
will
occur
0.14
0.60
times
per
year
at
2.0°C
above
preindustrial
GMT.
Additionally,
we
find
learned
relationships
between
daily
GMT
are
influenced
by
seasonality
forced
response
meteorological
conditions.
Our
results
broadly
agree
with
other
techniques,
suggesting
learning
can
be
used
perform
rapid,
low-cost
events.
Weather and Climate Dynamics,
Год журнала:
2024,
Номер
5(3), С. 943 - 957
Опубликована: Июль 24, 2024
Abstract.
Central
European
winters
have
warmed
markedly
since
the
mid-20th
century.
Yet
cold
are
still
associated
with
severe
societal
impacts
on
energy
systems,
infrastructure,
and
public
health.
It
is
therefore
crucial
to
anticipate
storylines
of
worst-case
winter
conditions
understand
whether
an
extremely
winter,
such
as
coldest
historical
record
Germany
in
1963
(−6.3
°C
or
−3.4σ
seasonal
December–January–February
(DJF)
temperature
anomaly
relative
1981–2010),
possible
a
warming
climate.
Here,
we
first
show
based
multiple
attribution
methods
that
similar
circulation
would
lead
extreme
about
−4.9
−4.7
(best
estimates
across
methods)
under
present-day
This
rank
second-coldest
last
75
years.
Second,
conceive
two
independent
rare
event
sampling
(climate
model
boosting
empirical
importance
sampling):
physically
central
Europe
today,
albeit
very
unlikely.
While
hazards
become
less
frequent
intense
climate
overall,
it
remains
possibility
avoid
potential
maladaptation
increased
vulnerability.
Weather and Climate Dynamics,
Год журнала:
2025,
Номер
6(1), С. 1 - 15
Опубликована: Янв. 7, 2025
Abstract.
Extreme
cold
winter
temperatures
in
Europe
have
huge
societal
impacts.
Being
able
to
simulate
worst-case
scenarios
for
such
events
present
and
future
climates
is
hence
crucial
short-
long-term
adaptation.
In
this
paper,
we
are
interested
persistent
events,
whose
probability
will
decrease
with
climate
change.
Large
ensembles
of
simulations
allow
us
better
analyse
the
mechanisms
characteristics
but
can
require
significant
computational
resources.
Rather
than
simulating
very
large
normal
trajectories,
rare-event
algorithms
sampling
tail
distributions
more
efficiently.
Such
been
applied
extreme
heat
waves.
They
emphasized
role
atmospheric
circulation
extremes.
The
goal
study
evaluate
dynamics
spells
simulated
by
a
algorithm.
We
focus
on
that
occurred
France
from
1950
2021.
investigate
mean
(December,
January
February)
identify
record-shattering
event
1963.
find
although
frequency
decreases
time,
their
intensity
stationary.
apply
stochastic
weather
generator
(SWG)
approach
importance
coldest
winters
could
occur
factual
counterfactual
climate.
thus
worst
consistent
reanalysis
data.
few
reach
colder
historical
This
shows
present-day
conditions
trigger
as
record
spite
global
warming.
prevails
during
those
analysed
compared
observed
record-breaking
showing
no
main
change
leading
type
event.
Environmental Research Letters,
Год журнала:
2023,
Номер
18(9), С. 094061 - 094061
Опубликована: Сен. 1, 2023
Abstract
Human
bodies,
ecosystems
and
infrastructures
display
a
non-linear
sensibility
to
extreme
temperatures
occurring
during
heatwave
events.
Preparing
for
such
events
entails
know
how
high
surface
air
can
go.
Here
we
examine
the
maximal
reachable
in
Western
Europe.
Taking
July
2019
record-breaking
as
case
study
employing
flow
analogues
methodology,
find
that
exceeding
50
∘
C
cannot
be
ruled
out
most
urban
areas,
even
under
current
climate
conditions.
We
analyze
changes
upper
bound
of
between
past
(1940–1980)
present
(1981–2021)
periods.
Our
results
show
significant
increase
daily
maximum
period
is
only
partially
explained
by
bound.
suggest
warming
result
from
strengthened
diabatic
fluxes
rather
than
free
troposphere
warming.
Climate Services,
Год журнала:
2023,
Номер
30, С. 100363 - 100363
Опубликована: Фев. 10, 2023
Climate
is
arguably
one
of
the
most
important
factors
determining
quality
wine
from
any
given
grapevine
variety.
This
study
focuses
on
three
wine-growing
regions
in
northern
Portugal:
Vinho
Verde,
Trás-os-Montes
and
Douro,
latter
coinciding
with
Porto.
High
rainfall
during
late
spring
(April
to
June)
can
promote
growth
vines
but
increases
risk
fungal
disease.
harvest
time
(August
October)
also
bears
potential
for
severe
operational
disruption
heavy
economic
losses.
The
probability
unprecedented
totals
season
over
Portugal
has
been
assessed.
A
large
ensemble
initialised
climate
model
simulations
analysed,
each
quantified.
Seasonal
considerably
higher
than
observed
are
possible
current
climate.
An
event
either
could
occur
a
between
0.01
0.05
present
Extreme
value
analysis
was
applied
observations
ensemble,
return
periods
known
extreme
events
calculated.
Similar
probabilities
were
year
similar
1993,
when
both
seasons
exceptionally
wet,
would
be
expected
occur,
average,
just
once
next
70–80
years
These
results
inform
requirements
improved
vineyard
management
resilience,
such
as
design
drainage
channels,
access
roads
terraces.
Physical review. E,
Год журнала:
2024,
Номер
110(4)
Опубликована: Окт. 4, 2024
Statistical
physics
and
dynamical
systems
theory
are
key
tools
to
study
high-impact
geophysical
events
such
as
temperature
extremes,
cyclones,
thunderstorms,
geomagnetic
storms,
many
others.
Despite
the
intrinsic
differences
between
these
events,
they
all
originate
temporary
deviations
from
typical
trajectories
of
a
system,
resulting
in
well-organized,
coherent
structures
at
characteristic
spatial
temporal
scales.
While
statistical
extreme
value
analysis
techniques
capable
providing
return
times
probabilities
occurrence
certain
not
apt
account
for
their
underlying
physics.
Their
focus
is
compute
probability
that
large
or
small
with
respect
some
specific
observable
(e.g.,
temperature,
precipitation,
solar
wind),
rather
than
relate
rare
phenomena
anomalous
regimes.
This
paper
outlines
this
knowledge
gap,
presenting
related
challenges,
new
formalisms
briefly
commenting
on
how
stochastic
approaches
tailored
can
help
advance
understanding.
The
field
of
extreme
event
attribution
(EEA)
has
rapidly
developed
over
the
last
two
decades.
Various
methods
have
been
and
implemented,
physical
modelling
capabilities
generally
improved,
impact
emerged,
assessments
serve
as
a
popular
communication
tool
for
conveying
how
climate
change
is
influencing
weather
events
in
lived
experience.
However,
number
non-trivial
challenges
still
remain
that
must
be
addressed
by
community
to
secure
further
advancement
whilst
ensuring
scientific
rigour
appropriate
use
findings
stakeholders
associated
applications.
As
part
concept
series
commissioned
World
Climate
Research
Programme,
this
article
discusses
contemporary
developments
six
key
domains
relevant
EEA,
provides
recommendations
where
focus
EEA
should
concentrated
coming
decade.
These
are:
(1)
observations
context
EEA;
(2)
definitions;
(3)
statistical
methods;
(4)
(5)
attribution;
(6)
communication.
Broadly,
call
increased
capacity
building,
particularly
more
vulnerable
regions;
guidelines
assessing
suitability
models;
establishing
best-practice
methodologies
on
compound
record-shattering
extremes;
co-ordinated
interdisciplinary
engagement
develop
scaffolding
their
broader
applications;
ongoing
investment
To
address
these
requires
significant
multiple
fields
either
underpin
(e.g.,
monitoring;
modelling)
or
are
closely
related
events;
impacts)
well
working
consistently
with
experts
outside
science
generally.
if
approached
investment,
dedication,
coordination,
tackling
next
decade
will
ensure
robust
analysis,
tangible
benefits
global
community.