Remote Sensing,
Год журнала:
2024,
Номер
16(22), С. 4339 - 4339
Опубликована: Ноя. 20, 2024
Wildfires
increasingly
threaten
ecosystems
and
infrastructure,
making
accurate
burn
severity
mapping
(BSM)
essential
for
effective
disaster
response
environmental
management.
Machine
learning
(ML)
models
utilizing
satellite-derived
vegetation
indices
are
crucial
assessing
wildfire
damage;
however,
incorporating
many
can
lead
to
multicollinearity,
reducing
classification
accuracy.
While
principal
component
analysis
(PCA)
is
commonly
used
address
this
issue,
its
effectiveness
relative
other
feature
extraction
(FE)
methods
in
BSM
remains
underexplored.
This
study
aims
enhance
ML
classifier
accuracy
by
evaluating
various
FE
techniques
that
mitigate
multicollinearity
among
indices.
Using
composite
index
(CBI)
data
from
the
2014
Carlton
Complex
fire
United
States
as
a
case
study,
we
extracted
118
seven
Landsat-8
spectral
bands.
We
applied
compared
13
different
techniques—including
linear
nonlinear
such
PCA,
t-distributed
stochastic
neighbor
embedding
(t-SNE),
discriminant
(LDA),
Isomap,
uniform
manifold
approximation
projection
(UMAP),
factor
(FA),
independent
(ICA),
multidimensional
scaling
(MDS),
truncated
singular
value
decomposition
(TSVD),
non-negative
matrix
factorization
(NMF),
locally
(LLE),
(SE),
neighborhood
components
(NCA).
The
performance
of
these
was
benchmarked
against
six
classifiers
determine
their
improving
Our
results
show
alternative
outperform
computational
efficiency.
Techniques
like
LDA
NCA
effectively
capture
relationships
critical
BSM.
contributes
existing
literature
providing
comprehensive
comparison
methods,
highlighting
potential
benefits
underutilized
As
wildfires
become
more
frequent
and
severe
in
the
face
of
global
environmental
change,
it
becomes
crucial
not
only
to
assess,
prevent,
suppress
them
but
also
manage
aftermath
effectively.
Given
temporal
interconnections
between
these
issues,
we
explored
concept
“wildfire
science
loop”—a
framework
categorizing
wildfire
research
into
three
stages:
“before”,
“during”,
“after”
wildfires.
Based
on
this
partition,
performed
a
systematic
review
by
linking
particular
topics
keywords
each
stage,
aiming
describe
one
quantify
volume
published
research.
The
results
from
our
identified
substantial
imbalance
landscape,
with
post-fire
stage
being
markedly
underrepresented.
Research
focusing
is
1.5
times
(or
46%)
less
prevalent
than
that
“before”
1.8
77%)
“during”
stage.
This
discrepancy
likely
driven
historical
emphasis
prevention
suppression
due
immediate
societal
needs.
Aiming
address
overcome
imbalance,
present
perspectives
regarding
strategic
agenda
enhance
understanding
processes
outcomes,
emphasizing
socioecological
impacts
management
recovery
multi-level
transdisciplinary
approach.
These
proposals
advocate
integrating
knowledge-driven
burn
severity
ecosystem
mitigation/recovery
practical,
application-driven
strategies
policy
development.
supports
comprehensive
spans
short-term
emergency
responses
long-term
adaptive
management,
ensuring
landscapes
are
better
understood,
managed,
restored.
We
emphasize
critical
importance
“after-fire”
breaking
negative
planning
cycles,
enhancing
practices,
implementing
nature-based
solutions
vision
“building
back
better”.
Strengthening
balanced
focused
will
ability
close
loop
involved
improve
alignment
international
agendas
such
as
UN’s
Decade
Ecosystem
Restoration
EU’s
Nature
Law.
By
addressing
can
significantly
restore
ecosystems,
resilience,
develop
suited
challenges
rapidly
changing
world.
Forests,
Год журнала:
2025,
Номер
16(2), С. 269 - 269
Опубликована: Фев. 4, 2025
Accurate
monitoring
and
assessment
of
forest
disturbance
recovery
dynamics
are
essential
for
sustainable
management,
particularly
in
ecological
transition
zones.
This
study
analyzed
patterns
China’s
Funiu
Mountains
from
1991
to
2020
by
integrating
the
LandTrendr
algorithm
with
space-time
cube
analysis.
Using
Landsat
time
series
data
Geodetector
method,
we
examined
both
spatiotemporal
characteristics
driving
factors
change
across
three
periods.
The
results
showed
that
(1)
between
2020,
area
experienced
131.19
km2
495.88
recovery,
processes
most
active
during
1990s;
(2)
analysis
revealed
were
predominantly
characterized
cold
spots,
suggesting
relatively
stable
conditions
despite
localized
changes;
(3)
human
activities
primary
drivers
early
period,
while
was
consistently
influenced
combined
effects
topographic
precipitation.
Additionally,
fires
emerged
as
an
important
factor
affecting
after
2010.
These
findings
enhance
our
understanding
zones
provide
empirical
support
regional
management
strategies.
also
highlight
importance
considering
spatial
temporal
dimensions
when
long-term
changes.
Remote Sensing,
Год журнала:
2025,
Номер
17(5), С. 897 - 897
Опубликована: Март 4, 2025
Forest
ecosystems
in
the
Mediterranean
basin
are
significantly
affected
by
summer
wildfires.
Drought,
extreme
temperatures,
and
strong
winds
increase
fire
risk
Greece.
This
study
explores
potential
of
NDVI
for
assessing
forecasting
post-fire
regeneration
burnt
areas
Peloponnese
(2007)
Evros
(2011).
data
from
Landsat
7
9
were
analyzed
to
identify
stages
process
dominant
vegetation
species
at
each
stage.
Comparing
pre-fire
values
highlighted
recovery
rate,
while
trendline
slope
indicated
rate.
combined
analysis
forms
a
methodology
that
allows
drawing
conclusions
about
type
prevails
after
fire.
Validation
was
conducted
using
photointerpretation
techniques
CORINE
land
cover
data.
The
findings
suggest
sclerophyllous
regenerate
faster,
fir
forests
recover
slowly
may
be
replaced
sclerophylls.
To
predict
regrowth,
two
time
series
models
(ARMA,
VARIMA)
machine
learning-based
ones
(random
forest,
XGBoost)
tested.
Their
performance
evaluated
comparing
predicted
actual
numerical
values,
calculating
error
metrics
(RMSE,
MAPE),
analyzing
how
patterns
align
with
observed
ones.
results
showed
overperformance
multivariate
need
introduce
additional
variables,
such
as
soil
characteristics
effect
climate
change
on
weather
parameters,
improve
predictions.
