Frontiers in Environmental Science,
Год журнала:
2024,
Номер
12
Опубликована: Июль 31, 2024
In
the
context
of
global
climate
change
and
rising
anthropogenic
loads,
outbreaks
both
endemic
invasive
pests,
pathogens,
diseases
pose
an
increasing
threat
to
health,
resilience,
productivity
natural
forests
forest
plantations
worldwide.
The
effective
management
such
threats
depends
on
opportunity
for
early-stage
action
helping
limit
damage
expand,
which
is
difficult
implement
large
territories.
Recognition
technologies
based
analysis
Earth
observation
data
are
basis
tools
monitoring
spread
degradation
processes,
supporting
pest
population
control,
management,
conservation
strategies
in
general.
this
study,
we
present
a
machine
learning-based
approach
recognizing
damaged
using
open
source
remote
sensing
images
Sentinel-2
supported
with
Google
example
bark
beetle,
Polygraphus
proximus
Blandford,
polygraph.
For
algorithm
development,
first
investigated
annotated
channels
corresponding
color
perception—red,
green,
blue—available
at
Earth.
Deep
neural
networks
were
applied
two
problem
formulations:
semantic
segmentation
detection.
As
result
conducted
experiments,
developed
model
that
quantitative
assessment
changes
target
objects
high
accuracy,
achieving
84.56%
F1-score,
determining
number
trees
estimating
areas
occupied
by
withered
stands.
obtained
masks
further
integrated
medium-resolution
achieved
81.26%
opened
operational
systems
recognize
region,
making
solution
rapid
cost-effective.
Additionally,
unique
dataset
has
been
collected
polygraph
region
study.
Current Forestry Reports,
Год журнала:
2023,
Номер
9(4), С. 219 - 229
Опубликована: Июнь 15, 2023
Abstract
Purpose
of
Review
Forest
models
are
becoming
essential
tools
in
forest
research,
management,
and
policymaking
but
currently
under
deep
transformation.
In
this
review
the
most
recent
literature
(2018–2022),
we
aim
to
provide
an
updated
general
view
main
topics
attracting
efforts
modelers,
trends
already
place,
some
current
future
challenges
that
field
will
face.
Recent
Findings
Four
major
on
modelling
efforts:
data
acquisition,
productivity
estimation,
ecological
pattern
predictions,
management
related
ecosystem
services.
Although
may
seem
different,
they
all
converging
towards
integrated
approaches
by
pressure
climate
change
as
coalescent
force,
pushing
research
into
mechanistic,
cross-scale
simulations
functioning
structure.
Summary
We
conclude
is
experiencing
exciting
challenging
time,
due
combination
new
methods
easily
acquire
massive
amounts
data,
techniques
statistically
process
such
refinements
mechanistic
incorporating
higher
levels
complexity
breaking
traditional
barriers
spatial
temporal
scales.
However,
available
also
creating
challenges.
any
case,
increasingly
acknowledged
a
community
interdisciplinary
effort.
As
such,
ways
deliver
simplified
versions
or
easy
entry
points
should
be
encouraged
integrate
non-modelers
stakeholders
since
its
inception.
This
considered
particularly
academic
modelers
increasing
mathematical
models.
Ecology and Evolution,
Год журнала:
2021,
Номер
11(10), С. 5713 - 5727
Опубликована: Март 26, 2021
Abstract
Invasive
species
have
considerably
increased
in
recent
decades
due
to
direct
and
indirect
effects
of
ever‐increasing
international
trade
rates
new
climate
conditions
derived
from
global
change.
We
need
better
understand
how
the
dynamics
early
invasions
develop
these
result
impacts
on
invaded
ecosystems.
Here
we
studied
distribution
severe
defoliation
processes
box
tree
moth
(
Cydalima
perspectalis
W.),
a
defoliator
insect
native
Asia
invasive
Europe
since
2007,
through
combination
models
based
landscape
composition
information.
The
results
showed
that
data
areas
was
most
effective
methodology
for
appropriate
modeling.
not
influenced
by
overall
factors,
but
only
presence
its
host
plant,
dispersal
capacity,
suitability.
Such
suitability
described
low
precipitation
seasonality
minimum
annual
temperatures
around
0°C,
defining
continentality
effect
throughout
territory.
emphasize
studying
separately
because
identified
slightly
involved
limiting
spread
strongly
constrained
ecosystem
impact
terms
before
reaches
equilibrium
with
environment.
New
studies
habitat
recovery
after
disturbance,
ecological
consequences
such
impact,
community
context
change
are
required
understanding
this
species.
Remote Sensing,
Год журнала:
2024,
Номер
16(12), С. 2060 - 2060
Опубликована: Июнь 7, 2024
The
fall
armyworm
(Spodoptera
frugiperda)
(J.
E.
Smith)
is
a
widespread,
polyphagous,
and
highly
destructive
agricultural
pest.
Global
climate
change
may
facilitate
its
spread
to
new
suitable
areas,
thereby
increasing
threats
host
plants.
Consequently,
predicting
the
potential
distribution
for
plants
under
current
future
scenarios
crucial
assessing
outbreak
risks
formulating
control
strategies.
This
study,
based
on
remote
sensing
assimilation
data
plant
protection
survey
data,
utilized
machine
learning
methods
(RF,
CatBoost,
XGBoost,
LightGBM)
construct
prediction
models
120
Hyperparameter
stacking
ensemble
method
(SEL)
were
introduced
optimize
models.
results
showed
that
SEL
demonstrated
optimal
performance
in
armyworm,
with
an
AUC
of
0.971
±
0.012
TSS
0.824
0.047.
Additionally,
LightGBM
47
30
plants,
respectively.
Overlay
analysis
suggests
overlap
areas
interaction
links
between
will
generally
increase
future,
most
significant
rise
RCP8.5
scenario,
indicating
threat
further
intensify
due
change.
findings
this
study
provide
support
planning
implementing
global
intercontinental
long-term
pest
management
measures
aimed
at
mitigating
impact
food
production.
Frontiers in Environmental Science,
Год журнала:
2024,
Номер
12
Опубликована: Июль 31, 2024
In
the
context
of
global
climate
change
and
rising
anthropogenic
loads,
outbreaks
both
endemic
invasive
pests,
pathogens,
diseases
pose
an
increasing
threat
to
health,
resilience,
productivity
natural
forests
forest
plantations
worldwide.
The
effective
management
such
threats
depends
on
opportunity
for
early-stage
action
helping
limit
damage
expand,
which
is
difficult
implement
large
territories.
Recognition
technologies
based
analysis
Earth
observation
data
are
basis
tools
monitoring
spread
degradation
processes,
supporting
pest
population
control,
management,
conservation
strategies
in
general.
this
study,
we
present
a
machine
learning-based
approach
recognizing
damaged
using
open
source
remote
sensing
images
Sentinel-2
supported
with
Google
example
bark
beetle,
Polygraphus
proximus
Blandford,
polygraph.
For
algorithm
development,
first
investigated
annotated
channels
corresponding
color
perception—red,
green,
blue—available
at
Earth.
Deep
neural
networks
were
applied
two
problem
formulations:
semantic
segmentation
detection.
As
result
conducted
experiments,
developed
model
that
quantitative
assessment
changes
target
objects
high
accuracy,
achieving
84.56%
F1-score,
determining
number
trees
estimating
areas
occupied
by
withered
stands.
obtained
masks
further
integrated
medium-resolution
achieved
81.26%
opened
operational
systems
recognize
region,
making
solution
rapid
cost-effective.
Additionally,
unique
dataset
has
been
collected
polygraph
region
study.