Jurnal Tanah dan Sumberdaya Lahan,
Год журнала:
2025,
Номер
12(1), С. 9 - 19
Опубликована: Янв. 1, 2025
A
rice
field
land-use
map
is
essential
in
the
sustainable
land
management
of
fields
for
physical
monitoring
and
planning.
Such
maps
are
usually
created
using
multitemporal
image
data
with
a
spectral
approach,
but
this
method
can
only
be
applied
to
certain
areas
cannot
easily
other
different
characteristics.
While
has
been
widely
used
by
researchers
proven
effective,
single-date
imagery
more
efficient.
This
study
aimed
based
on
Sentinel-2
landform
maps.
These
were
derived
through
visual
interpretation
false
colour
composite
bands,
DEMNAS,
system
map.
The
resulted
eleven
classes.
landscape
ecology
approach
assumed
influence
landforms
land-use.
use
ten
optical
bands
multispectral
classification
maximum
likelihood
algorithm
convolutional
neural
network
twelve
cover
implemented
two-dimensional
ecological
spatial
relationship
matrix
that
produced
nine
obtained
an
overall
accuracy
90,28%
Kappa
0,87.
result
was
better
than
algorithm,
which
86,81%
0,83.
class
had
total
area
33.686,69
ha
mean
absolute
error
(MAE)
value
0,0241,
while
29.590,21
larger
MAE
0,0343.
Environmental Monitoring and Assessment,
Год журнала:
2024,
Номер
196(12)
Опубликована: Ноя. 6, 2024
The
spatiotemporal
dynamics
of
forest
cover
are
essential
for
understanding
the
patterns
and
processes
change
over
time
space.
This
research
focused
on
trends
drivers
in
Metekel
Zone
Northwest
Ethiopia.
Landsat
5,
7,
8
imagery,
covering
period
from
1986
to
2019,
were
used
land
use/cover
classification.
Land
classification
was
performed
using
random
(RF)
support
vector
machine
(SVM)
algorithms
Google
Earth
Engine
(GEE)
platform,
with
training
samples
obtained
through
visual
image
interpretation.
Spectral
indices,
such
as
normalized
difference
vegetation
index,
soil-adjusted
leaf
area
water
analyzed
examine
time.
In
addition,
key
informant
interviews
(KIIs)
focus
group
discussions
(FGDs)
conducted.
Findings
revealed
that
decreased
significantly
51.37%
17.20%
driven
largely
by
human
activities
agricultural
expansion,
increased
demand
firewood,
urban
expansion.
spectral
indices
further
corroborated
finding
study
region
(mainly
southwestern
part)
substantially
2019.
Concerning
depletion,
lack
local
community
awareness
has
become
a
challenge.
problem
is
attributed
communities
prioritizing
immediate
needs
fuel
agriculture
long-term
conservation.
To
combat
ongoing
deforestation,
Administration,
collaboration
administration
office
other
stakeholders,
revisited
strengthened
existing
policies
control
systems.
It
also
suggested
awareness,
chiefly
among
youth,
should
be
enhanced
strategic
expansion
formal
nonformal
educational
initiatives,
which
empower
youth
agents
promote
dissemination
knowledge
throughout
community.
Land,
Год журнала:
2024,
Номер
13(10), С. 1566 - 1566
Опубликована: Сен. 26, 2024
Rapid
urbanization
in
developing
countries
leads
to
significant
land-use
and
land-cover
change
(LULCC),
which
contributes
increased
carbon
dioxide
(CO2)
emissions
the
degradation
of
storage.
Studying
spatio-temporal
changes
storage
is
crucial
for
guiding
sustainable
urban
development
toward
neutrality.
This
study
integrates
machine-learning
random
forest
algorithm,
CA–Markov,
InVEST
models
predict
distribution
Shenzhen,
China,
under
various
scenarios.
The
findings
indicate
that,
over
past
two
decades,
Shenzhen
has
experienced
changes.
transformation
from
high-
low-carbon-density
land
uses,
particularly
conversion
forestland
construction
land,
primary
cause
loss.
Forestland
mainly
influenced
by
natural
factors,
such
as
digital
elevation
model
(DEM)
precipitation,
while
other
(LULC)
types
are
predominantly
affected
socio-economic
demographic
factors.
By
2030,
projected
vary
significantly
across
different
scenarios,
with
greatest
decline
expected
scenario
(NDS)
least
ecological
priority
(EPS).
RF-CA–Markov
outperforms
traditional
CA–Markov
accurately
simulating
use,
small
scattered
types.
Our
conclusions
can
inform
future
low-carbon
city
optimization.
Eng—Advances in Engineering,
Год журнала:
2024,
Номер
5(4), С. 3397 - 3426
Опубликована: Дек. 16, 2024
Deteriorating
road
infrastructure
is
a
global
concern,
especially
in
low-income
countries
where
financial
and
technological
constraints
hinder
effective
monitoring
maintenance.
Traditional
methods,
like
inertial
profilers,
are
expensive
complex,
making
them
unsuitable
for
large-scale
use.
This
paper
explores
the
integration
of
cost-effective,
scalable
smartphone
technologies
surface
monitoring.
Smartphone
sensors,
such
as
accelerometers
gyroscopes,
combined
with
data
preprocessing
techniques
filtering
reorientation,
improve
quality
collected
data.
Machine
learning
algorithms,
particularly
CNNs,
utilized
to
classify
anomalies,
enhancing
detection
accuracy
system
efficiency.
The
results
demonstrate
that
smartphone-based
systems,
paired
advanced
processing
machine
learning,
significantly
reduce
cost
complexity
traditional
surveys.
Future
work
could
focus
on
improving
sensor
calibration,
synchronization,
models
handle
diverse
real-world
conditions.
These
advancements
will
increase
scalability
urban
areas
requiring
real-time
rapid
Jurnal Tanah dan Sumberdaya Lahan,
Год журнала:
2025,
Номер
12(1), С. 9 - 19
Опубликована: Янв. 1, 2025
A
rice
field
land-use
map
is
essential
in
the
sustainable
land
management
of
fields
for
physical
monitoring
and
planning.
Such
maps
are
usually
created
using
multitemporal
image
data
with
a
spectral
approach,
but
this
method
can
only
be
applied
to
certain
areas
cannot
easily
other
different
characteristics.
While
has
been
widely
used
by
researchers
proven
effective,
single-date
imagery
more
efficient.
This
study
aimed
based
on
Sentinel-2
landform
maps.
These
were
derived
through
visual
interpretation
false
colour
composite
bands,
DEMNAS,
system
map.
The
resulted
eleven
classes.
landscape
ecology
approach
assumed
influence
landforms
land-use.
use
ten
optical
bands
multispectral
classification
maximum
likelihood
algorithm
convolutional
neural
network
twelve
cover
implemented
two-dimensional
ecological
spatial
relationship
matrix
that
produced
nine
obtained
an
overall
accuracy
90,28%
Kappa
0,87.
result
was
better
than
algorithm,
which
86,81%
0,83.
class
had
total
area
33.686,69
ha
mean
absolute
error
(MAE)
value
0,0241,
while
29.590,21
larger
MAE
0,0343.