Geomatics Natural Hazards and Risk,
Год журнала:
2024,
Номер
15(1)
Опубликована: Июль 9, 2024
Identifying
an
inundation
area
after
a
flood
event
is
essential
for
planning
emergency
rescue
operations.
In
this
study,
we
propose
method
to
automatically
determine
inundated
road
segments
by
floods
using
image
recognition
technology,
deep
learning
model,
and
elevation
data.
First,
develop
training
model
aerial
photographs
captured
during
event.
Then,
the
applied
another
The
visualizes
status
of
roads
on
100-m
mesh-by-mesh
basis
integrating
information
whether
mesh
includes
targeted
segments.
Our
results
showed
that
F-score
was
higher,
89%–91%,
when
only
with
15
m
or
less.
Moreover,
visualizing
in
GIS
facilitated
classification
roads,
even
within
same
mesh,
which
relevant
finding
complements
object
detection.
Applied Sciences,
Год журнала:
2025,
Номер
15(1), С. 465 - 465
Опубликована: Янв. 6, 2025
Emergency
situation
awareness
during
sudden
natural
disasters
presents
significant
challenges.
Traditional
methods,
characterized
by
low
spatial
and
temporal
resolution
as
well
coarse
granularity,
often
fail
to
comprehensively
capture
disaster
situations.
However,
social
media
platforms,
a
vital
source
of
sensing,
offer
potential
supplement
situational
awareness.
This
paper
proposes
an
innovative
framework
for
based
on
multimodal
data
from
identify
content
related
typhoon
disasters.
Integrating
text
image
facilitates
near
real-time
monitoring
the
public
perspective.
In
this
study,
Typhoon
Haikui
(Strong
No.
11
2023)
was
chosen
case
study
validate
effectiveness
proposed
method.
We
employed
ERNIE
language
processing
model
complement
Deeplab
v3+
deep
learning
semantic
segmentation
extracting
damage
information
media.
A
visualization
analysis
disaster-affected
areas
performed
categorizing
types.
Additionally,
Geodetector
used
investigate
heterogeneity
its
underlying
factors.
approach
allowed
us
analyze
spatiotemporal
patterns
evolution,
enabling
rapid
assessment
facilitating
emergency
response
efforts.
The
results
show
that
method
significantly
enhances
effectively
identifying
different
types
sensing
data.
Water,
Год журнала:
2024,
Номер
16(17), С. 2476 - 2476
Опубликована: Авг. 30, 2024
In
the
context
of
increasing
frequency
urban
flooding
disasters
caused
by
extreme
weather,
accurate
and
timely
identification
monitoring
flood
risks
have
become
increasingly
important.
This
article
begins
with
a
bibliometric
analysis
literature
on
identification,
revealing
that
since
2017,
this
area
has
global
research
hotspot.
Subsequently,
it
presents
systematic
review
current
mainstream
technologies,
drawing
from
both
traditional
emerging
data
sources,
which
are
categorized
into
sensor-based
(including
contact
non-contact
sensors)
big
data-based
social
media
surveillance
camera
data).
By
analyzing
advantages
disadvantages
each
technology
their
different
focuses,
paper
points
out
largely
emphasizes
more
“intelligent”
technologies.
However,
these
technologies
still
certain
limitations,
sensor
techniques
retain
significant
in
practical
applications.
Therefore,
future
risk
should
focus
integrating
multiple
fully
leveraging
strengths
sources
to
achieve
real-time
flooding.