Urban Informatics,
Journal Year:
2024,
Volume and Issue:
3(1)
Published: May 31, 2024
Abstract
The
resilience
of
internet
service
is
crucial
for
ensuring
consistent
communication,
situational
awareness,
facilitating
emergency
response
in
our
digitally-dependent
society.
However,
due
to
empirical
data
constraints,
there
has
been
limited
research
on
disruptions
during
extreme
weather
events.
To
bridge
this
gap,
study
utilizes
observational
datasets
performance
quantitatively
assess
the
extent
disruption
two
recent
Taking
Harris
County
United
States
as
region,
we
jointly
analyzed
hazard
severity
and
associated
context
results
show
that
events
significantly
impacted
regional
connectivity.
There
exists
a
pronounced
temporal
synchronicity
between
magnitude
severity:
hazards
intensifies,
correspondingly
escalate,
eventually
return
baseline
levels
post-event.
spatial
analyses
can
happen
even
areas
are
not
directly
by
hazards,
demonstrating
repercussions
extend
beyond
immediate
area
impact.
This
interplay
synchronization
variance
underscores
complex
relationships
Internet
disruption.
Furthermore,
socio-demographic
analysis
suggests
vulnerable
communities,
already
grappling
with
myriad
challenges,
face
exacerbated
these
events,
emphasizing
need
prioritized
disaster
mitigation
strategies
interventions
improving
services.
best
knowledge,
among
first
studies
examine
hazardous
using
quantitative
dataset.
insights
obtained
hold
significant
implications
city
administrators,
guiding
them
towards
more
resilient
equitable
infrastructure
planning.
Hydrology,
Journal Year:
2025,
Volume and Issue:
12(2), P. 25 - 25
Published: Feb. 4, 2025
Despite
significant
advancements
in
flood
forecasting
using
machine
learning
(ML)
algorithms,
recent
events
have
revealed
hydrological
behaviors
deviating
from
historical
model
development
trends.
The
record-breaking
2019
the
Ottawa
River
basin,
which
exceeded
100-year
threshold,
underscores
escalating
impact
of
climate
change
on
extremes.
These
unprecedented
highlight
limitations
traditional
ML
models,
rely
heavily
data
and
often
struggle
to
predict
extreme
floods
that
lack
representation
past
records.
This
calls
for
integrating
more
comprehensive
datasets
innovative
approaches
enhance
robustness
adaptability
changing
climatic
conditions.
study
introduces
Next-Gen
Group
Method
Data
Handling
(Next-Gen
GMDH),
an
leveraging
second-
third-order
polynomials
address
models
predicting
events.
Using
HEC-RAS
simulations,
a
synthetic
dataset
river
flow
discharges
was
created,
covering
wide
range
potential
future
with
return
periods
up
10,000
years,
accuracy
generalization
predictions
under
evolving
GMDH
addresses
complexity
standard
by
incorporating
non-adjacent
connections
optimizing
intermediate
layers,
significantly
reducing
computational
overhead
while
enhancing
performance.
Gen
demonstrated
improved
stability
tighter
clustering
predictions,
particularly
scenarios.
Testing
results
exceptional
predictive
accuracy,
Mean
Absolute
Percentage
Error
(MAPE)
values
4.72%
channel
width,
1.80%
depth,
0.06%
water
surface
elevation.
vastly
outperformed
GMDH,
yielded
MAPE
25.00%,
8.30%,
0.11%,
respectively.
Additionally,
reduced
approximately
40%,
33.88%
decrease
Akaike
Information
Criterion
(AIC)
width
impressive
581.82%
improvement
depth.
methodology
integrates
hydrodynamic
modeling
advanced
ML,
providing
robust
framework
accurate
prediction
adaptive
floodplain
management
climate.
Climate
hazards
are
escalating
in
frequency
and
severity,
with
flooding
as
a
major
threat.
The
limitations
of
the
existing
analytical
necessitate
computational
tools
for
flood
risk
management
necessitates
shift
towards
more
data-driven
strategies
informed
by
AI-driven
methods.
This
paper
explores
forefront
focusing
on
integrating
artificial
intelligence
(AI),
specifically
machine
learning
(ML)
deep
(DL)
technologies.
By
reviewing
hundreds
relevant
studies,
we
present
comprehensive
analysis
AI
applications
examining
types,
models,
spatial
scales,
input
data,
practical
applications,
to
provide
holistic
view
current
landscape
future
potential
AI-enhanced
management.
We
highlight
extent
which
solutions
can
complement
enhance
reliability
predictions
inform
mitigation
response
strategies.
also
address
prevailing
challenges,
including
data
bias
need
explainable
proposes
pathways
research
fully
harness
AI's
mitigating
risks.
underscores
promising
improving
adaptive
management,
is
crucial
safeguarding
communities
infrastructures
against
challenges
posed
floods.
AIMS environmental science,
Journal Year:
2025,
Volume and Issue:
12(1), P. 72 - 105
Published: Jan. 1, 2025
<p>Floods
have
been
identified
as
one
of
the
world's
most
common
and
widely
distributed
natural
disasters
over
last
few
decades.
Floods'
negative
impacts
could
be
significantly
reduced
if
accurately
predicted
or
forecasted
in
advance.
Apart
from
large-scale
spatiotemporal
data
greater
attention
to
Internet
Things,
worldwide
volume
digital
is
increasing.
Artificial
intelligence
plays
a
vital
role
analyzing
developing
corresponding
flood
mitigation
plan,
prediction,
forecast.
Machine
learning
(ML)-based
models
recently
received
much
due
their
self-learning
capabilities
without
incorporating
any
complex
physical
processes.
This
study
provides
comprehensive
review
ML
approaches
used
forecasting,
classification
tasks,
serving
guide
for
future
challenges.
The
importance
challenges
applying
these
techniques
prediction
are
discussed.
Finally,
recommendations
directions
analysis
presented.</p>
