Frontiers in Environmental Science,
Journal Year:
2024,
Volume and Issue:
12
Published: Sept. 25, 2024
Multi-site
PM2.5
prediction
has
emerged
as
a
crucial
approach,
given
that
the
accuracy
of
models
based
solely
on
data
from
single
monitoring
station
may
be
constrained.
However,
existing
multi-site
methods
predominantly
rely
recurrent
networks
for
extracting
temporal
dependencies
and
overlook
domain
knowledge
related
to
air
quality
pollutant
dispersion.
This
study
aims
explore
whether
superior
architecture
exists
not
only
approximates
performance
through
feedforward
but
also
integrates
PM2.5.
Consequently,
we
propose
novel
spatio-temporal
attention
causal
convolutional
neural
network
(Causal-STAN)
predicting
concentrations
at
multiple
sites
in
Yangtze
River
Delta
region
China.
Causal-STAN
comprises
two
components:
feature
integration
module,
which
identifies
local
correlation
trends
spatial
correlations
data,
extracts
inter-site
directional
residual
block
delineate
features
concentration
dispersion
between
sites;
captures
internal
information
long-term
time
series.
was
evaluated
using
one-year
247
mainland
Compared
six
state-of-the-art
baseline
models,
achieves
optimal
6-hour
future
predictions,
surpassing
model
reducing
error
by
8%–10%.
Remote Sensing,
Journal Year:
2024,
Volume and Issue:
16(3), P. 467 - 467
Published: Jan. 25, 2024
Long-term
exposure
to
high
concentrations
of
fine
particles
can
cause
irreversible
damage
people’s
health.
Therefore,
it
is
extreme
significance
conduct
large-scale
continuous
spatial
particulate
matter
(PM2.5)
concentration
prediction
for
air
pollution
prevention
and
control
in
China.
The
distribution
PM2.5
ground
monitoring
stations
China
uneven
with
a
larger
number
southeastern
China,
while
the
sites
also
insufficient
quality
control.
Remote
sensing
technology
obtain
information
quickly
macroscopically.
possible
predict
based
on
multi-source
remote
data.
Our
study
took
as
research
area,
using
Pearson
correlation
coefficient
GeoDetector
select
auxiliary
variables.
In
addition,
long
short-term
memory
neural
network
random
forest
regression
model
were
established
estimation.
We
finally
selected
(R2
=
0.93,
RMSE
4.59
μg
m−3)
our
by
evaluation
index.
across
2021
was
estimated,
then
influence
factors
high-value
regions
explored.
It
clear
that
not
only
related
local
geographical
meteorological
conditions,
but
closely
economic
social
development.
Toxics,
Journal Year:
2025,
Volume and Issue:
13(4), P. 254 - 254
Published: March 28, 2025
Surface
air
pollution
affects
ecosystems
and
people’s
health.
However,
traditional
models
have
low
prediction
accuracy.
Therefore,
a
hybrid
model
for
accurately
predicting
daily
surface
PM2.5
concentrations
was
integrated
with
wavelet
(W),
convolutional
neural
network
(CNN),
bidirectional
long
short-term
memory
(BiLSTM),
gated
recurrent
unit
(BiGRU).
The
data
meteorological
factors
pollutants
in
Guangzhou
City
from
2014
to
2020
were
utilized
as
inputs
the
models.
W-CNN-BiGRU-BiLSTM
demonstrated
strong
performance
during
phase,
achieving
an
R
(correlation
coefficient)
of
0.9952,
root
mean
square
error
(RMSE)
1.4935
μg/m3,
absolute
(MAE)
1.2091
percentage
(MAPE)
7.3782%.
Correspondingly,
accurate
is
beneficial
control
urban
planning.
