Abstract.
Understanding
the
spatiotemporal
characteristics
of
long-
and
short-term
exposure
to
ground
ozone
is
crucial
for
improving
environmental
management
health
studies.
However,
such
studies
have
been
constrained
by
availability
high-resolution
data.
To
address
this,
we
characterized
ground-level
variations
risks
across
multiple
spatial
(pixel,
county,
region,
national)
temporal
(daily,
monthly,
seasonal,
annual)
scales
using
daily
1-km
data
from
2000
2020,
derived
satellite
LST
via
a
machine-learning
method.
The
model
provided
reliable
estimates,
validated
through
rigorous
cross-validation
direct
comparison
with
external
measurements.
Our
long-term
estimates
revealed
seasonal
shifts
in
high-exposure
centers:
spring
eastern
China,
summer
North
China
Plain
(NCP),
autumn
Pearl
River
Delta
(PRD).
A
non-monotonous
trend
was
observed,
levels
rising
2001–2007
at
rate
0.47
μg/m3/year,
declining
after
2008
(-0.58
μg/m3/year),
increasing
significantly
2016–2020
(1.16
accompanied
regional
fluctuations.
Notably,
increased
0.63
μg/m3/year
NCP
during
second
phase,
6.38
PRD
third
phase.
Exposure
over
100
μg/m3
shifted
June
May,
exceeding
160
were
primarily
seen
NCP,
showing
an
expanding
trend.
day-to-day
analysis
highlights
influence
meteorological
factors
on
extreme
events.
These
findings
emphasize
need
stronger
mitigation
efforts.
Remote Sensing,
Год журнала:
2024,
Номер
16(7), С. 1298 - 1298
Опубликована: Апрель 7, 2024
Air
pollution
has
been
standing
as
one
of
the
most
pressing
global
challenges.
The
changing
patterns
air
pollutants
at
different
spatial
and
temporal
scales
have
substantially
studied
all
over
world,
which,
however,
were
intricately
disturbed
by
COVID-19
subsequent
containment
measures.
Understanding
fine-scale
stages
epidemic’s
course
is
necessary
for
better
identifying
region-specific
drivers
preparing
environmental
decision
making
during
future
epidemics.
Taking
China
an
example,
this
study
developed
a
multi-output
LightGBM
approach
to
estimate
monthly
concentrations
six
major
(i.e.,
PM2.5,
PM10,
NO2,
SO2,
O3,
CO)
in
revealed
distinct
spatiotemporal
each
pollutant
course.
5-year
period
2019–2023
was
selected
observe
changes
from
pre-COVID-19
era
lifting
performance
our
model,
assessed
cross-validation
R2,
demonstrated
high
accuracy
with
values
0.92
0.95
0.90
0.79
0.82
CO.
Notably,
there
improvement
particulate
matter,
particularly
although
PM10
exhibited
rebound
northern
regions.
SO2
CO
consistently
declined
across
country
(p
<
0.001
p
0.05,
respectively),
while
O3
southern
regions
experienced
notable
increase.
Concentrations
Beijing–Tianjin–Hebei
region
effectively
controlled
mitigated.
findings
provide
critical
insights
into
trends
quality
public
health
emergencies,
help
guide
development
targeted
interventions,
inform
policy
aimed
reducing
disease
burdens
associated
pollution.
Environmental Science & Technology,
Год журнала:
2024,
Номер
58(36), С. 15938 - 15948
Опубликована: Авг. 28, 2024
Accurately
mapping
ground-level
ozone
concentrations
at
high
spatiotemporal
resolution
(daily,
1
km)
is
essential
for
evaluating
human
exposure
and
conducting
public
health
assessments.
This
requires
identifying
understanding
a
proxy
that
well-correlated
with
variation
available
high-resolution
data.
study
introduces
modeling
method
utilizing
the
XGBoost
algorithm
satellite-derived
land
surface
temperature
(LST)
as
primary
predictor.
Focusing
on
China
in
2019,
our
model
achieved
cross-validation
