Climate
models
typically
provide
air
temperature
estimates
at
lower
resolutions,
lacking
the
necessary
details
for
urban
climate
studies.
These
require
significant
computational
resources
and
time
to
estimate
temperatures
higher
resolution,
which
are
not
easily
accessible
city
scale.
In
contrast,
data-driven
approaches
offer
accuracy
speed
in
downscaling.
this
study,
a
framework
downscaling
derived
from
such
as
UrbClim
was
developed.
The
proposed
utilized
morphological
features
extracted
LiDAR
data.
To
extract
features,
first
three-dimensional
building
model
created
using
data
deep
learning
models.
Then,
these
were
integrated
with
meteorological
parameters
wind,
humidity,
etc.,
downscale
machine
algorithms.
results
demonstrated
that
developed
effectively
Deep
algorithms
played
crucial
role
generating
extracting
aforementioned
features.
Also,
evaluation
of
various
indicated
LightGBM
had
best
performance
an
RMSE
0.352°K
MAE
0.215°K.
Furthermore,
examination
final
maps
showed
successfully
estimated
enabling
identification
local
patterns
street
level.
source
codes
corresponding
research
paper
available
on
GitHub
via
https://github.com/FatemehCh97/Air-Temperature-Downscaling
Research Square (Research Square),
Journal Year:
2024,
Volume and Issue:
unknown
Published: May 7, 2024
Abstract
Campus
areas
as
a
microcosm
of
urban
areas;
given
the
context
global
warming,
are
becoming
more
vulnerable
to
rising
temperatures.
This
study
focuses
on
outdoor
environment
and
microclimate
effects
Ozyegin
campus
by
considering
surface
plantation
types.
Urban
green
spaces
offer
potential
solution
lowering
air
temperatures
through
shading
evapotranspiration.
The
selection
appropriate
plant
types
is
crucial
for
effective
temperature
reduction,
leaves
act
barriers
solar
radiation.
Measurements
were
conducted
in
November–December
2023
at
15
designated
points
campus.
measurements
especially
autumn
diffuse
daylight
prevent
effect
direct
radiation
high
difference
trees.
research
seeks
address
fundamental
questions
about
how
different
surfaces,
both
hard
soft,
influence
thermal
conditions,
explore
university
campuses,
strategies
improvement.
Employing
comprehensive
field
surveys
data
analysis,
including
statistical
techniques
like
ANOVA,
Bonferroni
post-hoc
test,
reveals
under
broad-leaved
trees
1.5
degrees
cooler
than
surfaces.
With
practical
objective,
aims
measure
conditions
make
recommendations
creating
comfortable
environments.
ISPRS International Journal of Geo-Information,
Journal Year:
2024,
Volume and Issue:
13(6), P. 190 - 190
Published: June 7, 2024
Understanding
solar
radiation
in
urban
street
spaces
is
crucial
for
comprehending
residents’
environmental
experiences
and
enhancing
their
quality
of
life.
However,
existing
studies
rarely
focus
on
the
patterns
over
time
across
different
suburban
areas.
In
this
study,
view
images
from
summers
2013
2019
Shanghai
were
used
to
calculate
spaces.
The
results
show
a
general
decrease
compared
2013,
with
an
average
drop
12.34%.
was
most
significant
October
(13.47%)
least
May
(11.71%).
terms
data
gathered
sampling
points,
76.57%
showed
decrease,
while
23.43%
increase.
Spatially,
decreased
by
79.66%
every
additional
1.5
km
city
centre.
summary,
generally
shows
decreasing
trend,
variations
between
These
findings
are
vitally
important
guiding
planning,
optimising
green
infrastructure,
ecological
environment,
further
promoting
sustainable
development
improving
Climate
models
typically
provide
air
temperature
estimates
at
lower
resolutions,
lacking
the
necessary
details
for
urban
climate
studies.
These
require
significant
computational
resources
and
time
to
estimate
temperatures
higher
resolution,
which
are
not
easily
accessible
city
scale.
In
contrast,
data-driven
approaches
offer
accuracy
speed
in
downscaling.
this
study,
a
framework
downscaling
derived
from
such
as
UrbClim
was
developed.
The
proposed
utilized
morphological
features
extracted
LiDAR
data.
To
extract
features,
first
three-dimensional
building
model
created
using
data
deep
learning
models.
Then,
these
were
integrated
with
meteorological
parameters
wind,
humidity,
etc.,
downscale
machine
algorithms.
results
demonstrated
that
developed
effectively
Deep
algorithms
played
crucial
role
generating
extracting
aforementioned
features.
Also,
evaluation
of
various
indicated
LightGBM
had
best
performance
an
RMSE
0.352°K
MAE
0.215°K.
Furthermore,
examination
final
maps
showed
successfully
estimated
enabling
identification
local
patterns
street
level.
source
codes
corresponding
research
paper
available
on
GitHub
via
https://github.com/FatemehCh97/Air-Temperature-Downscaling
