Geoderma,
Год журнала:
2023,
Номер
439, С. 116697 - 116697
Опубликована: Окт. 24, 2023
Optical
remote
sensing
satellites
provide
rapid
access
to
regional
topsoil
salinization
mapping.
However,
mapping
based
on
spectral
reflectance
is
always
affected
by
background
material
like
vegetation
cover,
straw
mulching
and
soil
types.
In
light
of
these
challenges,
this
study
investigates
the
potential
image
fusion,
where
images
original
bare
pixels
were
combined,
minimize
impact
cover
salinity
A
case
was
presented
for
typical
area
using
synchronized
Sentinel-2
MSI
(named
image)
255
ground-truth
data
collected
in
October
2020,
aligning
with
periods
salt
return.
Furthermore,
obtain
novel
pixels,
multi-temporal
acquired
during
two
distinct
intervals:
March
May
September
November,
spanning
years
from
2018
2021.
The
synthetic
(SYSI)
obtained
extracting
images.
Two
(original,
SYSI)
fused
non-negative
matrix
factorization
(NMF)
method,
named
SYSIfused.
Then,
stacking
machine
algorithm
used
under
different
types,
evaluating
SYSIfused
accuracy
prediction.
results
showed
outperformed
(the
R2
best
models
increased
0.054–0.242,
RMSE
MAE
decreased
0.049–0.780
0.012–0.546,
respectively).
Based
SYSIfused,
order
effect
types
coastal
bog
solonchaks
>
alluvial
cinnamon
coral
saline
overall
samples,
their
roles
improving
model
0.141,
0.085,
0.022,
0.012,
respectively.
Besides,
provided
prediction
performances
(R2
=
0.742,
0.377,
0.362).
This
introduces
concept
merging
SYSI,
resulting
a
significant
improvement
areas
covered
vegetation.
ISPRS Journal of Photogrammetry and Remote Sensing,
Год журнала:
2023,
Номер
199, С. 40 - 60
Опубликована: Апрель 3, 2023
The
use
of
remote
sensing
data
methods
is
affordable
for
the
mapping
soil
properties
plowed
layer
over
croplands.
Carried
out
in
framework
ongoing
STEROPES
project
European
Joint
H2020
Program
SOIL,
this
work
focused
on
feasibility
Sentinel-2
based
approaches
high
resolution
topsoil
clay
and
organic
carbon
(SOC)
contents
at
within-farm
or
within-field
scales,
cropland
sites
contrasted
climates
types
across
Northern
hemisphere.
Four
pixelwise
temporal
mosaicking
methods,
using
a
two
years-Sentinel-2
time
series
several
spectral
indices
(NDVI,
NBR2,
BSI,
S2WI),
were
developed
compared
i)
pure
bare
condition
(maxBSI),
ii)
driest
(minS2WI),
iii)
average
(Median)
iv)
dry
conditions
excluding
extreme
reflectance
values
(R90).
Three
modeling
approaches,
bands
output
mosaics
as
covariates,
tested
compared:
(i)
Quantile
Regression
Forest
(QRF)
algorithm;
(ii)
QRF
adding
longitude
latitude
covariates
(QRFxy);
(iii)
hybrid
approach,
Linear
Mixed
Effect
Model
(LMEM),
that
includes
spatial
autocorrelation
properties.
We
pairs
mosaic
ten
Türkiye,
Italy,
Lithuania,
USA
where
samples
collected
SOC
content
measured
lab.
RPIQ
best
performances
among
test
was
2.50
both
(RMSE
=
0.15%)
3.3%).
Both
accuracy
level
uncertainty
mainly
influenced
by
site
characteristics
cloud
frequency,
management.
Generally,
models
including
component
(QRFxy
LMEM)
performing,
while
mostly
Median
R90.
most
frequent
optimal
combination
model
type
R90
QRFxy
SOC,
LMEM
estimation.
Journal of Soils and Sediments,
Год журнала:
2024,
Номер
24(11), С. 3556 - 3571
Опубликована: Окт. 5, 2024
Abstract
Purpose
Accurately
assessing
soil
organic
carbon
(SOC)
content
is
vital
for
ecosystem
services
management
and
addressing
global
climate
challenges.
This
study
undertakes
a
comprehensive
bibliometric
analysis
of
estimates
SOC
using
remote
sensing
(RS)
machine
learning
(ML)
techniques.
It
showcases
the
historical
growth
thematic
evolution
in
research,
aiming
to
amplify
understanding
estimation
themes
provide
scientific
support
change
adaptation
mitigation.
Materials
Methods
Employing
extensive
literature
database
analysis,
network
clustering
techniques,
reviews
1,761
articles
on
RS
technologies
490
employing
both
ML
technologies.
Results
Discussion
The
results
indicate
that
satellite-based
RS,
particularly
Landsat
series,
predominant
other
associated
studies,
with
North
America,
China,
Europe
leading
evaluations
Africa
having
low
adopting
technology.
Trends
research
demonstrate
an
from
basic
mapping
advanced
topics
such
as
(C)
sequestration,
complex
modeling,
big
data
utilization.
Thematic
clusters
co-occurrence
suggest
interplay
between
technology
development,
environmental
surveys,
properties,
dynamics.
Conclusion
highlights
synergy
ML,
techniques
proving
be
critical
accurate
estimation.
These
findings
are
crucial
estimation,
informed
strategic
decision-making.
Geoderma,
Год журнала:
2024,
Номер
449, С. 116987 - 116987
Опубликована: Авг. 1, 2024
Sustainable
cropland
management
requires
quantitative
and
up-to-date
information
on
the
spatial
pattern
of
soil
organic
carbon
(SOC)
at
scales
relevant
for
implementing
targeted
conservation
measures.
Spectra-based
remote
sensing
SOC
in
croplands
is
promising,
but
it
extraction
high-quality
bare
pixels
that
enable
spatially
continuous
coverage.
Here,
we
aim
to
compare
predictive
capability
single-date
versus
multitemporal
compositing
images
an
intensively
cultivated
region
(4,700
km2)
northeast
China.
A
series
12
within
2017–2022
were
processed
passed
onto
three
approaches
(geometric
median,
univariate
mean
median)
create
mosaics.
Both
spectral
images,
together
with
laboratory-simulated
Sentinel-2
benchmark
data,
used
develop
partial
least
squares
regression,
Cubist
random
forest
models
via
100
bootstrapped
validations.
With
consistently
being
best
performing
algorithm
all
data
sources,
results
show
exhibited
temporally
unstable
performance
(R2:
0.30–0.67).
Among
approaches,
high-dimensional
geometric
median
composite
was
most
suitable
because
(i)
its
close
resemblance
laboratory
reference
robustness
outliers,
which
yielded
a
model
0.64;
RMSE:
2.24
g/kg)
outperforming
11
out
models;
(ii)
ability
retain
between-band
relationships
allowed
further
incorporation
SOC-relevant
indexes,
led
6.5
%
increase
prediction
accuracy.
The
resultant
map
highlighted
imaging
reveal
field-scale
degradation
patterns.
Future
work
should
explore
possibility
extending
purely
spectra-based
framework
integrated
mapping
monitoring
additional
biophysical
information.
Geoderma,
Год журнала:
2024,
Номер
446, С. 116905 - 116905
Опубликована: Май 7, 2024
Erosion-induced
lateral
soil
redistribution
leads
to
spatially
heterogenous
composition,
which
can
be
captured
through
the
distinctive
spectral
reflectance
of
soils
under
varying
levels
erosion
influence.
This
points
potential
using
remotely
sensed
spectra
detect
severe
and
deposition
hotspots
in
exposed
croplands
and,
importantly,
further
differentiate
intra-class
variability
moderate
that
often
occupies
largest
proportion.
Here,
we
aim
develop
a
two-step
classification
mapping
approach
based
on
multitemporal
compositing
Sentinel-2
bare
images
typical
agricultural
region
(11,500
km2)
northeast
China.
A
random
forest
classifier
was
firstly
trained
against
ground-truth
data
derived
from
very
high
resolution
(VHR)
imagery
Google
Earth,
with
an
overall
accuracy
91
%
allowed
for
clear
delineation
areas
their
distinct
topographic
features
particularly
red
red-edge
bands.
In
second
step,
remaining
area
(60.30
%)
differentiated
Iterative
Self-Organizing
cluster
unsupervised
yield
five-class
map
at
10
m
spatial
resolution.
The
predicted
successfully
validated
by
independent
Caesium-137
(137Cs)
organic
carbon
observations
catchment
regional
scales,
as
revealed
significant
inter-class
differences
rates
estimated
137Cs
inventory.
class
had
loss
rate
5.5
mm
yr−1,
suggesting
previous
assessments
have
underestimated
severity.
accordance
crop
growth
intensity,
localized
settings,
highlighted
imaging
spatiotemporal
development
its
response
targeted
sustainable
cropland
management
efforts.