Water,
Год журнала:
2024,
Номер
16(4), С. 553 - 553
Опубликована: Фев. 11, 2024
Remote
sensing
technology
applications
for
water
quality
inversion
in
large
rivers
are
common.
However,
their
application
to
medium/small-sized
bodies
within
rural
areas
is
limited
due
the
low
spatial
resolution
of
remote
images.
In
this
work,
a
typical
small
river
was
selected,
and
high-resolution
unmanned
aerial
vehicle
(UAV)
multispectral
images
ground
monitoring
data
were
obtained.
Then,
comparative
analysis
three
univariate
regression
models
nine
machine
learning
(Ridge
Regression
(RR),
Support
Vector
(SVR),
Grid
Search
(GS-SVR),
Random
Forest
(RF),
(GS-RF),
eXtreme
Gradient
Boosting
(XGBoost),
Deep
Neural
Networks
(DNN),
Convolutional
(CNN),
Catboost
(CBR))
accuracy
prediction
turbidity
(TUB),
total
nitrogen
(TN),
phosphorus
(TP)
performed.
TUB
can
be
achieved
by
simple
statistical
models.
The
CBR
model
exhibited
best
performance
index
inversions
on
test
set
evaluation
metrics:
R2
(0.90~0.92),
RMSE
(7.57
×
10−3~1.59
mg/L),
MAE
(0.01~1.30
RPD
(3.21~3.56),
NSE
(0.84~0.92).
pollution
study
area
closely
related
its
land-use
pattern,
excessive
irrational
fertilizer
application,
distribution
pollutant
outlets.
