Water Resources Research,
Journal Year:
2023,
Volume and Issue:
59(3)
Published: Feb. 20, 2023
Abstract
Three‐dimensional
(3d)
numerical
models
are
state‐of‐the‐art
for
investigating
complex
hydrodynamic
flow
patterns
in
reservoirs
and
lakes.
Such
full‐complexity
computationally
demanding
their
calibration
is
challenging
regarding
time,
subjective
decision‐making,
measurement
data
availability.
In
addition,
physically
unrealistic
model
assumptions
or
combinations
of
parameters
may
remain
undetected
lead
to
overfitting.
this
study,
we
investigate
if
how
so‐called
Bayesian
aids
characterizing
faulty
setups
driven
by
parameter
combinations.
builds
on
recent
developments
machine
learning
uses
a
Gaussian
process
emulator
as
surrogate
model,
which
runs
considerably
faster
than
3d
model.
We
Bayesian‐calibrate
Delft3D‐FLOW
pump‐storage
reservoir
function
the
background
horizontal
eddy
viscosity
diffusivity,
initial
water
temperature
profile.
consider
three
scenarios
with
varying
degrees
different
velocity
measurements.
One
forces
completely
unrealistic,
rapid
lake
stratification
still
yields
similarly
good
accuracy
more
correct
global
statistics,
such
root‐mean‐square
error.
An
uncertainty
assessment
resulting
from
indicates
that
scenario
fast
through
highly
uncertain
mixing‐related
parameters.
Thus,
describes
quality
correctness
geometric
characteristics
posterior
distributions.
For
instance,
most
likely
values
(posterior
distribution
maxima)
at
range
limit
widespread
characterize
poor
calibration.
Advances in Artificial Neural Systems,
Journal Year:
2015,
Volume and Issue:
2015, P. 1 - 12
Published: June 9, 2015
In
this
study,
two
artificial
neural
network
models
(i.e.,
a
radial
basis
function
network,
RBFN,
and
an
adaptive
neurofuzzy
inference
system
approach,
ANFIS)
multilinear
regression
(MLR)
model
were
developed
to
simulate
the
DO,
TP,
Chl
,
SD
in
Mingder
Reservoir
of
central
Taiwan.
The
input
variables
MLR
determined
using
linear
regression.
performances
evaluated
ANFIS,
based
on
statistical
errors,
including
mean
absolute
error,
root
square
correlation
coefficient,
computed
from
measured
model-simulated
values.
results
indicate
that
performance
ANFIS
is
superior
those
RBFN
models.
study
show
suitable
for
simulating
water
quality
with
reasonable
accuracy,
suggesting
can
be
used
as
valuable
tool
reservoir
management
Freshwater Science,
Journal Year:
2013,
Volume and Issue:
32(1), P. 39 - 55
Published: Jan. 28, 2013
Temperature
is
a
primary
driver
of
the
structure
and
function
stream
ecosystems.
However,
lack
temperature
(ST)
data
for
vast
majority
streams
rivers
severely
compromises
our
ability
to
describe
patterns
thermal
variation
among
streams,
test
hypotheses
regarding
effects
on
macroecological
patterns,
assess
altered
STs
ecological
resources.
Our
goal
was
develop
empirical
models
that
could:
1)
quantify
watershed
alteration
(SWA)
STs,
2)
accurately
precisely
predict
natural
(i.e.,
reference
condition)
in
conterminous
USA
rivers.
We
modeled
3
ecologically
important
elements
regime:
mean
summer,
winter,
annual
ST.
To
build
reference-condition
(RCMs),
we
used
daily
ST
obtained
from
several
thousand
US
Geological
Survey
sites
distributed
across
iteratively
with
Random
Forests
identify
condition.
first
created
set
dirty
(DMs)
related
both
factors
(e.g.,
climate,
area,
topography)
measures
SWA,
i.e.,
reservoirs,
urbanization,
agriculture.
The
performed
well
(r2
=
0.84–0.94,
residual
square
error
[RMSE]
1.2–2.0°C).
For
each
DM,
partial
dependence
plots
SWA
thresholds
below
which
response
minimal.
then
only
upstream
these
RCMs
as
predictors
0.87–0.95,
RMSE
1.1–1.9°C).
Use
reference-quality
caused
suffer
modest
loss
predictor
space
spatial
coverage,
but
this
associated
parts
curves
were
flat
and,
therefore,
not
responsive
further
space.
compared
predictions
made
DMs
0.
most
DMs,
setting
SWAs
0
resulted
biased
estimates
Water,
Journal Year:
2024,
Volume and Issue:
16(4), P. 514 - 514
Published: Feb. 6, 2024
In
regions
where
drought
has
become
a
common
occurrence
for
most
of
the
year
and
agriculture
is
main
economic
activity,
development
hydro-agricultural
systems
made
it
possible
to
improve
water
management.
Despite
this,
intensification
combined
with
climate
change
leads
potential
decrease
in
quality
management
practices
are
essential
agro-environmental
sustainability.
The
aim
this
study
was
assess
irrigation
ecological
status
reservoir
(using
support
chemical
parameters).
results
showed
biological
oxygen
demand
values
above
maximum
stipulated
an
excellent
all
sampling
periods
except
April
2018
December
2020
(with
highest
10
mg
L−1
O2
dry
periods).
Most
total
nitrogen
concentrations
(TN)
surpassed
those
good
(0.96
≤
TN
2.44
N).
fact,
suspended
solids
were
parameters
used
classification.
From
perspective
according
FAO
guidelines
regarding
infiltration
rate,
these
waters
presented
light
moderate
levels
restrictions.
Thus,
revealed
that
its
impact
on
soil
rate
can
be
related,
part,
meteorological
conditions
intensive
agricultural
developed
around
drainage
basin.
that,
as
Lage
part
Brinches–Enxoé
hydraulic
circuit,
recirculation
also
important
factor
may
have
affected
obtained.
Furthermore,
experimental
design,
integrating
status,
parameters,
systems;
using
same
from
different
perspectives;
allowed
us
global
idea
contamination
agroecosystems,
improving
river
basin
processes.
