Journal of Physics Conference Series,
Journal Year:
2023,
Volume and Issue:
2676(1), P. 012004 - 012004
Published: Dec. 1, 2023
Abstract
This
paper
presents
the
development
and
evaluation
of
an
Artificial
Neural
Network
(ANN)
based
on
model
for
predicting
salinity
Warta
River.
The
study
focused
prediction
river
water
salinity,
expressed
in
terms
electrical
conductivity
(EC),
using
proposed
ANN
structure
7-10-1.
network
showed
a
satisfactory
ability
to
capture
interrelationships
between
input
data:
sulphates,
chlorides,
calcium,
magnesium,
total
hardness,
pH,
dissolved
solids.
correlation
coefficient
(R)
values
training,
validation
test
sets
were
0.99444,
0.96988
0.97174,
respectively.
From
results,
it
can
be
concluded
that
developed
is
suitable
EC
river.
Frontiers in Freshwater Science,
Journal Year:
2025,
Volume and Issue:
3
Published: April 14, 2025
Identifying
commonalities
in
how
fish
navigate
rivers
near
infrastructure
will
enhance
water
operations
and
design
by
improving
our
ability
to
predict
engineering
outcomes
(e.g.,
barrier
construction/removal,
passage
installation)
novel
settings
before
the
cost
of
real-world
implementation.
Evidence
from
intermediate-scale
computer
models
(time
scales
minutes
days
spatial
<2
km)
suggests
that
movement
behavior
is
frequently
governed
responses
one
or
more
following
hydrodynamic
features:
(1)
flow
direction
(i.e.,
rheotaxis),
(2)
velocity
magnitude,
(3)
turbulence,
(4)
depth,
plus
(5)
integration
information
over
recent
time
periods
memory/experience).
However,
lack
consistent
modeling
approaches,
infrequent
assessment
each
response
isolation
combination,
a
focus
on
limited
number
species
means
generality
these
uncertain.
We
use
model,
specifically
pattern-oriented
approach
incorporating
individual
based
(IBMs),
apply
four
features
memory/experience
different
combinations
study
their
value
for
reproducing
an
infrequently
modeled
lifestage,
upriver
migrating
adult
sea
lamprey,
Petromyzon
marinus
.
The
site
was
region
downstream
Sault
Ste.
Marie
lock
dam
complex
located
between
Canada
U.S.A
St.
Marys
River
joining
Lake
Superior
Huron.
Our
analysis
indicates
rheotaxis
magnitude
as
well
past
experience
improve
lamprey
spatio-temporal
prediction
compared
other,
simpler
forms
behavior.
Sea
also
biased
toward
lower
levels
turbulence
turbulent
kinetic
energy)
its
precursor
gradient
speed).
A
depth
not
found
be
important,
but
domain
two-dimensional
which
assessment.
As
similar
are
very
fish,
appear
underlie
river
navigation
across
range
life
stages
share
goal-oriented
downriver
movement.
systematic
highlights
accuracy
trade-offs
response,
individually
often
accompany
alternative
behavioral
formulations
model
structure
provides
framework
future
findings
analyses
additional
contexts
can
added.
Journal of Geophysical Research Oceans,
Journal Year:
2025,
Volume and Issue:
130(4)
Published: April 1, 2025
Abstract
The
Rhine‐Meuse
Delta
is
a
low‐lying
delta
in
the
Netherlands
that
subject
to
both
salt
intrusion
events
and
storm
surges.
Typically,
surges
only
temporarily
cause
increased
do
not
severe
problems
for
freshwater
availability.
However,
during
surge
of
December
2013,
reached
closed
southern
branch
higher
salinities
were
observed
weeks
after
surge.
purpose
this
study
examine
mechanisms
controlling
event.
A
three‐dimensional
hydrodynamic
model
(Delft3D‐FM)
was
developed
successfully
reproduces
normal
conditions.
During
storm,
high
water
levels
northern
caused
flux
toward
branch.
off
by
an
estuarine
dam,
consequently
retained
landward
dam.
Local
stratification
remain
deeper
parts,
limiting
effectiveness
flushing
In
post‐storm
period,
gradually
released
from
branch,
raising
salinity
adjacent
channel.
river
discharge
just
below
yearly
average,
showing
prolonged
can
also
occur
outside
dry
periods.
Journal of Geophysical Research Biogeosciences,
Journal Year:
2024,
Volume and Issue:
129(4)
Published: April 1, 2024
Abstract
Wetland
ecosystems
hold
nearly
a
third
of
the
global
soil
carbon
pool,
but
as
wetlands
rapidly
disappear
fate
this
stored
is
unclear.
The
aim
study
was
to
quantify
and
then
link
potential
rates
microbial
decomposition
after
vertical
drowning
vegetated
tidal
marshes
in
coastal
Louisiana
known
drivers
anaerobic
altered
by
vegetation
loss.
Profiles
CH
4
CO
2
production
(surface
60
cm
deep)
were
measured
during
incubations,
organic
matter
chemistry
assessed
with
infrared
spectroscopy,
porewater
nutrients
redox
potentials
field
along
chronosequence
wetland
After
drowning,
pond
soils
had
lower
potentials,
higher
pH
values,
nitrogen
concentrations,
lignin:
polysaccharide
ratios,
more
NH
+
PO
3−
,
release
than
marsh
soils.
Potential
similar
open
water
ponds,
depth‐dependent
decreases
concentrations
increased.
In
these
anoxic
soils,
loss
exerts
primary
control
on
because
flooding
drives
sustained
increases
nutrient
availability
(NH
3
dissolved
carbon)
(from
−150
−500
mV)
that
lead
fluxes
within
few
years.
Without
new
inputs
following
loss,
ponds
may
losses
could
influence
budgets.
Journal of Fish Biology,
Journal Year:
2024,
Volume and Issue:
105(2), P. 459 - 471
Published: July 4, 2024
Estuaries
are
essential
habitats
for
recreational
and
commercial
fish
that
shaped
by
both
natural
anthropogenic
processes.
In
Louisiana
a
combination
of
climate
change
planned
coastal
restoration
actions
is
predicted
to
increase
freshwater
introduction
estuaries.
As
such
there
need
quantify
the
relationships
between
estuarine
ecology
salinity
aid
in
predicting
how
species
will
respond
shifts
salinity.
We
investigated
relative
abundance
dietary
niches
adult
(24.5
±
5.4
cm
standard
length)
spotted
seatrout
Cynoscion
nebulosus
across
varying
regimes
(oligohaline,
mesohaline,
polyhaline)
within
Barataria
Bay,
Louisiana,
using
net
sampling
gut
content
stable
isotopes
analysis.
found
C.
was
lowest
at
oligohaline
site,
translating
approximately
five
fewer
captured
every
single
psu
decrease
site's
average
annual
contrast,
we
diets
and,
lesser
extent,
isotopic
had
high
degree
overlap
sites
with
differing
regimes.
Fish
penaeid
shrimp
were
most
common
important
prey
taxa
recovered
from
guts
all
sites.
The
small
differences
among
likely
due
spatial
variation
hydrogeochemical
baselines,
observed
provides
support
idea
move
adjacent
forage
throughout
Bay.
Our
results
contribute
greater
understanding
preference
trophic
can
their
responses
future
habitat
changes
Bay
associated
actions.
Water Resources Research,
Journal Year:
2024,
Volume and Issue:
60(10)
Published: Oct. 1, 2024
Abstract
Transformer
neural
networks
(TNNs)
have
caused
a
paradigm
shift
in
deep
learning
domains
like
natural
language
processing,
gathering
immense
interest
due
to
their
versatility
other
fields
such
as
time
series
forecasting
(TSF).
Most
current
TSF
applications
of
TNNs
use
only
historic
observations
predict
future
events,
ignoring
information
available
weather
forecasts
inform
better
predictions,
and
with
little
attention
given
the
interpretability
model's
explanatory
inputs.
This
work
explores
potential
for
perform
across
multiple
environmental
variables
(streamflow,
stage,
water
temperature,
salinity)
two
ecologically
important
regions:
Peace
River
watershed
(Florida)
northern
Gulf
Mexico
(Louisiana).
The
TNN
was
tested
its
prediction
uncertainty
quantified
each
response
variable
from
one‐to
fourteen‐day‐ahead
using
past
spatially
distributed
forecasts.
A
sensitivity
analysis
(SA)
performed
on
trained
TNNs'
weights
identify
relative
influence
input
windows.
Overall
model
performance
ranged
good
very
(0.78
<
NSE
0.99
all
forecast
horizons).
Through
SA,
we
found
that
able
learn
physical
patterns
behind
data,
adapt
forecast,
increasingly
windows
increased.
TNN's
excellent
flexibility,
along
intuitive
highlighting
logic
models'
decision‐making
process,
provide
evidence
applicability
this
architecture
locations.