Flow intermittence prediction using a hybrid hydrological modelling approach: influence of observed intermittence data on the training of a random forest model
Hydrology and earth system sciences,
Год журнала:
2024,
Номер
28(4), С. 851 - 871
Опубликована: Фев. 23, 2024
Abstract.
Rivers
are
rich
in
biodiversity
and
act
as
ecological
corridors
for
plant
animal
species.
With
climate
change
increasing
anthropogenic
water
demand,
more
frequent
prolonged
periods
of
drying
river
systems
expected,
endangering
ecosystems.
However,
understanding
predicting
the
hydrological
mechanisms
that
control
periodic
rewetting
rivers
is
challenging
due
to
a
lack
studies
observations,
particularly
non-perennial
rivers.
Within
framework
Horizon
2020
DRYvER
(Drying
River
Networks
Climate
Change)
project,
modelling
study
flow
intermittence
being
carried
out
three
European
catchments
(Spain,
Finland,
France)
characterised
by
different
climate,
geology,
use.
The
objective
this
represent
spatio-temporal
dynamics
at
reach
level
mesoscale
networks
(between
120
350
km2).
daily
spatially
distributed
condition
(flowing
or
dry)
predicted
using
J2000
model
coupled
with
random
forest
classification
model.
Observed
data
from
sources
(water
measurements,
photo
traps,
citizen
science
applications)
used
build
predictive
This
aims
evaluate
impact
observed
dataset
(sample
size,
spatial
temporal
representativity)
on
performance
Results
show
hybrid
approach
developed
allows
patterns
be
accurately
catchments,
sensitivity
criterion
above
0.9
prediction
dry
events
Finnish
French
case
0.65
Spanish
study.
shows
value
combining
reduce
uncertainty
intermittence.
Язык: Английский
How low can you go? Widespread challenges in measuring low stream discharge and a path forward
Limnology and Oceanography Letters,
Год журнала:
2023,
Номер
8(6), С. 804 - 811
Опубликована: Сен. 30, 2023
Low
flows
pose
unique
challenges
for
accurately
quantifying
streamflow.
Current
field
methods
are
not
optimized
to
measure
these
conditions,
which
in
turn,
limits
research
and
management.
In
this
essay,
we
argue
that
the
lack
of
measuring
low
streamflow
is
a
fundamental
challenge
must
be
addressed
ensure
sustainable
water
management
now
into
future,
particularly
as
climate
change
shifts
more
streams
increasingly
frequent
flows.
We
demonstrate
pervasive
flows,
present
decision
support
tool
(DST)
navigating
best
practices
highlight
important
method
developmental
needs.
Water
resource
facing
mounting
associated
with
scarcity,
including
interactive
effects
changing
increased
demand
(Craig
et
al.
2017).
Climate
increasing
drought
severity
many
regions
(Cook
2020),
while
limited
supplies
depletes
resources
(de
Graaf
2019).
Combined,
stressors
result
lower
variable
rivers
(Zipper
2021),
arid
(Hammond
2021).
Despite
posed
by
low-flow
majority
(e.g.,
time,
funding)
monitoring
have
historically
focused
on
high-water
concerns,
such
ensuring
navigation
predicting
floods
(Vörösmarty
2001;
Ruhi
2018),
larger,
perennially-flowing
systems
(Krabbenhoft
2022).
Low-flow
conditions
(Mauger
define
or
little
downstream
surface
flow
caused
small
volumes
very
velocities
(i.e.,
slackwater),
prevalent
thus
necessitate
greater
focus
quantification
approaches.
Streamflow
underlying
physical
template
structuring
biotic
abiotic
processes,
biogeochemical
cycling,
ecological
communities
river
systems;
thus,
inaccurate
measurements
can
propagate
hinder
diverse
analyses
requiring
accurate
data,
ranging
from
characterization
2022),
environmental
allocations
(Neachell
Petts
2019),
function
assessments
(Leigh
Datry
2017),
species
conservation
plans
(Lopez
forecasting
(Forzieri
2014).
posit
measurement
techniques
leaves
networks
ill-equipped
inform
management,
future.
Our
objectives
to:
(1)
widespread
across
an
existing
network
United
States,
(2)
discuss
limitations
current
(3)
DST
choosing
among
methods,
(4)
methodological
developments
needed
improve
monitoring.
Such
progress
prerequisite
understanding
how
will
respond
human
demands,
thereby
supporting
policy
actions
seeking
avoid
minimize
impacts.
Point
essential
short-
long-term
studies
monitoring,
made
using
different
(Turnipseed
Sauer
2010).
If
conducted
over
range
discrete
used
develop
rating
curve
relates
stage
discharge,
allowing
long-term,
continuous
discharge
via
sensors
our
analysis
discussion
point
streamflow,
but
emphasize
approaches
implications
accuracy
longer-term
development.
To
quantify
prevalence
substandard
measurements,
examined
manual
8008
U.S.
Geological
Survey
(USGS)
gages
continental
States
GAGES
II
dataset
(Falcone
2011),
sites
either
20+
years
since
1950
were
operational
2009
(Appendix
S1).
For
each
measurement,
collected
quality
code
assigned
USGS
hydrographers
immediately
after
making
measurement:
"Poor"
when
uncertainty
estimated
above
8%,
"fair"
less
than
good
5%,
excellent
2%
These
codes
qualitative
estimating
individual
based
suitability
channel
cross-section,
state,
other
gage,
identified
minimum
value
"good"
calculated
percent
gage's
daily
record
below
threshold.
results
overly
sensitive
threshold,
also
compared
percentage
two
additional
thresholds:
corresponding
average
thresholds
(see
Table
S1
details),
obtained
comparable
results.
The
"minimum
good"
metric
provides
conservative
estimate
duration
high
site;
it
only
considers
related
does
account
stemming
conditions.
interrogated
because
represents
standard
investigators
use
benchmark,
provided
large
relating
quality/uncertainty.
performed
all
R
version
4.2.1
(R
Core
Team
2022)
data
National
Information
System
DataRetrieval
Package
(De
Cicco
Across
network,
records
was
8.4%,
indicating
overall
measurements.
However,
found
393
(~
5.5%)
had
at
least
50%
value,
68
95%
threshold
(Fig.
1A).
Sites
widely
distributed
climatic
zones,
land
uses,
hydrologic
settings,
although
greatest
density
concentrated
southwestern
where
issues
linked
scarcity
(Brown
provide
example
difficulties
gage
Kings
Creek
near
Manhattan
KS
(USGS
Gage
06879650),
well-studied,
grassland
stream
long
(1979–present).
Only
73
238
31%)
considered
"excellent"
1B).
relatively
incidence
resulted
58.6%
(from
1980
2021)
being
lowest
proportion
given
year
2.5%
100%.
This
underscores
even
site,
relative
importance
vary
year-to-year,
impact
during
dry
1C).
Furthermore,
uncertainties
may
subsequent
estimates
nutrient
export,
lead
some
annual
load
much
certain
others.
Systems
flashy
face
highly
uncertain
end
curve,
leading
sources
uncertainty.
While
sensitivity
propagation
beyond
scope
paper,
highlights
areas
poorly
suited
capture
Three
general
categories
comprise
toolbox
available
most
practitioners.
include:
velocity-area
methods;
tracer-based
salt
dye;
known
streambed
geometry
flume
weir)
capturing
constriction
(WMO
Most
tend
unusable
under
(Hamilton
2008)
three
reasons:
and/or
shallow
depths
2A,B),
mobile
streambeds
irregular
channels
2D,E),
proportions
subsurface
2C–E).
Many
transition
visible
slow
imperceptible
movement
water,
sometimes
spatially
discontinuous
pooled.
poor
tracer
mixing
recovery
dilution
gaging
2A).
High
width-to-depth
ratios
wide
water)
inability
fully
submerge
velocimeters
2E).
bed
elevations
rocks
boulders)
emergent
vegetation
further
reduce
velocity
render
them
impossible
2D).
Finally,
surface-water
therefore
directly
estimates,
flow.
relevant
often
exhibit
hyporheic
problems
mutually
exclusive;
indeed,
multiple
arise
leaving
practitioners
unsure
about
considerable
Given
challenges,
reflects
collective
experience
working
systems,
describes
approach
applying
complicating
factors
dominate
3).
aim
offer
guidance
systematic
way
apply
consistent
complex
systems.
assumes
chosen
location
site
there
no
better
within
reasonable
distance
upstream
downstream)
what
should
avoided
selection.
intended
data-driven
study
optimal
rather
offering
informed
opinions
work
specific
contexts
experts
who
frequently
attempt
non-ideal
compiling
DST,
development
prioritized,
hope
catalyzes
advances
community.
initial
bifurcation
separates
whether
visibly
flowing
material
leaves)
observed
moving
downstream.
movement,
fewer
options
exist
visible,
prompts
series
questions
regarding
cross-section
depth
help
identify
suitable
their
acknowledge
pathways
nodes
equally
likely
encountered.
example,
few
locations
natural
points
bucket
2F),
though
appears
twice
terminate
"no
methods."
experience,
(numbering
dozens,
examples
Fig.
2)
fall
characterized
method"
part
year,
us
unable
fluxes
limiting
like
flux
estimates.
possible
recommended
suboptimal
error.
Selecting
requires
degree
precision
consider
trade-offs
between
costs.
studies,
parameters
easier
measure—like
depth,
wetted
width/area,
approximate
state—may
sufficient
(Jaeger
2023).
contrast,
biogeochemistry
key
calculating
loads
(Gómez-Gener
require
aquatic
habitat.
Other
trade-offs,
personnel
costs,
frequency,
time
conduct
outweigh
scientific
considerations
3.
At
take
hours
days,
minutes
hour
required
moderate
addition,
non-optimal
breakthrough
curves
incomplete
Portable
flumes/weirs
faster
implement
modifying
channel,
manually
creating
berms
concentrate
through
2C),
reasons.
recommendations
modifications
accommodate
variations
application
gaging;
suggestions
situations
desirable,
those
S1.
rivers,
communities.
Discharge
assess
connectivity
tributaries
solutes
network.
Time
inputs
models
ecosystem
desired
output
identifying
driving
responses
anthropogenic
change.
All
applications
full
variability.
universal
answer
question
"what
error
acceptable
flows,"
need
clear.
Although
absolute
changes
0.01
0.02
m3/s),
system
(100%).
Small
substantial
consequences
habitat
extent
(Rolls
2012).
Detection
trends
hampered
imprecise
cause
vulnerable
go
unquantified
(Whitfield
Hendrata
2006).
Environmental
regulations
precise
enforcement
complicate
implementation
enforcement.
large,
difficult-to-measure
norm
prevent
flow-duration
minimal
purposes.
represent
smaller
component
solute
they
critical
hydrological,
ecological,
dynamics
without
robust
addition
providing
systematically
deciding
employ
determining
3)
assessment.
cases,
modification
optimization
entirely
new
developed
refined,
as:
slackwater
pools
2A);
wide,
shallow,
irregular,
threaded
2E),
opportunity
modification;
reaches
dense
vegetation;
wind
strongly
affects
velocities.
commonly
freshwaters
share
similarities
coastal
opening
up
potential
transfer
to/from
hydrology
Birgand
There
promising
recent
technological
micro
(Osorno
time-lapse
imagery
trail
cameras
videos
(Birgand
2022;
Chapman
Dolcetti
radar
altimetry
(Bandini
presence/absence
(Chapin
2014;).
Emerging
tools
spatiotemporal
variation
state
assessment
presence
absence
done
advance
presence/absence,
unresolved.
settings
modeling
mathematical
relationship
option
(Gao
suggest
effort
urgently
needed,
numerous
viable
Methods
accelerates,
variability
around
world.
manage
future
managers
achieve
this,
flexibility
extreme
Without
improvements,
able
sustain
predict
continuing
trajectories
Understanding
managing
integrity,
promoting
quality,
safeguarding
access.
first
step
high-quality
Data
Supplementary
Information.
Please
note:
publisher
responsible
content
functionality
any
information
supplied
authors.
Any
queries
(other
missing
content)
directed
author
article.
supported
NSF-DEB
Grant
#1754389
Dry
Rivers
Research
Coordination
Network
NSF-IOA
#2019603
Aquatic
Intermittency
Microbiomes
Streams
(AIMS)
project.
would
thank
members
AIMS
team
thoughtful
conversations
contributions
helped
ideas
explored
manuscript.
Язык: Английский
Spatial stream network modeling of water temperature within the White River Basin, Mount Rainier National Park, Washington
Scientific investigations report,
Год журнала:
2025,
Номер
unknown
Опубликована: Янв. 1, 2025
Язык: Английский
Citizen scientists can help advance the science and management of intermittent rivers and ephemeral streams
BioScience,
Год журнала:
2023,
Номер
73(7), С. 513 - 521
Опубликована: Июль 1, 2023
Abstract
Intermittent
rivers
and
ephemeral
streams
are
the
world's
dominant
type
of
river
ecosystem
becoming
more
common
because
global
change.
However,
inclusion
intermittent
in
water
policies
management
plans
remains
largely
limited
monitoring
schemes
tools
designed
for
perennial
rivers.
In
present
article,
we
discuss
how
smartphone
applications
used
by
citizen
scientists
can
quantify
extent
occurrence
streams.
We
also
introduce
a
new
app,
DRYRivERS,
specifically
to
monitor
After
year
use,
counted
than
3600
observations
from
1900
reaches
across
19
countries
four
continents.
Through
three
case
studies,
then
show
that
science
improve
our
knowledge
prevalence
landscape,
enhance
hydrological
modeling
calibration,
guide
managers
setting
abstraction
restrictions.
Together,
approach
demonstrates
be
incorporated
into
environmental
better
inform
policy.
Язык: Английский
Intersection of machine learning and mobile crowdsourcing: a systematic topic-driven review
Personal and Ubiquitous Computing,
Год журнала:
2024,
Номер
unknown
Опубликована: Июнь 10, 2024
Язык: Английский
The importance of source data in river network connectivity modeling: A review
Limnology and Oceanography,
Год журнала:
2024,
Номер
unknown
Опубликована: Окт. 7, 2024
Abstract
River
network
connectivity
(RC)
describes
the
hydrologic
exchange
of
water,
nutrients,
sediments,
and
pollutants
between
river
channel
other
“sites”
via
heterogenous
flowpaths
along
corridor.
As
water
moves
downstream
it
carries
these
constituents,
creating
a
stream‐to‐ocean
continuum
that
regulates
global
carbon,
nutrient
cycling.
models
have
developed
over
many
decades,
culminating
in
recent
years
with
network‐scale
RC
explicitly
simulate
transport
elements
from
headwaters
to
coasts,
sometimes
requiring
contain
tens
millions
reaches.
These
advances
provide
transformative
insights
into
aggregate
effects
on
material
across
scales
local
global.
Yet,
reviews
pointed
several
challenges
need
be
overcome
continue
advancing
modeling.
In
service
goals,
I
summarize
maps
identify
similarities
differences
large‐scale
modeling
landscape.
Although
our
computational
upscaling
abilities
significantly
improved
revealed
new
insights,
current
are
still
limited
by
quantity,
quality,
resolution,
lack
standardization
available
situ
databases
source
data
necessary
for
This
suggests
we
can
extend
if
keep
improving
datasets,
while
continuously
revisiting
physics
theory
explain
those
data.
doing
so,
will
expand
role
informing
quality
management
future.
Язык: Английский
Spatial Variability of Water Temperature within the White River Basin, Mount Rainier National Park, Washington
EarthArXiv (California Digital Library),
Год журнала:
2024,
Номер
unknown
Опубликована: Авг. 16, 2024
Water
temperature
is
a
primary
control
on
the
occurrence
and
distribution
of
cold-water
species.
Rivers
draining
Mount
Rainier
in
western
Washington,
including
White
River
along
its
northern
flank,
support
several
fish
populations,
but
spatial
water
temperatures,
particularly
during
late-summer
base
flow
between
August
September,
climatic,
hydrologic,
physical
processes
regulating
this
are
not
well
understood.
Spatial
stream
network
(SSN)
models,
which
generalized
linear
models
that
incorporate
streamwise
autocovariance
structures,
were
fit
to
mean
seven-day
average
daily
maximum
for
September
basin
located
with
National
Park.
The
SSN
calibrated
using
measurements
collected
2010
2020.
Significant
covariates
within
best-fit
included
proportion
ice
cover
forest
basin,
air
temperature,
consolidated
geologic
units,
snow
equivalent.
Statistical
structures
had
better
predictive
performance
than
those
did
not.
In
addition,
September.
Predictions
similar
general
warming
downstream
part
main-stem
compared
cooler
temperatures
high-elevation
headwater
streams.
Estimated
upper
model
three
four
degrees
Celsius
warmer
tributaries
one
two
main
stem
regional-scale
model.
Differences
attributed
observations
specific
River,
whereas
from
lower
elevation
streams
Rainer
Park
boundary
used
Язык: Английский
Zero‐Flow Dynamics for Headwater Streams in a Humid Forested Landscape
Hydrological Processes,
Год журнала:
2024,
Номер
38(12)
Опубликована: Дек. 1, 2024
ABSTRACT
Much
of
our
understanding
on
temporary
headwater
streams
is
from
arid
and
sub‐humid
environments.
We
know
less
about
zero‐flow
periods
in
humid
catchments
that
experience
seasonal
snow
cover.
Our
study
characterised
the
temporal
spatial
patterns
for
forested
a
snow‐dominated
landscape.
used
36
years
streamflow
data
13
within
Turkey
Lakes
Watershed
located
Canadian
Shield
Ontario,
Canada,
near
eastern
shores
Lake
Superior.
These
differ
substantially
their
number
May–November
days
(0–166
per
year)
despite
being
clustered
small
geographical
area
with
similar
geology,
physiography
vegetation
The
also
continental
climatic
conditions
relatively
even
precipitation
inputs
throughout
year
(mean
annual
1210
mm/year).
Inter‐annual
variability
was
primarily
associated
evapotranspiration.
Despite
large
snowpacks
form
this
region,
amount
did
not
appear
to
influence
extent
periods.
found
between‐catchment
occurrences
related
differences
catchment
properties
typically
greater
groundwater
influence.
suggests
zero‐flows
can
be
highly
variable
over
regions
flow
permanence
may
more
sensitive
spring
fall
weather
than
due
partly
shallow
soils
Shield.
Язык: Английский