Abstract.
Accurate
hydrological
modeling
is
vital
to
characterizing
how
the
terrestrial
water
cycle
responds
climate
change.
Pure
deep
learning
(DL)
models
have
shown
outperform
process-based
ones
while
remaining
difficult
interpret.
More
recently,
differentiable,
physics-informed
machine
with
a
physical
backbone
can
systematically
integrate
equations
and
DL,
predicting
untrained
variables
processes
high
performance.
However,
it
was
unclear
if
such
are
competitive
for
global-scale
applications
simple
backbone.
Therefore,
we
use
–
first
time
at
this
scale
differentiable
hydrologic
(fullname
δHBV-globe1.0-hydroDL
shorthanded
δHBV)
simulate
rainfall-runoff
3753
basins
around
world.
Moreover,
compare
δHBV
purely
data-driven
long
short-term
memory
(LSTM)
model
examine
their
strengths
limitations.
Both
LSTM
provide
competent
daily
simulation
capabilities
in
global
basins,
median
Kling-Gupta
efficiency
values
close
or
higher
than
0.7
(and
0.78
subset
of
1675
long-term
records),
significantly
outperforming
traditional
models.
regionalized
demonstrated
stronger
spatial
generalization
ability
(median
KGE
0.64)
parameter
regionalization
approach
0.46)
even
ungauged
region
tests
Europe
South
America.
Nevertheless,
relative
LSTM,
hampered
by
structural
deficiencies
cold
polar
regions,
highly
arid
significant
human
impacts.
This
study
also
sets
benchmark
estimates
world
builds
foundations
improving
simulations.
Environmental Research Letters,
Journal Year:
2024,
Volume and Issue:
19(7), P. 074005 - 074005
Published: May 31, 2024
Abstract
Global
hydrological
models
(GHMs)
are
widely
used
to
assess
the
impact
of
climate
change
on
streamflow,
floods,
and
droughts.
For
‘model
evaluation
attribution’
part
current
round
Inter-Sectoral
Impact
Model
Intercomparison
Project
(ISIMIP3a),
modelling
teams
generated
historical
simulations
based
observed
direct
human
forcings
with
updated
model
versions.
Here
we
provide
a
comprehensive
daily
maximum
annual
discharge
ISIMIP3a
from
nine
GHMs
by
comparing
observational
data
644
river
gauge
stations.
We
also
low
flows
effects
different
routing
schemes.
find
that
can
reproduce
variability
in
discharge,
but
tend
overestimate
both
quantities,
as
well
flows.
Models
perform
better
at
stations
wetter
areas
lower
elevations.
Discharge
routed
CaMa-Flood
improve
performance
some
models,
for
others,
is
overestimated,
leading
reduced
performance.
This
study
indicates
future
development
include
improving
simulation
processes
arid
regions
cold
dynamics
high
further
suggest
studies
attributing
changes
using
ensemble
will
be
most
meaningful
humid
areas,
elevations,
places
regular
seasonal
these
where
underlying
seem
best
represented.
Remote Sensing,
Journal Year:
2024,
Volume and Issue:
16(3), P. 550 - 550
Published: Jan. 31, 2024
The
land
surface
model
(LSM)
is
extensively
utilized
to
simulate
terrestrial
processes
between
and
atmosphere
in
the
Earth
system.
Hydrology
simulation
key
component
of
model,
which
can
directly
reflect
capability
LSM.
In
this
study,
three
offline
LSM
simulations
were
conducted
over
China
using
Community
Land
Model
version
5.0
(CLM5)
driven
by
different
meteorological
forcing
datasets,
namely
Meteorological
Forcing
Dataset
(CMFD),
Global
Soil
Wetness
Project
Phase
3
(GSWP3),
bias-adjusted
ERA5
reanalysis
(WFDE5),
respectively.
Both
gridded
situ
reference
data,
including
evapotranspiration
(ET),
soil
moisture
(SM),
runoff,
employed
evaluate
performance
levels
CLM5-based
across
its
ten
basins.
general,
all
realistically
replicate
magnitudes,
spatial
patterns,
seasonal
cycles
ET
when
compared
with
remote-sensing-based
observations.
Among
basins,
Yellow
River
Basin
(YRB)
basin
where
are
best,
supported
higher
KGE
value
0.79.
However,
substantial
biases
occur
Northwest
Rivers
(NWRB)
significant
overestimation
for
CMFD
WFDE5
underestimation
GSWP3.
addition,
both
grid-based
or
site-based
evaluations
SM
indicate
that
systematic
wet
exist
CLM5
shallower
layer
nine
basins
China.
Comparatively,
simulating
deeper
slightly
better.
Moreover,
types
reasonable
runoff
among
capture
more
detailed
information,
but
GSWP3
presents
comparable
change
trends
data.
summary,
study
explored
capacity
assessment
results
may
provide
important
insights
future
developments
applications
Sensors,
Journal Year:
2024,
Volume and Issue:
24(10), P. 3024 - 3024
Published: May 10, 2024
Soil–Vegetation–Atmosphere
Transfer
(SVAT)
models
are
a
promising
avenue
towards
gaining
better
insight
into
land
surface
interactions
and
Earth’s
system
dynamics.
One
such
model
developed
for
the
academic
research
community
is
SimSphere
SVAT
model,
popular
software
toolkit
employed
simulating
among
layers
of
vegetation,
soil,
atmosphere
on
surface.
The
aim
present
review
two-fold:
(1)
to
deliver
critical
assessment
model’s
usage
by
scientific
wider
over
last
15
years,
(2)
provide
information
current
developments
implemented
in
model.
From
conducted
herein,
it
clearly
evident
that
from
models’
inception
day,
has
received
notable
interest
worldwide,
dissemination
continuously
grown
years.
been
used
so
far
several
applications
study
interactions.
validation
performed
worldwide
shown
able
produce
realistic
estimates
parameters
have
validated,
whereas
detailed
sensitivity
analysis
experiments
with
further
confirmed
its
structure
architectural
coherence.
Furthermore,
recent
inclusion
novel
functionalities
as
outlined
review,
resulted
improving
capabilities
opening
up
new
opportunities
use
community.
also
ongoing
different
aspects,
advancing
our
understanding
both
educational
points
view
anticipated
grow
coming
Journal of Geophysical Research Atmospheres,
Journal Year:
2024,
Volume and Issue:
129(23)
Published: Nov. 28, 2024
Abstract
Offline
land
surface
models
(LSMs)
require
atmospheric
forcing
data
sets
for
simulating
water,
energy,
and
biogeochemical
fluxes.
However,
available
remain
highly
uncertain
can
introduce
additional
differences
in
LSM
simulations.
This
study
explored
the
impact
of
various
sets,
ranging
from
widely
used
to
newly
developed,
on
hydrological
simulations
using
Common
Land
Model
2024
(CoLM2024).
We
conducted
12
global
experiments
different
force
CoLM2024.
evaluated
model's
performance
against
plot‐scale
observations
globally
gridded
reference
data.
examined
uncertainties
forcings
their
output
variables
such
as
latent
heat,
sensible
net
radiation,
total
runoff.
Globally,
precipitation
has
highest
degree
uncertainty
at
4.4%.
The
propagate
model
cause
significant
simulated
variables.
Runoff
is
about
15.7%
globally,
with
a
greater
low
latitudes.
Our
evaluation
shows
that
developed
CRUJRA
ERA5LAND,
generally
outperform
others.
optimal
set
varies
depending
variable
interest
targeted
region.
Partial
Least
Squares
Regression
analysis
reveals
are
associated
dominant
variables,
highlighting
importance
selecting
specific
applications
regions.
confirms
improving
quality
consistency
meteorological
would
help
reduce
simulation
biases
guide
improvement
structure
parameterization