Abstract.
The
recent
surge
in
reservoir
construction
has
increased
global
surface
water
storage,
with
Mainland
Southeast
Asia
(MSEA)
being
a
significant
hotspot.
Such
infrastructural
evolution
demands
updates
management
strategies
and
hydrological
models.
However,
information
on
actual
storage
is
hard
to
acquire,
especially
for
transboundary
river
basins.
To
date,
no
high
spatio-temporal
dataset
absolute
time
series
available
reservoirs
MSEA.
address
this
gap,
we
present
(1)
comprehensive,
open-access
database
of
(sub-monthly)
185
(larger
than
0.1
km3)
MSEA
spanning
the
period
1985–2023,
(2)
an
analysis
dynamics.
MSEA-Res
includes
static
(Area-Elevation-Storage
curves,
frequency,
extent)
dynamic
(area,
level,
series)
components
each
reservoir.
collectively
store
around
175
km³
(140
–
210
km³)
water,
covering
aggregated
area
8,700
km²
(6,500
10,000
km²).
We
show
that
combined
average
from
70
160
(+130
%)
2008
2017,
primarily
contributed
by
dams
Irrawaddy,
Red,
Upper
Mekong,
Lower
Mekong
Our
in-situ
validation
provides
good
match
between
estimated
observations,
60
%
sites
(12
out
20)
showing
R²
>
0.65
nRMSE
<
15
%.
indirect
(based
altimetry-converted
storage)
shows
even
better
results,
0.7
12
(14
reservoirs.
Furthermore,
2019–2020
drought
event
reveals
nearly
30–40
region
experienced
more
five
months
drought,
most
impact
Cambodia
Thailand.
As
result,
departures
ranged
up
-40
some
reservoirs,
highlighting
impacts
availability.
Overall,
demonstrates
potential
inferred
assessing
real-life
water-related
problems
Asia,
possibility
applications
other
parts
world.
associated
Python
code
are
publicly
Zenodo
at
https://doi.org/10.5281/zenodo.12787699
(Mahto
et
al.,
2024).
IEEE Geoscience and Remote Sensing Letters,
Journal Year:
2024,
Volume and Issue:
21, P. 1 - 5
Published: Jan. 1, 2024
Emerging
deep
learning
methods
for
satellite-derived
bathymetry
(SDB),
in
which
water
depth
is
estimated
using
satellite
band
reflectance
values,
typically
treat
the
problem
as
either
classification
or
regression
tasks,
can
underperform,
particularly
when
data
exhibits
a
skewed
distribution.
In
this
work,
we
propose
novel
jointly-trained
classification-regression
(JTCR)
model
SDB
that
first
classifies
input
values
to
correspond
range
and
then
performs
within
each
range.
Using
Shetrunji
reservoir,
an
inland
reservoir
India,
case
study,
with
Sentinel-2
demonstrate
our
proposed
outperforms
other
competitive
models,
including
derived
from
separate
training
of
tasks
architecture.
Concretely,
observe
Root
Mean
Square
Error
(RMSE),
Absolute
(MAE),
R-squared
(R2)
0.17,
0.05,
0.99,
respectively,
JTCR
model,
compared
0.71,
0.85
feedforward
neural
network
model.
Environmental Research,
Journal Year:
2024,
Volume and Issue:
262, P. 119860 - 119860
Published: Aug. 29, 2024
Dam
reservoirs
are
at
the
core
of
local
water
storage
and
supply,
especially
in
water-stressed
regions
world
with
acute
shortage
problems.
However,
evaporative
losses
from
these
their
efficiency
often
overlooked
budgeting.
We
offer
a
mechanistic
approach
that
combines
physically-based
modeling
remote
sensing
information
reservoir
characteristics
to
reliably
predict
dam
reservoirs.
The
developed
framework
is
used
potential
different
basins
worldwide.
apply
this
10
largest
world's
quantify
losses.
Our
analysis,
spanning
2000
2020,
reveals
considerable
variations
annual
evaporation
rates
located
water-deprived
exceeding
3200
mm/year
during
study
period
total
loss
reaching
26.5
km
Journal of Degraded and Mining Lands Management,
Journal Year:
2025,
Volume and Issue:
12(2), P. 7337 - 7351
Published: Jan. 1, 2025
A
crucial
component
of
water
supply
in
arid
tropical
regions
is
the
construction
small
reservoirs.
Interestingly,
various
problems
arise
and
management
reservoirs,
so
role
reservoirs
providing
surface
considered
less
than
optimal.
This
study
aimed
to
identify
that
cause
function
be
optimal
provide
direction
for
reservoir
dry
areas.
The
investigation
was
carried
out
two
primary
stages:
site
analysis
issues
with
usability
regional
physical
characteristics
regionally.
Determining
points
using
SPOT
6/7
imagery
a
resolution
1.5
m.
Land
system
maps
morphometry
were
used
analyze
site.
Field
surveys
in-depth
interviews
conducted
benefits
limiting
factors
results
revealed
are
found
locations
low
altitudes
(0-100
masl)
undulating
terrain
(8-15%).
According
findings,
95.3%
still
water-filled.
However,
many
resulted
suboptimal
utilization
Specifically,
faced
structural
damage
due
1)
erosion-landslides
(74.77%),
2)
sedimentation
(33.64%),
3)
seepage
(7.48%),
4)
embankment
collapse
(6.54%),
5)
leakage
(2.80%).
Most
community
does
not
utilize
limited
infrastructure.
Pipes
distribute
only
available
at
around
46.73%,
while
tanks
43.93%.
Directions
future
adding
infrastructure,
sediment
management,
community-based
management.
International Journal of Agricultural and Environmental Information Systems,
Journal Year:
2025,
Volume and Issue:
16(1), P. 1 - 17
Published: April 18, 2025
The
reasonable
storage
and
retrieval
of
spatial
data
in
rivers
lakes
can
promote
the
development
river
lake
management
protection
projects.
In
order
to
efficiently
store
retrieve
data,
this
study
adopts
computer
technology
design
a
architecture
method
based
on
types.
structured
relational
databases
document
with
indexing
characteristics.
Use
geospatial
abstraction
library
read
write
raster
image
from
unstructured
use
Elasticsearch
metadata.
test
results
show
that
minimum
latency
is
13ms,
average
response
time
78ms,
maximum
throughput
14000
req/s,
failure
rate
0.106%.
designed
database
performance
are
excellent,
providing
technical
support
for
efficient
data.
CATENA,
Journal Year:
2024,
Volume and Issue:
242, P. 108090 - 108090
Published: May 11, 2024
Naturally
closed
lakes
located
on
the
Tibetan
Plateau
provide
a
more
authentic
depiction
of
climate
change
and
have
undergone
significant
but
dissimilar
changes
over
past
four
decades.
Although
previous
research
has
concentrated
lake
at
regional
continental
scales,
dominant
factors
related
to
catchment
scale
remain
elusive.
This
study
employed
hierarchical
minimum
variance
clustering
method,
six
representative
topographies
land
surface
descriptors
classify
catchments
322
(>1
km2)
across
from
1981
2020.
To
better
understand
driving
in
catchment-scale
areas,
we
assigned
two
distinct
attribute
classes
resulting
classifications:
hydro-meteorological
characteristics
geographical
environmental
aspects.
Our
analysis
revealed
that
five
clusters
are
geographically
coherent
exhibit
characteristics.
By
assessing
decadal-scale
time
for
with
area
greater
than
±
50
%,
found
most
driver
2000
was
continuous
decline
snowmelt,
followed
by
substantial
increase
precipitation,
temperature,
glacial
meltwater
Regarding
trends
magnitude,
different
responses
may
occur
areas
similar
physical
geography
climatology
(i.e.,
same
cluster).
During
decades,
only
3
%
(19)
shrank
decreased
rate.
Also,
intense
evaporation
corresponds
period
decrease
area.
Despite
variation
among
clusters,
precipitation
dominates
inconsistent
under
climatology,
some
exceptions
strongly
groundwater
melted
permafrost
recharge.
findings
laid
groundwork
understanding
this
region's
complex
hydrological
behavior.
Computers & Fluids,
Journal Year:
2024,
Volume and Issue:
278, P. 106321 - 106321
Published: May 25, 2024
Knowledge
of
the
bottom
topography,
also
called
bathymetry,
rivers,
seas
or
ocean
is
important
for
many
areas
maritime
science
and
civil
engineering.
While
direct
measurements
are
possible,
they
time
consuming,
expensive
inaccurate.
Therefore,
approaches
have
been
proposed
how
to
infer
bathymetry
from
surface
waves.
Mathematically,
this
an
inverse
problem
where
unknown
system
state
needs
be
reconstructed
observations
with
a
suitable
model
flow
as
constraint.
In
cases,
shallow
water
equations
can
used
describe
flow.
theoretical
studies
efficacy
such
PDE-constrained
optimisation
approach
reconstruction
exist,
there
seem
few
publications
that
study
its
application
data
obtained
real-world
measurements.
This
paper
shows
can,
at
least
qualitatively,
reconstruct
Gaussian-shaped
in
wave
flume
free
level
up
three
points.
Achieved
normalized
root
mean
square
errors
(NRMSE)
line
other
approaches.
Water Resources Research,
Journal Year:
2024,
Volume and Issue:
61(1)
Published: Dec. 28, 2024
Abstract
Accurate
reservoir
representation
in
large‐scale
river
models
remains
challenging
owing
to
limited
access
data
on
operations.
We
contribute
model
development
by
introducing
a
global
machine‐learning
based
flood
storage
capacity
(FSC)
set
and
satellite‐based
target
operation
scheme
(SBTS).
The
FSC
for
1,178
control
reservoirs
is
constructed
using
multiple
attributes
reported
data.
Integrating
these
FSCs
into
SBTS
enables
its
applicability
with
generic
formulations
of
zoning.
Then,
we
develop
monthly
median
values
satellite
as
parameters.
With
seasonal
patterns
constrains,
improvements
simulation
results
are
achieved.
When
simulated
observed
inflow,
performed
significantly
better
(median
Kling‐Gupta
efficiency
0.52
0.17
outflow
simulations
among
289
reservoirs),
compared
the
previous
linearly
interpolated
parameter
(0.41
−0.19).
Compared
two
existing
schemes
without
storages,
demonstrates
improved
performance
many
whose
inflow
pattern
more
regular.
coupled
model,
it
discharge
across
293
downstream
gauges,
overall
performance,
peak,
low
flow
improving
at
40%,
21%,
35%
respectively,
reservoirs.
However,
do
not
improve
notably
due
biases
demonstrated
that
observations
help
parameterizations,
found
other
aspects
modeling
essential
accurately
reproducing
patterns.