bioRxiv (Cold Spring Harbor Laboratory),
Год журнала:
2022,
Номер
unknown
Опубликована: Янв. 28, 2022
Abstract
Continental-scale
increases
in
aquatic
system
eutrophication
are
linked
with
increased
cyanobacteria
threats
to
recreational
water
use
and
drinking
resources
globally.
Increasing
evidence
suggests
that
diurnal
vertical
migration
of
key
factors
must
be
considered
cyanobacterial
bloom
risk
management.
While
this
has
been
discussed
marine
eutrophic
freshwater
contexts,
reports
oligotrophic
lakes
scant.
Typical
monitoring
protocols
do
not
reflect
these
dynamics
frequently
focus
only
on
surface
sampling
approaches,
either
ignore
time
or
recommend
large
midday
timeframes
(e.g.,
10AM-3PM),
thereby
preventing
accurate
characterization
community
dynamics.
To
evaluate
the
impact
migrations
column
stratification
abundance
composition,
communities
were
characterized
a
shallow
well-mixed
lake
interconnected
thermally
stratified
Turkey
Lakes
Watershed
(Ontario,
Canada)
using
amplicon
sequencing
16S
rRNA
gene
across
multi-time
point
series
2018
2022.
This
work
showed
present
their
structure
varies
(i)
diurnally,
(ii)
depth
column,
(iii)
interannually
within
same
(iv)
between
different
closely
watershed.
It
underscored
need
for
integrating
multi-timepoint,
multi-depth
discrete
guidance
into
reservoir
programs
describe
signal
change
inform
management
associated
potential
cyanotoxin
production.
Ignoring
variability
(such
as
reported
herein)
reducing
sample
numbers
can
lead
false
sense
security
missed
opportunities
identify
mitigate
changes
trophic
status
risks
such
toxin
taste
odor
production,
especially
sensitive,
systems.
Graphical
Highlights
■
Cyanobacterial
populations
fluctuate
sporadically
cycles
vary
significantly
Significant
annual
shifts
higher
Cyanobacteria
should
incorporate
Remote Sensing,
Год журнала:
2023,
Номер
15(5), С. 1361 - 1361
Опубликована: Фев. 28, 2023
Timely
and
accurate
crop
yield
information
can
ensure
regional
food
security.
In
the
field
of
predicting
yields,
deep
learning
techniques
such
as
long
short-term
memory
(LSTM)
convolutional
neural
networks
(CNN)
are
frequently
employed.
Many
studies
have
shown
that
predictions
models
combining
two
better
than
those
single
models.
Crop
growth
be
reflected
by
vegetation
index
calculated
using
data
from
remote
sensing.
However,
use
pure
sensing
alone
ignores
spatial
heterogeneity
different
regions.
this
paper,
we
tested
a
total
three
models,
CNN-LSTM,
CNN
LSTM
(ConvLSTM),
for
annual
rice
at
county
level
in
Hubei
Province,
China.
The
model
was
trained
ERA5
temperature
(AT)
data,
MODIS
including
Enhanced
Vegetation
Index
(EVI),
Gross
Primary
Productivity
(GPP)
Soil-Adapted
(SAVI),
dummy
variable
representing
heterogeneity;
2000–2019
were
employed
labels.
Data
download
processing
based
on
Google
Earth
Engine
(GEE).
downloaded
images
processed
into
normalized
histograms
training
prediction
According
to
experimental
findings,
included
represent
had
stronger
predictive
ability
just
data.
performance
CNN-LSTM
outperformed
or
ConvLSTM
model.
Water Research,
Год журнала:
2024,
Номер
252, С. 121199 - 121199
Опубликована: Янв. 26, 2024
Cyanobacteria
increasingly
threaten
recreational
water
use
and
drinking
resources
globally.
They
require
dynamic
monitoring
to
account
for
variability
in
their
distribution
arising
from
diel
cycles
associated
with
oscillatory
vertical
migration.
While
this
has
been
discussed
marine
eutrophic
freshwater
contexts,
reports
of
diurnal
migration
cyanobacteria
oligotrophic
lakes
are
scant.
Typical
protocols
do
not
reflect
these
dynamics
frequently
focus
only
on
surface
sampling
approaches,
either
ignore
time
or
recommend
large
midday
timeframes
(e.g.,
10AM-3PM),
thereby
preventing
accurate
characterization
cyanobacterial
community
dynamics.
To
evaluate
the
impact
migrations
column
stratification
abundance
composition,
communities
were
characterized
a
shallow
well-mixed
lake
interconnected
thermally
stratified
Turkey
Lakes
Watershed
(Ontario,
Canada)
using
amplicon
sequencing
16S
rRNA
gene
across
multi-time
point
series
2018
2022.
This
work
showed
that
present
structure
varies
(i)
diurnally,
(ii)
depth
column,
(iii)
interannually
within
same
(iv)
between
different
closely
watershed.
It
underscored
need
integrating
multi-timepoint,
multi-depth
discrete
guidance
into
reservoir
programs
describe
signal
change
inform
risk
management
potential
cyanotoxin
production.
Ignoring
(such
as
reported
herein)
reducing
sample
numbers
can
lead
false
sense
security
missed
opportunities
identify
mitigate
changes
trophic
status
risks
such
toxin
taste
odor
production,
especially
sensitive,
systems.
