Tip:
Highlight text to annotate it
X
can you hear me
good morning everyone
My name's Dinh from Politecnico di milano
today, I would like to present anlysis results on the forest temporal
decorrelation in frame of TropiScat experiment
the team includes many
radar scientists from Politecnico di Milano, Onera
and cesbio
recent studies from the TropiSAR
campaign show
that the forest vertical structure represents a
key element to characterize scattering in tropical forests
and provide a more direct link to the forest biomass
and polinsar and particularly
SAR tomography
can provide forest vertical structure
as a consquence
biomass mission
is planed to operate in a tomography phase
During this phase the system would be able to to gather
multiple acquisitions
such characterized by a small baseline and a repeat pass time
in the order of a few days
that's allowing tomographic imaging
of the vegetation layer
figure you see the left
depicts a concept
of tomography configurations
and on the right one is a potential result
which is reconstructed from the tropiSAR data
this definitely allows us to see
the vertical forest structure
This phase is forseen with a repeat pass time of about 3 or 4 days
for tomography imaging
after which the time would be increased
about seventeen day to twenty five day to allow single or due baseline Polinsar measurements
Clearly uh...
the availability of multiple pass is a necessary condition for sar tomography
after which
it follows that the replicability of the results above
based on spaceborne data
is strictly connected
to the stability of the forest scattering over time
unfortunately there is a orbit constrained
in
sun synchronous satellite configuration
the revisit time is at least one day
so therefore the forest scattering changes over time
accordingly, evaluation the impact
of the temporal decorrelation in the forest areas
is a crucial issue
the goal of this talk
is to report the study of temporal decorrelation of scattering mechanism of radar signals
in a tropical forest as a function of forest heigh
and polarization
the TropiScat ground based experiment is planned
to give an answer to this question
this was chosen to be set up in the Paracou field station which is the same site with tropiSAR data
the equipment was installed on the Guyaflux tower
aa you can see here
on the top of a tower to radiate from p to l band radar signal to the forest below
on the left
the metalic tower 55m tall
that maybe see through
SAR tomography using images from
tropisar campaign data
the tomographic system has been designed
uh... to provide
full polarimetric
vertical resolution capability
uh...
to gather data continuously for about one year
and to provide us the
sufficient number of look
for reliable coherence evaluation at every pixel
at
several heigh level
within the forest
in this system at tomography mode the temporal sampling at p band is fifteen
minutes
just allowing making 96 sample per date for study both
short term and long term temporal correlation
as a function of
height in the forest
and this is the result
the slide shows shows what
the tomogram
at p band
for example
this one is
uh... hh top right hv
the bottom vh vv
and the last one vv
all of the panel has been flatten
so as to bring the terrain
level
to zero meter
so as to help
visualization and interpretation of the result
it is however important to note that
the terrain topography in the area is characterized
by .. area has a strong back slope
ie the terrain is tilted away from the tower
this leads to the absence of scattering contribution from the
ground trunk interactions
by closely
examining the
the tomogram you can see the structure of forest
for example we could see that
uh... area from uh...
fifty meter to one hundred meter so we can see
the contribution in the tomography of the canopy and ground from the ground
and top 30m above up
it is interesting to know there is a substantial gap
between
twolayer this suggests
a very dense
canopy is on the top
far away
far away from the tower about 200 meters the imaging is not certain
due to both
increasing distance and vertical resolution loss
thanks to very
high time sampling
it is
possible produce
a... tomography movie to capture the forest breathing
would you like to watch that movie
before
I open the movie please know that time index
in the title of panels ie 0H
is correspondent to
local midnight
as you can see
the forest motion during day hours is strong
which maybe due to wind disturbance or temperature variation
and at night time
it appears to be much more stable
an another interesting result
we found out that
location of forest center of mass
has been obseverd to go up and down
during the daytime
of course we can capture that motion too
so we go back
here is the law motion
we found that this motion has a strong correlation with the temperature
variation during one day
the figure on the top is the law of motion
and bottom the temperature
variation over one day
as a function of heigh
as you can see here
this strong correlation
phenomenon
maybe suggest
that
this phenomenon might be
connected to the water content inside
the vegetation layer
one possible explaination can be
considered
evapotranspiration phenomena
for example
with changing temperature
the exchange of oxygen and carbon dioxide
between leaves and atmosphere varies
this leads to the water content inside the trunk varying
for example you see forests
store
the water
inside them overnight
and then
afternoon with
higher temperaure
as it happens
sorry in the morning when with higher temperature and in afternoon
when the temperature goes down the center of mass ... shifts
0:09:07.140,0:09:10.220 okay we move to another interesting results
for example
how does coherence change as a function of forest height and what is the
best time
for daily data acquisition
the questions can be answered
by examining
the temporal
day and night
coherency matrix
we show
the HV coherency matrix in the slide
as a function of forest height
uh... on the top
on the top left zero meter
on the top right ten meter
20 meter
bottom right thirty meter
as you can see here all the pannel exhibit
the same behaviour
temporal
coherence drops
during day time
uh... but you can see that
in the ground it is a litle bit much stable
what about the other channels
similar obsevation can be made indeed we can even decompose the stable and
varying
mechanisms information
it is possible
to describe its main contribution from the polarimetric and temporal point of view
indeed we can use the
algebraic synthesis technique
to describe
to recognize and separate
its informations
correspondent with its contribution
resulting in the state and
varying mechanisms information
in this slide we show
all channel infomation as well as with
stable and varying component
what we can see here that uh... there
the interferometric
amplitude and the phase
variation during day and night
as a function of height
zero meter
ten meter
twenty meter
again we could see that
in the ground layer
the variation is much more stable
there is so interesting because suggestion that
in tropical forests
biomass performance can be optimized
their data
in night time for example from dusk to dawn
based on this data analysis we use this
dusk dawn time to investigate
the long time span for more investigation coherence
for example
uh... what you think
what is
temporal decorrelation in the forest at p band after three days
for tomography
processing
as you can see here
coherency three day
the average
coherence is remarkable high
uh... and
about what ten meter above the ground we can see
is a little bit lower
and what is about polinsa for example
how much is the temporal of decorrelation in forest the for p band
after 25 day for polinsar inversion
as that what you see
it is still good
it means that
the good polinsar can be expected as well
again there is
a little bit lower
in ten meter
in the ground they are is quite stable and very high
due to many data available
we can even model
the decorrelation as follows
we model temporal coherence as
exponential decay with an initial value
and time constant
these models are well known in sar literature
uh... time constantly will give us an idea about the lifetime scatterer
the identification of the parameter can be esimated
by
trying many combinations and choosing the best one
based on this figure of merit
this is the result it's also the last slide
in the slide we show here
quantitative expectation of temporal coherence as a function of height and
polarization
for example
in every channel
uh... we can see up to three day
and
twenty-five day value for example up to twenty five and we can see
their value about
uh... to 0.8 on ground
and 0.7
in the middle to upper layer thus witness the coherence here is the sensitivity to
heigh
the black and green lines
that denote third day and 25 day
which is
forseen for the biomass mission
in summary the first time ever at least in our knowledge
tomography movie capturing the forest daily changing is produced
and the analyses reveals the diurnal
motion of forest center of mass
the temporal decorrelation in one day suggests
biomass should be position at dawn dusk time since it will
minimize not only ionosphere disturbances
but also temporal coherence loss
finally from the long term coherence and we can see that
expectation of temporal coherence
in HV channel
has been observed about zero eight at ground and
zero seven in the middle layer so far we have seen their
sensitivity to height
and that's all thanks for your attention