This document analyzes landslides that occurred in Morretes, Brazil in March 2011 using synthetic aperture radar (SAR) data. Several image processing techniques were applied to detect landslides, including image differencing, normalized difference scattering index, polarimetric decompositions, and amplitude analysis. Difference images and normalized difference scattering index images best identified landslide areas. While landslides could be detected, it was difficult to precisely threshold landslide regions from the SAR data alone. Future work involves integrating SAR data into weather monitoring systems and developing an operational landslide monitoring system using optical and SAR data.
Slide deck for the IPCC Briefing to Latvian Parliamentarians
Case Study - Landslide in Morretes/Brazil
1. Analysis
of
Landslides
in
Morretes・Brazil
2011
Tokyo-‐Japan,
August
7th
2012
Brazil
–
Eng.
Fábio
Sato
-‐
SIMEPAR
2. Case
Study:
Morretes
2011-‐03-‐11
Small
tourisNc
and
rural
city
in
Paraná
state,
Brazil.
Has
recently
suffered
flooding
and
several
landslides
in
March
11th
2011,
caused
by
3
days
of
conNnuous
rainfall.
Civil
Defense
Sta.s.cs
Damaged
houses:
2.720
Homeless
people:
967
Removed
people:
8.453
Deaths:
04
Affected
people:
23.828
Affected
ciNes:
7
5. Case
Study
Data
Before
Event
(Master
data)
2
years
before
ObservaNon
date:
2009/03/25
Processing
level:
1.1
Scene
shi[:
-‐4
Mode:
FBS
Scene
id:
ALPSRP168626680
A?er
Event
(Master
data)
2
weeks
a[er
ObservaNon
date:
2011/03/31
Processing
level:
1.1
Scene
shi[:
-‐4
Mode:
FBS
Scene
id:
ALPSRP275986680
6. Analysis
Methodology
1. Amplitude
image
analysis
1. Image
Difference
(HH
&
VV)
2. NSDI
(HH
&
VV)
2. Polarimetric
Analysis:
1. Pauli
Image
2. Freeman
decomposiNon
3. H/A/alpha
decomposiNon
Interferometry
and
phase
coherence
analysis
could
not
be
conducted
7. Analysis
Flowchart
1/3
Master
Level
1.1
data
Slave
Level
1.1
data
Polsar
Pro
• Geocode
T3
Matrix
ASF
MapReady
3x3
Complex
Coherency
T3
files
• Extract
MulNlook
Image
(4x4
window)
Geocoded
T3
files
8. Analysis:
Flowchart
2/3
Geocoded
T3
files
Polsar
Pro
Freeman3
decomposiNon
H
/
A
/
Alpha
decomposiNon
Freeman
RGB
Composite
H/A/Alpha
RGB
Composite
9. Analysis:
Flowchart
3/3
Master
Level
1.1
data
Slave
Level
1.1
data
ASF
MapReady
STRM
DEM
GeoTiff
files
ArcGIS
Difference
Image
NDSI
Image
• Geocode
• Terrain
CorrecNon
with
DEM
18. Conclusions
• Landslides
could
be
idenNfied
on
all
generated
images/products
• In
this
case,
befer
results
where
provided
by
Difference
and
NDSI
images
• The
analysis
of
polarimetric
products
are
more
difficult
to
interpret
• Although
it
was
possible
to
detect
landslides,
it
is
difficult
to
produce
a
final
threshold
image
of
the
landslides
regions
from
SAR
data
19. Future
Plans
• Short
Term:
IntegraNon
of
SAR
data
into
meteorological
visualizaNon
systems
at
SIMEPAR
• Medium
Term:
Develop
operaNonal
system
for
landslide
detecNon/
monitoring
with
satellite
opNcal
and
SAR
data
in
Parana
• Long
Term:
Pursue
research
on
landslide
forecasNng
• OpportuniNes:
Propose
projects
about
landslide
monitoring
in
Brazil
and
LaNn
America