Diapositiva 1 - ReSyLAB 2015

2nd ReSyLAB, 14-16 May 2015, Belgrade - Serbia
Satellite multi-temporal interferometry
for landslide hazard detection
in the peri-Adriatic area
Janusz WASOWSKI1
F Bovenga2, R Nutricato3, DO Nitti3, MT Chiaradia4
(1) CNR-IRPI (National Research Council), Bari, Italy
(2) CNR-ISSIA (National Research Council), Bari, Italy
(3) GAP srl c/o Politecnico di Bari, Bari, Italy
(4) Dept. Physics, c/o Politecnico di Bari, Italy
j.wasowski@ba.irpi.cnr.it
Key message
Big Data from radar satellites, high spatial (1-3 m) and temporal
resolution (days-weeks) sensors e.g., COSMO-SkyMed & TerraSAR-X,
and better image processing techniques
 Multi-temporal interferometry (MTI) now more effective for both
regional & local scale slope hazard detection & landslide monitoring
 Expect greater use of MTI for regular (weekly-monthly), long-term
(years), wide-area (>1000 km2) monitoring of landslide hazards
Content
I.
Background information on MTI, available satellite radar data and
new generation sensors
II.
Regional & local scale MTI applications for slope hazard detection
a) Hilltop town instability in the S. Apennines (Apulia Region)
b) Landslide activity in the Dinarides Mts of Central Albania
III. High resolution CSK data for MTI in SERBIA: a possible landslide case
application
IV. Early detection of slope hazards  where? possible via high res. MTI
but temporal predictions  when? difficult
Satellite MTI methods using good (persistent) radar targets
(persistent scatterers - PS) e.g. PSInSAR (PSI), SBAS
Synthetic Aperture Radar (SAR): ra>dar sensors that work using
microwaves  = [1, 100] cm:
• Active sensors
• Negligible atmospheric absorption
Day & Night / All weather
observations
> background information  Wasowski & Bovenga, 2014, Eng. Geology, 174, 103-138
Good radar targets - PS = those that remain measurable through time
mainly buildings and other human-made structures
from >1000 to 10,000 PS/km2
Good radar targets: also rock outcrops, ≈bare ground
GOOD
GOOD
NO
NO
~OK
~OK
≈10-100 PS/km2
MAYBE
Differential interferometry / MTI: 2 / >15-20 radar acquisitions
during successive satellite passes over the same area
What is measured?
distance (R) changes in time (t)
between satellite sensor and
targets on the ground
MTI products: example of Rome
SLC
InSAR Processing
Rome - Tiber River
MTI products: color coded PS map & PS displacement trend
PS displacement trend
SLC
InSAR Processing
PS map = {PSi}
Moving away from the SAT
No moving
Rome - Tiber River
settlement of alluvials
Moving toward the SAT
Advantages and products of MTI
•
Systematic, precise (mm-cm) measurements of ground surface
deformations over large areas ( >1000 km2) by exploiting locally dense
( >100 - >1000 points/km2 ) ”natural” radar targets
•
Ground surface deformation maps and temporal deformation trends
(time series) for each radar target (Lat, Long, Height position)
•
Retrospective studies, exploiting archives of space agencies (since 1992- )
Data: “historic” & new radar satellite missions
1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 …
ERS-1
ERS-2
retired in 2001
RADARSAT-1
RADARSAT-2
† 2011
ENVISAT
JERS
2013 2014
ALOS
† 2011
COSMO-SkyMed
TerraSAR-X
KOMPSAT-5
C band ( = 5.6 cm)
L band ( = 23.6 cm)
X band ( = 3.1 cm)
SENTINEL-1
ALOS-2
New generation X-band sensors & SENTINEL-1
 higher temporal / spatial resolution
Sensor
Band Wavelength λ
Spatial
Resolution
az x range (m)
Repeat Cycle
dt
Max Velocity
vmax=λ/(4∙dt)
ERS/ENVISAT †
C - 5.6 cm
≈ 6 x 24
35 days
14.6 cm/y
RADARSAT-2
C - 5.5 cm
≈3x3
≈8x8
24 days
20.4 cm/y
SENTINEL-1
C - 5.6 cm
≈ 5 x 20
12 days
6 days
21.3 cm/y
42.6 cm/y
ALOS †
L - 23.6 cm
≈ 5 x 8.6
46 days
46.8 cm/y
COSMO-SkyMed
(CSK)
X - 3.1 cm
≈ 1.0 x 1.0
≈ 2.5 x 2.5
16 days
8 days
4 days
17.7 cm/y
35.4 cm/y
70.7 cm/y
TerraSAR-X
X - 3.1 cm
≈ 1.0 X 1.0
≈ 3.3 x 2.8
11 days
25.7 cm/y
KOMPSAT-5
X - 3.1 cm
≈ 1.0 X 1.0
≈ 3.0 x 3.0
28 days
10.4 cm/y
Review article on landslide investigation using MTI
2014
Vol. 174, 103-138
Overview of remote sensing techniques for landslide motion
measurement with emphasis on MTI
Wasowski & Bovenga, 2014
Remote Sensing of Landslide Motion with
Emphasis on Satellite Multitemporal
Interferometry Applications: An Overview
http://dx.doi.org/10.1016/B978-0-12396452-6.00011-2
MTI-based landslide investigations in the world
versus non-seismic fatal landslides in 2004-2010 (Petley, 2012)
S. Apennines – Cascini et al.
Slovenian Alps – Komac et al.
Examples of MTI applications in the peri-Adriatic region:
1 - SE Apennine Mts, Italy; 2 - Dinarides Mts, Central Albania
12
Petley, 2012
Example from the SE Apennines Mts - many hilltop towns & roads
affected by slope instability problems Adriatic Sea
From regional to local scale investigation – SE Apennines
overview of hilltop towns stability using 20 m ENVISAT data
Local scale MTI investigation using 20 m ENVISAT desc & asc data detecting extremely slow landslide in a small hilltop town
Comparison of MTI results from 3 m TSX & 20 m ENVISAT data
for a small town in a landslide-prone area
high resolution (3 m)
--> 18100 radar targets
medium resolution (20 m)
--> 1140 radar targets
Regional scale investigation – SE Apennines
overview of hilltop towns stability using high res. 3 m TSX data
Adriatic Sea
From regional  local scale investigation – SE Apennines
hilltop towns stability using high res. 3 m TSX data
Adriatic Sea
SE Apennines: local scale investigation using 3 m TSX data
ground / road instabilities around a hilltop town
SE Apennines: local scale investigation using 3 m TSX data
ground / road instabilities around a hilltop town
A
B
A
B
MTI results for Central Albania – high res. 3 m CSK data
Tirana
MTI results for Central Albania: from regional  local scale
Tirana
Detecting activity of very slow landslides in the Dinarides,
Central Albania
Activity of very slow landslides in the Dinarides, Central Albania
example of displacement trend
Activity of very slow landslides in the Dinarides, Central Albania
example of displacement trend
Monitoring engineering structure (roads, buildings) instability
in landslide-prone areas: examples from Southern Italy
MTI velocity map showing unstable & stable parts of road network
around a town in the SE Apennines
High resolution (3 m) MTI velocity map showing instability of a viaduct
(A3 Highway) crossing a landslide, Calabria, Italy
Wasowski & Bovenga, 2014
Engineering Geology, 174
MTI velocity map and displacement time series showing instability of
a viaduct crossing a landsliide, Calabria, Italy
Wasowski & Bovenga, 2014
Engineering Geology, 174
A
B
(A)
(B)
Conclusion
MTI now more effective for regional and local scale
landslide hazard monitoring thanks to improved spatial
(1-3m) and temporal (~weekly) resolutions of new radar
satellites, as well as better data processing techniques
SERBIA + around: coverage by high res. COSMO-SkyMed data
SERBIA + around: coverage by high res. COSMO-SkyMed data
Belgrade – Umka area: coverage by CSK data
The Umka landslide
Early detection of slope hazards?
January 2014 slope failure & train accident
near the town of Marina di Andora, NW Italy
Jan 2014 slope failure & train accident
near the town of Marina di Andora, NW Italy
MTI velocity map along the railway near Marina di Andora
CSK 3 m data
MTI velocity map for the failed slope area
Displacement time series of radar target from a building just above
the failed slope: 2008-2014 avg. velocity 9 mm/yr
Failure 17.01.14
Warning signal in displacement time series?
failure
Lessons from the Andora slope failure & train accident
 Small landslide, but big problem  need good spatial resolution
 Deformation trends from long-term, high res. MTI monitoring of slopes
 unique opportunity for early detection & warning of hazard  where
 Temporal predictions  when? remains very difficult