Download as pdf or txt
Download as pdf or txt
You are on page 1of 36

Introduction to SAR Interferometry

Theoretical Basis

www.dares.tech
1
Our

Objectives

• To prevent failure events related with ground deformation.


• Exploit and automate monitoring services together with InSAR
techniques for total control.
• Minimize the presence of personnel in the field without loosing
any information.
What are microwaves
useful for?
Introduction

Remote Sensing
Remote sensing is the science of acquiring information about the Earth's surface without being in contact with it.
There are 2 kinds of remote sensing sensors:
• Passive Sensors: optical sensors, radiometers, etc.
• Active Sensors: Radar, Synthetic Aperture Radar (SAR), Scatterometers, Altimeters, etc.

Passive sensors Active sensors

Measure the sun's Provide their own


energy that is energy source for
either reflected by illumination.
the Earth, or The sensor emits
absorbed and then radiation toward the
re-emitted, as it is target and
for thermal infrared measures the
wavelengths. backscattered
signal.
Difference between active and passive remote sensing sensors

Remote Sensing
Radar Image
Radar Sensors Optical Sensors

Active sensors Passive sensors

Provide their own energy source for Lack of an independent source of


illumination radiation

Observation during day/night & through Observation only during daylight and
cloud cover with no cloud cover
Optical Image

Side looking geometry Nadir looking geometry

Centimeter wavelength Up to the visible wavelengths

Coherent: amplitude and phase Non-Coherent: Reflectivity


Electromagnetic Spectrum

The Microwave Spectrum


Electromagnetic Spectrum Austrian Pine Example
Interaction between the electromagnetic waves and the
elements of the scene

X-band

C-band

L-band

Band Frequency Designations


P 0.3 - 1 GHz
L 1 - 2 GHz
S 2 – 4 GHz
C 4 – 8 GHz
X 8 – 12 GHz
Ku 12 – 18 GHz
X-band C-band L-band
K 18 – 26 GHz
λ = 3 cm λ = 5 cm λ = 20 cm
Ka 26 – 40 GHz
Constellations and Archive

Space-Borne
SAR Sensors
Sensor Wavelength Resolution Rev. Time
[cm] [meters] [days]

ALOS-2 22.9 3x10 14


ENVISAT 5.6 4x20 35
Radarsat-2 5.5 5x8 24
Sentinel 5.5 13x5 6-12
TerraSAR-X 3.1 3x3 11
Cosmo-SkyMed 3.1 3x3 1-8

SAR sensor choice should fit customer needs based


on the revisiting-time and penetration capabilities
required to tackle the ground displacement
phenomena of the area of interest
Synthetic Aperture Radar
Radar: Radio Detection and Ranging

Backscattered Signal and Round-Trip Delay


The objective of a is to measure the distance or range of targets in a scene and identify their structural type
Transmitted
p(t)
Signal
 (x) Scatterers
2 4
1 3

n

Pulse Duration
x1 x2 x3 x4 xn
TP

s (t) Received Echoed Signal

2 RCS Round-Trip
4 Delay
3
1
n

t1 t 2 t3 t4 … tn
t = 2x
c
Amplitude and phase information
Wrapped Phase Concept
Radar: Radio Detection and Ranging

Understanding the Phase Delay


The phase detected by the radar is wrapped k*2π cycles (module 2π)
Taking into account the two-way travel path of the radar pulse, the phase of a target does not change if its distance from the radar is a
multiple of λ/2
r3
r2
Depending on the distance of the
r1
target, the carrier frequency
modulated by the radar pulse is
received by the radar antenna
 =0 with a different phase shift with
time
respect to the original sinusoidal
signal used to generate the
transmitted pulse.

= Using phase differences the
2 time
phase delay due to the target
position can be cancelled and
3 detecting displacements
= between consecutive pairs with
2 time millimetric precision will be
possible
eff =  / 2
Definition

Synthetic Aperture Radar


Description and Acquisition geometry Range Resolution (Pulse compression)
SAR sensors are radar systems operating in the microwave LFM chirp signal
spectrum that allow obtaining all-weather terrain complex
reflectivity images.
c
orbit track Dr =
2× B
azimuth
vs direction
θoff
Azimuth Resolution (Synthetic Aperture)
ran

fs L sa
ge
dir

ti t obs tf
e
ctio
r
n

nadir

z
La
θ
l
Dx =
x y imaging
radar La 2
vg
sw

footprint
at
h

ground-range
plane

ground target
Acquisition Geometry

Radar Point of View


Optical Sensors Radar Sensors SAR Image
According to the point of view According to the distance The volcano in the SAR image appears with a
geometrical distortion due to the system acquisition
geometry
Acquisition Geometry

Geometrical Distortions
Foreshortening Layover Shadowing

B'
A' B'
A'
B' B
B C'
D'
C'
B CD
A C A
ground-range ground-range
ground-range

The foreshortening effect may be defined The layover is the opposite case of The shadowing effect occurs when a
as the compression of the slopes oriented foreshortening. The layover distortion slope opposite to the radar has a steeper
towards the satellite (A-B) when they are effect occurs when the angle of a slope slope than the incidence angle. This
projected onto the slant-range plane (A’- facing to the radar (A-B) exceeds the causes some parts to be not illuminated
B’). As the local slope is closer to the incidence angle. This inverse effect in the by the radar (C-D), the so-called shaded
incidence angle this effect is more echo arrival order provoke that those areas, and thus no reflectivity or phase
pronounced. points located at higher elevations, which information from these points can be
are closer to the radar, appear earlier in collected.
SAR images (B’-A’).
Acquisition Geometry

Ascending and Descending Orbits


Satellites with SAR sensors orbit the
Earth in a sun synchronous LEO polar
Descending
orbit and data acquisitions can be
made at any time of day or night and N N
independent of cloud coverage,
collecting both amplitude and phase
data. SAR satellites have repeating
paths and collect images in two
different orbit passes (ascending and
descending) taking profit of Earth’s
rotation.

The processing should be tailored


to the best orbit geometry in order
to minimize geometrical distortions
S
and maximize the detection S
Ascending
capabilities
Acquisition Geometry

Ascending and Descending Orbits


ASCENDING DESCENDING range
ORBIT ORBIT

azimuth
azimuth

Sentinel-1 Image Sentinel-2 Image of Mount Etna Sentinel-1 Image


range Ascending Descending

Choosing the correct orbit pass is mandatory in order to minimize geometrical distortions and maximize the
detection capabilities in the area of interest, specially over environments with steep topography
SAR Interferometry
SAR Amplitude

Amplitude and Phase


The phase of a SAR image is related with the distance the
electromagnetic pulse travels from the satellite to the ground
and back to the satellite.

Phase 
SAR Interferometry exploits phase-differences between
different satellite passes.

-
Differential Phase

InSAR Basics
A Radar measures the round-trip If different satellite passes
delay between the sensor and
are exploited
the target The differential phase is
sentitive to the
topography and the
t2 displacement
produced
t1

c  tdelay 
R= ΔR12
2

R1 R2

ΔR12

Ground deformation
InSAR

Interferometric Phase

Applications
SAR
Image
Sensitive to the topography if
S1 Interferometric the spatial baseline is
Phase significant.

Sensitive to displacement
phenomena if the temporal
baseline is significant.

4p BnDr 4p BnDh 4p
Dj int = + + Dr + Dj APS + Dj N
SAR l r0 tan a l r0 sin a l
Image
S2 Interf
Flat-Earth Topo. Disp. APS+Noise
Phase
InSAR
Flow Mont Fuji Example
DEM Chart ALOS-PALSAR | Bperp: 300 m | Btemp: 46 days

4 Bn r 4 Bn h 4
int = + +  +  APS +  N
Rationale SAR1 SAR2  r0 tan   r0 sin  

Interf Flat-Earth Topo. Disp. APS+Noise


Interferometric pairs with:
Registration SAR Amplitude Interferogram Flat Earth Compensation
p p
• Large spatial baseline Common-Band Filtering

Interf. Generation
• Short temp. baseline
Flat Earth Removal

PhU -p -p
Sensitivity to PhU 3800 m
topography Geocoding
0

DEM
E
p N -30 0m
DInSAR
Flow Central Catalonia Example
Motion Chart Sentilel-1 | Bperp: 100 m | Btemp: 60 days

4 Bn r 4 Bn h 4
int = + +  +  APS +  N
Rationale SAR1 SAR2 DEM Orbits
 r0 tan   r0 sin  

Interf Flat-Earth Topo. Disp. APS+Noise


Interferometric pairs with: Registration Inv. Geocoding
Common-Band
Synthetic SAR Amplitude Interferogram p Flat Earth p
• Short spatial baseline Filtering
Topographic
Interf Component
Generation
• Large temp. baseline
+ -
+
-p -p
Sensitivity to PhU
ground motion Flat Earth Compensation p Topographic Model p Diff. Interferogram p
Geocoding Displ. Bowls

Displacement
-p -p -p
Earthquake Example

Lombok
Wrapped Phase
Interferogram associated with the 5TH August
2018 6.9 magnitude earthquake in Lombok

One full cycle in the interferometric fringes


represent a change of 28 mm between pre-
and post-earthquake SAR images processed
30/07/2018 - 05/07/2018
Surface displacement

Displacement Map
25 centimeters of ground surface uplift has
been detected
Earthquake Example

Lombok
PhU Phase
Interferogram associated with the 5TH August
2018 6.9 magnitude earthquake in Lombok

One full cycle in the interferometric fringes


represent a change of 28 mm between pre-
and post-earthquake SAR images processed
30/07/2018 - 05/07/2018
Surface displacement

Displacement Map
25 centimeters of ground surface uplift has
been detected
Vertical
Ascending Orbit Descending Orbit
East

vertical

vertical
North

North
Theoretical Basis

Up-Down  

East-West  

Decomposition Horizontal East Horizontal East

SAR sensors are only sensitive to


displacements produced along the Up-Down and East West Decomposition
LOS direction. Therefore, the

vertical
measured displacements in the LOS
are rarely the real ones, but a A Linear Equation System can be inverted
projection of them. to retrieve the up-down and east-west
components of displacement  ASC  DSC

If the Ascending and Descending


acquisition modes are combined the
 − cos( ASC ) sin( ASC )  cos( ASC )   VUD   VLOS , ASC 
up-down and east-west    = V 
VEW

displacement components can be  − cos( DSC ) sin( DSC )  cos( DSC   EW 


) V  LOS ,D SC  Horizontal

computed from the incidence angle VUD

(𝜃) and the track heading (𝛿) of both


acquisition geometries.
ASCENDING
Wrapped Interferogram 08/04/2016 - 20/04/2016 PhU Interferogram 08/04/2016 - 20/04/2016
 150

Kumamoto Earthquake Study Case

Up-Down
East-West

[rads]
Decomposition
This example shows the - -150
interferograms in Ascending and
DESCENDING
Descending acquisition modes
Wrapped Interferogram 27/03/2016 - 20/04/2016 PhU Interferogram 27/03/2016 - 20/04/2016
associated with the 16TH April 2016  150
7 magnitude earthquake in
Kumamoto.
The different ground motion
detected in Ascending and

[rads]
Descending modes indicates that
DESCENDING

a significant horizontal
displacement is present. In the
case of pure vertical displacement
Ascending and Descending
modes present similar patterns. - -150
Kumamoto Earthquake Study Case

Up-Down
East-West
Decomposition
UP-DOWN EAST-WEST

EAST-WEST DISPLACEMENT
Tenerife Island Example
Interferometric Phase Quality C-Band Envisat satellite

Coherence Amplitude

Small correlation length errors:


• Misregistration
• Geometric decorrelation
• Doppler decorrelation
Affect
• Volumetric decorrelation Coherence
Coherence Temp Baseline: 35 days Coherence Temp Baseline: 560 days
• Temporal decorrelation
• Thermal noise

Long correlation length errors:


• Orbital errors Not Affect
Coherence
• Atmospheric artifacts

Quality Phase degradation due to temporal decorrelation


Advanced InSAR Methods
Persistent Scatterers
Interferometry
One full cycle is equivalent to
half the wavelength
Persistent Scatterers Interferometry displacement but there are

InSAR Limitations p Topo


more fringes…

Motion
InSAR Limitations APS Noise
• Not all the points provide reliable information.
• Topographic residue in the differential phase.
• Presence of Atmospheric Phase Screen (APS).

- p
t0
t1
Persistent Scatterers Interferometry
tn
Exploitation of a multi-temporal stack of
differential interferograms to obtain the linear
and non-linear components of displacement,
the topographic error, as well as the
atmospheric artifacts. Coherent Pixels Technique
Persistent Scatterers Interferometry

InSAR Limitations
InSAR Limitations
Persistent Scatterers Interferometry

InSAR Limitations
InSAR Limitations
Persistent Scatterers Interferometry

Multi-Temporal Explotation
APS and displacement
components can be separated
exploiting the difference in their
statistics along the spatio-
t3 temporal domain:
t2
Atmospheric phase • Atmosphere: smooth in the
N
t1 delays have a spatial domain and
negative impact in completely random in time
displacement
• Displacement: smooth in the
estimation
temporal domain
t3
t2
t1 Only the Persistent Scatterers
ΔR12
(with good radar response
ΔR23 along time), fulfill the phase
quality requirements to be
S Time included in the PSI processing
Linear Estimation Block Inputs Pixel Selection Block Linear Block

Flow Triangulation

Chart Differential
Interferograms
Minimization

Geometrical Links Filtering


The Linear Estimation Masks
Block is devoted to
estimate the linear
contribution of the Persistent Integration
Coherence Mean Scatterers
differential phase, i.e, the
Maps Coherence Candidates
linear component of
Map Deramping
ground displacement and
the residual topographic
error component
Outliers Filtering
remaining in the g ³ g TH
differential DA £ DA,TH
interferograms. SAR Image Amplitude
Amplitudes Dispersion Linear Displacement
&
Quality Threshold Topographic Error Maps
Non-Linear Estimation Block Inputs Non-Linear Estimation Block

Flow APS → low-pass spatial behavior / white noise along time


DISP → high-pass spatial behavior / low-pass along time

Chart Differential
Interferograms
+
+
-
Residues
+
+
-
PhU + SVD

When Non-Linear
Displacement
non-linear LP Spatial Filtering

component +
+
comapared +
with the
APS Linear
Displacement Linear
APS

correlation &
Topographic
Phase
Model
window Error Maps

IS NOT Filtered
Time Series
Temporal
Smoothing
Time
Series
expected
Persistent Scatterers Interferometry

Products
Accumulated Maps Time Series
Dares Technology 0

Displacement [cm]
employs the Coherent
Pixels Technique (CPT).
-6

This PSI technique has


been developed by the
-12
Remote Sensing Displacement
[cm] Jul 2017
Jul 2017 Nov 2017 Feb 2018
Laboratory of the +12
Universitat Politecnica de +9 Oct 2017 Cross Section
Catalunya since 2002. +6
Feb 2018 A’ Displacement over time and topography
+3
+1
CPT allows the estimation -1

H + Disp x 1000 [m]


-3
of the linear and non- -6
linear components of -9
A
-12
displacement, the
topographic error, as well
as the atmospheric
artifacts.
Distance [meters]
Thank you

You might also like