airborne campaigns for pol-insar applications...

1

Folie [email protected] - 22.01.2007

Airborne Campaignsfor Pol-InSAR Applications Development

Irena Hajnsek, Alberto Moreira, Malcolm DavidsonGerman Aerospace Center European Space Agency

Folie 2Microwaves and Radar Institute

Outline

Why do we need airborne campaigns?First airborne demonstration of Pol-InSARWhat is the status of application development with Pol-InSAR?

Forest ApplicationAgricultural ApplicationIce and Snow Application Urban Area Application

Contribution of airborne campaigns to spaceborne

2


Why do we need airborne campaigns?

DLR‘s aircrafts

Requirements to an airborne systemFlexible and modular SAR systemSystem availabilityComplete operational processing chainFast data delivery High data quality

InnovationSpecifications needed for future satellite sensors Test advanced imaging modes (Pol-InSAR, digital beamforming, etc.)

DevelopmentDevelopment of algorithm for quantitative parameter estimationDevelopment of new application productsObservations to with which to calibrate or validate satellite retrievals

Data AvailabilityDetailed information's in critical areasKey information that cannot currently be measured from spaceYoung researcher education and preparation to satellite SAR sensors


Agri/ForestBretigny/Le LandesX-/ L-&P-band1999/2004RAMSES @ ONERA

ForestCamp LejuneP-band2000GeoSAR @ JPL

Temp. forestCapitolP-band2000/2002TopoSAR @ Intermap

ForestPortage Lake MaineC-,L-&P-band1993AIRSAR @ JPL

AgricultureFoulumC-band1995EMISAR @ DNSC (DTU)

X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site

Agriculture2005 (?)INGARA @ DSTO

Forest2000/2001PI-SAR @ JAXA

Temp. forest1998E-SAR @ DLR-HR

ApplicationFirst Pol-InSARdemo

Airborne System/Institution

First airborne Pol-InSAR campaigns

3


AIRSAR‘s first Pol-InSAR campaign @ JPL, USA

Objectives: To measure attenuation at different frequencies in the canopyUnderstand temporal decorrelation in the canopySee how the interferometric phase varied with polarimetric decompositions

Summary: Airborne InSAR experiments have shown good correlation at L-bandGeoSAR: Dual Pol-InSAR @ P-band – Campaign in Camp Lejune NC USA 2000

C-band L-band P-bandInterferograms

Correlation

Date: 1993







X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site







4


EMISAR‘s first Pol-InSAR campaign @ DNSC, Denmark

EMISAR DEM Coherence magnitude(RGB-coded Pauli basis)

Test site: Foulum, DenmarkCampaign purpose: Crop classificationAcquisition date: May 3, 1995Acquisition/processing: Technical University of DenmarkFrequency: C-band







X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site







5


E-SAR‘s first Pol-InSAR campaign @ DLR, Germany

HH

Oberpfaffenhofen19983D- Representation of Tree Height

Inte

rfer

omet

ricC

oher

ence

in T

hree

Pol

aris

atio

ns

Inversion/Validation

Inte

rfer

omet

ricPh

ase

in T

hree

Pol

aris

atio

ns







X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site







6


RAMSES‘s first Pol-InSAR campaign @ ONERA, France

σ

|γ|

Αrg(γ)

Date: Sep 1999Campaign: ORANGEObjective: Polarimetry studiesTest site: Bretigny, FranceFrequency: X bandImaging mode: Single pass







X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site







7


PI-SAR‘s first Pol-InSAR campaign @ JAXA, Japan

1.27GHz1.27GHzFrequencyMt. FujiTottori, TsukubaTest site

Surface deformation with control flight (5m tube)

Tree height, InSAR capability (~90m Bp)

Objective

Nov. 21 2006Oct. 8, 2000March 3, 2001

First experiments







X-band

L-band

L-band

Frequency

Australia

Tutori/Tsukuba

OP

Test site







8


TopoSAR (formerly AeS-1)Simultaneous X-band HH and quad-pol P-bandP-band bandwidth: 70 MHz @ 415 MHz centre frequencyFirst Pol-InSAR demonstration in 2000First operational Pol-InSAR campagain in July 2002:

Capitol Forest (Washington State, USA) for the purposeof bald-earth DTM generation beneath forest canopy

Gulfstream Commanderwith X- and P-band antennas

Aerial Photograph Pol-InSAR derived DTMP-band Magnitude (HH)

Capitol Forest example:

TopoSAR first Pol-InSAR campaign @ Intermap, Germany


ForestVegetation

•Forest Height•Forest Biomass•Canopy Extinction•Forest Structure•Underlying Topography

•Forest Ecology•Forest Management•Ecosystem Change•Carbon Cycle

AgriculturalVegetation

•Soil Moisture Content•Soil Roughness •Height of Vegetation Layer•Extinction of Vegetation Layer•Moisture of Vegetation Layer

•Farming Management•Water Cycle•Desertification

Ice and Snow

•Topography•Penetration Depth •Depth of Snow-Ice Layer•Extinction•Density of Ice Bodies

•Ecosystem Change•Water Cycle•Water Management

•Urban MonitoringUrban Areas•Geometric Properties•Dielectric Properties

9


1994: SIR-C / X-SAR acquires the first POL-InSAR data set1996: First publication on Pol-InSAR.

SIR

-C/X

-SA

R /

Test

Site

: Kud

ara,

Rus

sia

1998: First Pol-InSAR forest height estimation.

1994: SIR-C / X-SAR acquires the first POL-InSAR data set1996: First publication on Pol-InSAR.

First Pol-InSAR Demonstration: Forest Application

L-band / Pauli RGB


Baseline=40m HH HV VV

HH HV VVBaseline=25m

E-SAR/Test Site: OberpfafenhoffenL-band

Azim

uth

Range

10


Three main forest types have been investigated:

• Temperate forest• Boreal forest• Mediterranean forest

1. Algorithm improvement2. Model validity 3. Validation 4. Product development

Airborne Campaigns with E-SAR @ Forest applications

Fichtelgebirge / GermanyCorridor / SpainBayrischer Wald / GermanyTraunstein / GermanyOberpfaffenhofen / Germany

Campaign Objectives:E-SAR campaigns over different forest types in Europe for forest height retrieval:


0 10 20 30 40

Forest Height (in Situ) [m]

Pic.abiesPi.silv.

Ab.piceaPseu.menz

Lar. decGg.sylv.

Querc.rab.Frax.exc.Bet.pend.Rob.pseu.Por.rob.

0 10 20 30 40 50

600

500

400

300

200

100

0

Forest Height [m]

Fore

st B

iom

ass

[t/ha

]Fo

rest

Bio

mas

s [t/

ha]

Airborne Campaigns @ Forest applications

11


Kalimantan - Indonesia

Tropical forest applicationsCampaign Objectives:

1. Investigation of L- and P-band overtropical forest for forest heightestimation (visebility of the ground, validity of the RVoGM etc.)

2. Is there a relation of forest height and forest biomass over tropical forest?

3. Is there an empirical relation between the radar backscatter and the forest

biomass using L- and/or P-band?


MAWAS: Tropical forest heightRGB Image X/L/P Band

Tropical Forest Height from Pol-InSAR

y

z

x

n

r

bl

h

⎥⎦

⎤⎢⎣

⎡=

VVVH

HVHH

SSSS

S ][ 1

⎥⎦

⎤⎢⎣

⎡=

VVVH

HVHH

SSSS

S ][ 2

1 2

Pol-InSAR

Forest height over tropical forest in Indonesia

az

rg

12


Preliminary results: Tropical forest applications1. Both frequencies L- & P-band are suitable for

forest height estimation

2. Research is still ongoing and a definite answer is pending – relation between height/biomass

3. Six different definitions of biomass have been found in the literature and have been used for comparison. No significant relation between biomass and radar backscatter

were found (270-390 t/ha)

P – Band (blue) L – Band (red) See presentation by Florian Kugler

0

5

10

15

20

25

30

35

40

45

H10

0[m

]

Top height of transects : 26m123

54

6

87

910

123

54

6

87

910

Plot design of ground truth:20 squares of 10m x 10m in two transects

L- against P-band derived heights

whole scene ground plot


Boreal Forest 2003 @ Helsinki AreaCampaign Objectives:1. Investigation of Pol-InSAR in L-band

over boreal forest areas.

2. Comparison of Pol-InSAR derived forest heights with HUT-SCAT data HUT-SCAT Profile

Cooperation with HUT

13


Forest height @ different frequencies …

P-ba

nd

L-ba

nd

C-b

and

@ 30º

50% HH & VV

0 0 0 1 20 0 1 4 0 0 16 0 0 1 8 00 2 0 0 0 2 20 0 2 4 0 0 26 0 0 2 8 00 3 0H u ts c a t lo o k

Forest Height @ L-band & HUTSCAT Height Profiles

Forest Height @ X-band & HUTSCAT Height Profiles

E-SAR / Test Site: Helsinki, Finland

L-ba

nd

)()(~

)exp()(~wm1wmγφiwγ V

0 r

rr

++

=V0 γφiwγ ~)exp()(~ =

r

X-ba

nd

Short summary:

1. Significant correlation between L-band & HUT-SCAT derived forest heights

2. Significant correlation between X-band & HUT-SCAT derived forest heights


ForestVegetation






Ice and Snow




14


E-SAR’s Agricultural Campaign in Tunesia 2005Airborne Campaign: AQUIFEREX within AQUIFER

support the national authorities and international institutions with Earth Observation (EO) based technology to better manage internationally shared water resources and aquifers strengthen overall and integrated water management practices build-up of an independent service provision capacity to ensure local service delivery after the project cycle to achieve the longer-term goal of service sustainability

Campaign ExecutionNovember 2005South Tunesia (Ben Gardane,GabesRegion)ImportantobservationParameter:Land-cover and use & soil moisture

E-SAR: C-band quad-pol

E-SAR: L-band quad-pol

AVIS(rgb=659, 550, 477 nm)


surface

scatteringplane

sensor

β

LOS

E-SAR / Test Site: Aquiferex, Tunisia

L-band / Pauli RGB Vol. Soil MoistureDielectric Constant

T11 T12

T21 T22

T33

T11 T12

T21 T22

T11

T22

T33

= +

X-Bragg Bragg RoughnessTerm

Pol.

Coh

. Mat

rix

Bare Surfaces: Soil Moisture Estimation @ AQUIFEREX

15


Agriculture Vegetation @ Alling/Germany 2000

SAR Image @ L-band 3-D Height Map

Differential Extinction0 dB

40 dB

)()()()exp()(

~~

wm1wmwγφiwγ V

0 r

rrr

++

=

0V I

Iwγ =)(~ r

∫

∫

⎟⎟⎠

⎞⎜⎜⎝

⎛=

⎟⎟⎠

⎞⎜⎜⎝

⎛=

V

V

h

0 00

h

0 0z

dzθ

zwσ2I

dzθ

zwσ2zκiI

'cos

')(exp

'cos

')(exp)'exp(

r

r

Interferometric Coherence:

Test

Site

: Kue

ttigh

offe

n, S

witz

erla

nd

E-SAR / Test Site: Alling, Germany


AGRISAR: Soil Surface and Crop Parameter (April-Aug.2006)1. Building up a data base for agricultural parameter

estimation over a whole vegetation growth period -has been started mid April 2006

2. support the space segment activities at ESA with respect to the Sentinel Program

answering open questions concerning system constellation (single, dual, qual polarisation, revisit time, etc)

• 16 Radar data acquisition flights during three months

PolarimetryFrequencyX-band interferometry

Soil Moisture Station

Bowen-Ratio Station

Laser-Profiler

16


AG

RIS

AR

@ L

-ban

dfo

r3 in

tens

ive

perio

ds

RGB Pauli

19/04/06 06/06/06 05/07/06


ForestVegetation






Ice and Snow




17


Snow First Pol-InSAR Snow Experiment in Austria

2004E-SAR: Kuehtai / Austria 2004

Cooperation with University of Innsbruck

X-band DEML-band / HH ImageHH-HH CoherenceL-band / Baseline 1

HH-HH CoherenceL-band / Baseline 2

Snow appears as a Volume Scatterer @ L-band

CampaignObjectives:

Investigation of Pol-InSAR forsnowcharacterisation


Snowθo

z0

z0 + hv

θr

)sin()sin( 0

rθθn =′

Baseline 1 (20M): Δφ=17° -> Δz=1.48m

Baseline 2 (40M): Δφ=28° -> Δz=1.22m

SnowSnow depth can be potentially estimated

HH-HH Coherence XX-XX CoherenceL-band / HH Image VV-VV Coherence

18


SVALEX 2005 – Svalbard Airborne ExperimentDLR-HR (Microwaves and Radar Institute)

* Pol-InSAR measurements over land ice* penetration depth at different frequencies* retrieval of glacier topography* mapping of internal ice structures

AWI-Potsdam (Alfred-Wegner Institute)* measurements of optical, chemical, and* physical properties of Arctic aerosols, in particular Arctic haze

AWI-Bremerhaven (Alfred-Wegner Institute)* boundary layer meteorology * transfer of atmospheric momentum to sea ice* use of radar imagery to quantify ice ks

Sensors:1. E-SAR Radar System2. Meteopod (wind velocity,

humidity, temp.)3. Laser Altimeter4. KT15 (surface temp.)5. Linescan camera

AWI-DLR Team

Sea/Land iceSAR data

acquisition


SVALEX 2005 – Svalbard Experiment 05Sea Ice (L-band quad pol)

azimuth

range

Land Ice L-band

Land Ice P-band

Airborne Radar Campaign at Spitzbergen• over Sea Ice in X- & L-band quad pol• over Land Ice (Glacier) in Pol-InSAR L- and P-band

19


SVALEX 2005 – Svalbard Experiment 05

Sea Ice & Meteopod

Corner Reflector @ Summit

Meteo Station

Preliminary Summary• Investigation of penetration depth at different frequencies,

where the corner reflector are used as a reference point. First results show that lower penetration occurs as expected in X-band, L-band and P-band.

• Implementation and modification of coherent scattering models for the characterisation of the ice volume and Pol-InSAR analysis at different frequencies showed good agreement with the observables.

See presentation by Jayanti Sharma


ForestVegetation






Ice and Snow




20


Campaigns over Urban Area over Munich 2005/6 Campaign Objective:1. Identification of Coherent Scatterers

at different Frequencies

2. Characterisation of Coherent Scatterer (dielectric and geometric properties)

Dihedral Size: Vertical: 80 x 80 cmHorizontal: 80 x 80 cm

Orientation: 1th dihedral: 0 °2nd dihedral: 5 °

Acquisition:13. Oktober 2005

Tracks: 0, 30, 90, 270m

3 independent

Ascending

Descending

Munich


Urban Areas: Coherent Scatterers @ Munich 2005/6

R: L-bandG: C-bandB: X-band

R: HH-VVG: HVB: HH+VV

2005-6: First CS’s dedicated experimentE-SAR / Test Site: Munich, Germany

21


Coherent Scatterers

Red: Dihedrals

Green: Dipoles

Blue: Flat Plates

Amplitude Entropy Alpha angle

06.2006

See presentation by Kostas Papathanassiou & Luca Marotti


Looking Forward …Development of parameter estimation methodology / algorithms: Snow / Ice / Agriculture …Information product definition and validation / Projection of product spec’s onto system design. Exploration of new and innovative observation spaces: From E to F ( SAR )

From S to Q ( Pol )From R to S ( Pass )From M to B ( Static )

ALOS / Pal-SAR TerraSAR-X RadarSAT-2 Tandem-X

airborne campaigns for pol-insar applications...

Documents