polarimetric remote sensing of soil moisture within tereno · tereno advisory board meeting,...
TRANSCRIPT
TERENO Advisory Board Meeting, September, 26-27, 2011, Rur Catchment, [email protected], [email protected]
Polarimetric Remote Sensing of Soil Moisture within TERENOThomas Jagdhuber1 & Irena Hajnsek1,2
1Microwaves and Radar Institute, Pol-InSAR Research Group, DLR2Institute of Environmental Engineering, Earth Observation, ETH Zurich
0 10 20 30 40 50[vol. %]
Page 2
E-SAR and F-SAR – Versatile Tools for Airborne Active Microwave Remote Sensing
E-SAR were and F-SAR is operated onboard DLR’s DO228-212 D-CFFU by the Microwaves and Radar Institute in cooperation with DLR’s Flight Facilities based in Oberpfaffenhofen.
Page 3
E-SAR Flight Campaign May 2008Strip 1: Selhausen – Hückelhoven, Length ca. 28kmStrip 2: Hückelhoven – Wüstebach, Length ca. 61kmStrip 3: Wüstebach – Selhausen, Length ca. 40km
Frequency: L-band (fully polarimetric)
rRollesbroich
Test sites
SelhausenStrip 1
Strip 2
Strip 3
Page 4
Ground Measurements 2008: Soil Moisture [vol. %]
Selhausen test site winter wheat fields winter barley field
sugar beet fields forested area
Page 5
vegetationground
Polarimetric Decompositions for Soil Moisture Inversion
Surface Dihedral VolumeRecorded signal
Removal of vegetation component and inversion for soil moisture
groundvegetation
- = +
Soil moisture [vol. %]
= + +
Page 6
vegetation
soil surface
dihedral
-30 -22 -14 -6 2 10[dB]
FZ JülichFZ Jülich
Volume intensity component from standard decomposition
Soil Moisture Estimation under Vegetation with Polarimetric SAR?
One solution: Estimation with polarimetric decompositions
Separation of vegetation from soil surface and dihedral
Normalized Pauli-RGB image of Selhausen test sitePhysically-constraint hybrid polarimetric decomposition
No physically constraint volume intensity
Page 7
Physically Constraint Hybrid Polarimetric Decomposition
θlocForward modelling:
surface component (αB )
αB
Calculus of data stack withphysically constraint fvp for
different εS -levels (using αB )
εS -level: 5
εS -level: …
εS -level: 10εS -level: 15
fvpεs…fvpεs15
fvpεs10fvpεs5
+
αs
αd
surface
dihedral
Hybrid polarimetric
decomposition&
Determination ofεS -level
Local incidence angle
Page 8
Inversion for Soil Moisture from Surface Component
( ){ } svcbestfs fmeanestvc
ααεαααε ⇒⇒⇒⇒→−
°<0)(
25)(11
( ){ } svcbestfs fmeanestvc
ααεαααε ⇒⇒⇒⇒→−
°<0)(
25)(11
{ }sms αα −minMinimization
εS
transfer function
of Topp et al.Soil moisture [vol. %]
Surface scattering componentfrom hybrid polarimetric
decomposition
αs
Polarimetric SAR data
⎟⎟⎠
⎞⎜⎜⎝
⎛+−
=HH
HHsm R
R
VV
VV
RRarctanα
Surface scattering with θloc and a variety of εS
RHH , RVV = f(εS ,θloc )
Bragg scattering model
Page 9
Results for Soil Moisture Inversion (Selhausen)
0
10
20
30
40
50
[vol. %]
Spatial standard deviation of inversion (10x10 pixels)
rang
e
azimuth
Estimated Soil Moisture (εS -level = 15)
Page 10
Results for Soil Moisture Inversion (Rollesbroich)
0
10
20
30
40
50
[vol. %]
Spatial standard deviation of inversion (10x10 pixels)
rang
e
azimuth
Estimated Soil Moisture (εS -level = 15)
Page 11
Validation for 3 Corp Types
Validation box:13 x 13 Pixels
winter wheat winter barley
sugar beet
Page 12
Validation for all fields of Selhausen
Validation box:13 x 13 PixelsRMSE=5.80
Page 13
Summary and OutlookSoil moisture estimation under vegetation cover has been performed:
The hybrid polarimetric decomposition including physical constraints has been successfully applied on the Rur Observatory (subsets Selhausen & Rollesbroich)
The developed approach results in a inversion rate of >90% for both test sites over agriculture (forest still critical)
Soil moisture inversion results are obtained for a variety of crop types with a RMSE < 6vol. % and a STDDEV < 3vol. %.
Further investigations:
Hybrid model improvement (focus on the volume part)
Investigation on steep incidence regions
Investigation of dihedral scattering component (αd) for soil moisture inversion
TERENO Advisory Board Meeting, September, 26-27, 2011, Rur Catchment, [email protected], [email protected]
TERENO F-SAR Airborne Campaign 2011 @ Rur, Bode and Ammer catchmentsoperated by DLR, FZJ, UFZ and KIT
Page 15
F-SAR L-Band Subsystem (2011)
up/down converter with chirp
antenna front end
system control
power supply
high power amplifier (HPA)
Page 16
F-SAR High-Gain L-Band Antenna
L-band without cowling
Page 17
Overview of TERENO Campaign 2011DLR‘s SAR Sensor: F-SAR
Frequency L-bandFully polarimetric Spatial Resolution (r/a): 2mx0.6mDate: KW 21-22 (23.05.-03.06.2011)
TERENO ObservatoriesAmmer - KITBode – UFZ/WESSRur – FZJ
Ground Measurementswere conducted by the research institutesof the observatories.DLR supported for theAmmer and the Bodecatchment.
Catchment DateData
acquisition @local time
No. of scenes
Rur 30/05/2011 09:35-12:26 13
Bode 31/05/2011 09:12-10:41 8
Ammer 07/06/2011 10:26-11:33 6
Catchment Date Ground measurements
Rur 30/05/2011
Soil moisture (cluster, mobile probes), soil roughness (Laser-
profiler), vegetation height, biomass, LAI
Bode 31/05/2011
Soil moisture (mobile probes, EMI), soil roughness (Laser-
scanning)vegetation height, phenology,
biomass, VWC, LAI
Ammer 07/06/2011 Soil moisture (mobile probes)
Page 18
F-SAR Flight Strip
near range
corner reflectors
TanDEM-X Strip
Test site – Bode WatershedFlight strip of F-SAR: 10 x 3 km Field measurements by UFZ/WESS/DLR:
Soil moisture (FDR [6cm], EMI [experimental] )Soil roughness (Laser scanning [experimental])Vegetation (height, phenology, biomass, VWC, LAI)
Zoom on corner reflector
Page 19
Test site – Ammer/Rott WatershedFlight strip of F-SAR: 11 x 3 kmField measurements by KIT/DLR: Soil moisture(mobile FDR/TDR probes)
TanDEM-X Strip
F-SAR Flight Strip
near
rang
e
Rott catchment
Tracks for mobilesoil moisture probes
Page 20
TanDEM-X Strip
F-SAR Flight Strip
Test site – Rur WatershedTriangular Flight configuration:Yellow AOI ares: 5 x 3 km (2) and 10 x3 kmField measurements by FZJ: Soil moisture, Vegetation
SoilNet (grassland (Rollesbroich), forest (Wüstebach))Active/passive arc experiment,Mobile TDR/FDR probes (Merzenhausen)
Triangular flight configuration
Page 21
Acknowledgement to the
Campaign Team