Calibration Assessment of RADARSAT-2 Polarimetry Using High Precision Transponders

Size: px

Start display at page:

Download "Calibration Assessment of RADARSAT-2 Polarimetry Using High Precision Transponders"

Violet Waters
5 years ago
Views:

1 Calibration Assessment of RADARSAT-2 Polarimetry Using High Precision Transponders R Touzi, S Côté, RK Hawkins CCRS/CSA Acknowledgments S Nedelcu (CCRS) S Muir (CSA) 1

2 Outline-Polarimetric RADARSAT-2 Independent assessment of RADARSAT-2 calibration parameters: Based on transponder measurements Takes advantage of higher signal levels at very good S/N conditions (lower than -33 db) LUT effects on RADARSAT-2 Calibration data quality 2

3 31 MODES BEAM MODE θ ( ) SWATH (km) RESOLUTION (m) R X Az LOOKS Polarization: HH+HV+VV+VH Fine quad-pol X 9 1 X 1 Standard quad-pol X 28 1 X 4 3

4 Upgraded for Operation at Various Polarizations High cross-pol isolation One antenna transmit One antenna receive 4

5 Ottawa 5

6 Results: FQ20, Ottawa, July 9,

Assessment of Transponder Polarization Orientation Transponder Scattering matrix as a function of polarization orientation (CEOS 09) cos sin cos sin Trans Trans θt θt Shh Svh θr θr [ S ( θ )] = σ

7 Assessment of Transponder Polarization Orientation Transponder Scattering matrix as a function of polarization orientation (CEOS 09) cos sin cos sin Trans Trans θt θt Shh Svh θr θr [ S ( θ )] = σ sinθt cosθ t Shv S vv sinθr cosθ r V hh hv Measured Voltage: [ ] jφ = Trans 3 Ae S ( θ ) V vh V V vv T 1 δ1 δ F δ 4 δ F 2 HV Transponder VH Transponder : : S S hv vh = 1; = 1; S S vh hv = = S S hh hh = = S S vv vv = = 0; θ r 0; θ t 0 ; θ t 0 ; θ r 0 V 0 V vh hv = = 0 0 7

Uncalibrated Data : Calibrated for Antenna Gains and slant

8 Assessment of RADARSAT-2 Antenna Parameters using Distortion Matrix-Off Product RADARSAT-2 Distortion off Product Uncalibrated Data : Calibrated for Antenna Gains and slant range (R 3 ) variations Transmitter-Receiver Distortion matrix not corrected 8

9 Transponder HV (Oct 2008) Uncalibrated FQ10_HV Distortion off Transponder Well Oriented (Θ=0) HV VH 9 VH Trans close to zero Peak to Clutter=-5dB

10 FQ10 Transponder HV (Oct 2008) Visible cross-talk at HH and VV HH VV 10

11 RADARSAT-2 Parameters using FQ6 and FQ10 HV Transponder Transmitter and receiver channel imbalance very good (internal transmitter-receiver measurements): F1 1 and F2 1 Stable phase difference (φ F2 - φ F1 ) Measured cross-talk (FQ6, FQ10 and FQ1) δ1-35 db and δ4-32 db Consistent with incidence angle (within 0.5 db and 30 ) Calibration independent of incidence angle One transmitter-receiver Distortion matrix for the 30 right-looking modes (and One for 30 left-looking) 11 Vhh Vhv jϕ δ Trans δ3 Ae S ( θ ) Vvh V = vv δ1 F 1 δ4 F 2

12 Transponder VH (Oct 2008) Uncalibrated FQ20 (Distortion off) Visible Cross-talk at HH and VV HH VV 12

13 FQ20 RADARSAT-2 Parameters Uncalibrated Data New method (VH Transponder + Forest) applied to estimate the SAR parameters: Transmitter and receiver Channel well balanced and phase difference stable (internal transmitter-receiver measurements): F1 1 and F2 1 Low cross-talk: δ 2-40 db and δ 3-38 db Note the sensitivity of transponder here 13 Vhh Vhv 1 δ2 1 δ jϕ Trans 3 Ae S ( ) Vvh V = θ vv δ1 F 1 δ4 F 2

14 Simple Calibration Method with No cross-talk removal Cross-talk lower than -32 db can be ignored RADARSAT-2 meets CEOS CAL-VAL requirements Antenna Gains and slant range variations removed Transponder RCS (within 0.5 db) Channel balanced: F 1 1 and F 2 1 Inter-channel relative phase requires calibration CR or Transponder +Azimuthally symmetric target: F1/F2 & φ F2 - φ F1 ; φ F1 and φ F2 (Zebker 90) 14

15 FQ Uncalibrated HH HV Inter-channel calibration: P=0.3 db Φ=200 15

16 Phase Calibrated FQ HH HV Inter-channel calibration: P=0.3 db Φ =5 16

17 In agreement with MDA Calibration FQ20, Ottawa, July 08 HH HV Inter-channel calibration: P=0.3 db Φ =5 17

18 FQ Uncalibrated VV VH Inter-channel calibration: P=0.3 db Φ=200 18

19 Calibrated phase within 5 degree Cross-talk lower than -32 db VV VH Inter-channel calibration: P=0.3 db; Φ =5 19

20 VV Canada Centre for Remote Sensing Earth Sciences Sector FQ20, Ottawa, July 9 MDA Calibrated Data VH Inter-channel calibration: P=0.3 db; Φ =5 20

21 Assessment of MDA Calibration residual cross-talk Using Azimuthally Symmetric target FQ20: δ 3 = db, δ 4 = db F 2 /F 1 =0.987 (-0.12 db) φ F1 - φ F2 =1 FQ6: δ 3 = db, δ 4 = db F 2 /F 1 =0.99 (-0.08 db) φ F1 - φ F2 =0.4 Residual cross-talk lower than -39 db Vhh Vhv 1 δ2 1 δ jϕ Trans 3 Ae S ( ) Vvh V = θ vv δ1 F 1 δ4 F 2 21

22 Effect of LUTs on RADARSAT-2 Polarimetric Data Quality 22

23 Applied Gain (db) LUT 16-bit Data are processed as 32-bit floating point complex data (I&Q) LUT applied to create 16-bit signed short integer Reduce Data Volume and presents a radiometrically flat product Canada Centre for Remote Sensing Earth Sciences Sector o β = Land Sea Ice Mixed Constant Beta Constant Sigma Constant Gamma Point Target Calibration-1 Incidence angle (deg) 1/ 2 I DN = fix( I FP GLUT ) 2 Q = fix( Q G 1/ ) DN I 2 DN + Q G LUT 2 DN FP LUT 23

24 LUT effect Canada Centre for Remote Sensing Earth Sciences Sector Coarse quantification of low-medium backscattering level Saturation of high backscattering level 24

25 LUT effect on SLC Canada Centre for Remote Sensing Earth Sciences Sector LUT applied separately on I and Q Potential phase distortion Radiometry: medium-low signal σ ο + texture affected HH-VV and mainly HV Scattering mechanisms: coarsely quantified Medium LUT misses double and single bounce scattering Low LUT misses scattering heterogeneity (Texture) 25

26 Selection of the best LUT suitable for your Application Good for applications of reasonable wide dynamic of scattering mechanisms (such as Agriculture) Signal variations fits within the 16-bit LUT Wetland and Urban scattering of high dynamic range Cannot be presented with 16-bit I and Q Saturation of single and double bounce scattering + phase distortion HV of low signal set to zero + phase distortion Data re-processed with several LUTs Reconstruction of floating point 32 bit I and Q necessary 26

27 Urban Application: LUT Point target/land up to 6 db error (Saturation) Building Point Target LUT Land LUT 27

28 Urban Application: LUT Point Target/Land db error in HV, HH and VV Farm Field Point Target LUT Land LUT 28

29 Cal1 and Reconstructed Cal1-Mixed (Calmix) Two LUTs Calibration1 and Mixed combined to reconstruct more accurate SLC (Calmix) 29

30 Radiometric Error related to LUT Calibration1 vs Mixed VH e (db) 30

31 Radiometric Error related to LUT Calibration1 versus Mixed HHe(dB) VVe(dB) 31

32 Phase Error related to LUT Calibration1 versus Mixed HV HH 32

33 Entropy; Reconstructed versus Mixed LUT Cloude-Pottier Entropy Reconstructed versus Mixed 33

34 Touzi Scattering phase Reconstructed versus Mixed 34

35 Conclusion Canada Centre for Remote Sensing Earth Sciences Sector Independently validated some modes of RADARSAT-2 as an excellent SAR with: Unique operational polarimetric capabilities -multiple incidence angle and wide swath (50 km thanks to T. Luscombe) Low noise floor: -38 db Oil slick, Ice applications Cross-talk << -32 db; so, RADARSAT-2 comfortably meets the CEOS Cal-Val requirements particularly important for marine apps Poor LUT choices can significantly affect the polarimetric properties Polarimetric RADARSAT-2 Floating 32 bit SLC ALOS is providing the polarimetric data in floating format 35

SENTINEL-1 Toolbox. Polarimetric Tutorial Issued March 2015 Updated August Luis Veci

SENTINEL-1 Toolbox. Polarimetric Tutorial Issued March 2015 Updated August Luis Veci SENTINEL-1 Toolbox Polarimetric Tutorial Issued March 2015 Updated August 2016 Luis Veci Copyright 2015 Array Systems Computing Inc. http://www.array.ca/ http://step.esa.int Polarimetric Tutorial The goal