1. Ionospheric mitigation schemes and their consequences for BIOMASS product quality O. French & S. Quegan, University of Sheffield, UK J. Chen, Beihang University, China ESA, Holland, 20 th May 2010 Task 200: Definition of Ionospheric Correction Schemes

2. Ionospheric Scintillation - Outline Summary of previous results Effect of 9m antenna GPS TEC data SAR simulator Correction strategies

3. Ionospheric Scintillation: Local Time Summary Scintillation effects severe for Boreal latitudes at all local times For temperate and equatorial zones, much reduced levels of scintillation in general......BUT: severe post-dusk scintillations in the equatorial zone for orbits with local time later than 20:00.

4. Effect of 9m Antenna Shorter antenna of 9m does not change the previous conclusions, in fact very little apparent effect on PSF. However, increased resolution drastically increases required computing power to simulate a given area. -Increased synthetic aperture; -Increased resolution on aperture; -Number of data points for 2D screen increases as d -4. Developed 1D slice simulator for which required number of datapoints scales as d -2.

5. GPS TEC data Spatial resolution at 2.5° latitude, 5 ° in longitude. Temporal resolution of 2 hours.

6. GPS TEC data

7. Still to do: perform comparison of IRI and GPS TEC data. Expect IRI to be “smoother” than GPS data - Compare IRI with window averaged GPS data GPS likely to be better data source - GPS data accurate within 3-5 TECU - IRI comprises a model fitted to experimental data

8. Correction Strategies 1D SAR simulator built that images point target Incorporates arbitrary phase perturbation across synthetic aperture - Polynomial, degree n; - 1D Ionospheric phase simulations. Simple multi-aperture mapdrift (MAM) correction strategy employed.

9. Correction Strategies: MAM Phase error modelled as Require N sub-apertures to estimate the a k ’s Decrease in SNR with increasing number of sub-apertures limits N Simulations performed for 2π phase error per term in polynomial, i.e. a k = 2π/L k

10. Correction Strategies: Mapdrift Uncorrected and corrected images for N=4

11. Correction Strategies: Mapdrift Uncorrected and corrected images for N=8

12. Correction Strategies: Mapdrift Uncorrected and corrected images for N=4 and with simulated ionospheric phase

13. Correction Strategies: Mapdrift Mapdrift ill-equipped to correct for ionospheric scintillation Probably due to high order of taylor expansion required to capture phase perturbations Alternative is PGA which does not rely on model for phase error

14. Ionospheric mitigation schemes and their consequences for BIOMASS product quality O. French & S. Quegan, University of Sheffield, UK J. Chen, Beihang University, China ESA, Holland, 20 th May 2010 Task 500: Definition of Calibration Scheme

15. Calibration - Outline Fujita three target approach Chen-Quegan compact polarisation approach Preliminary results for HV full polarisation approach Further work

16. Fujita Three Target Approach Assumptions crosstalk same for transmit and receive channel; discount all quadratic terms; zero-mean complex Gaussian noise. Five parameters (4 complex, 1 real): transmit and receive channel imbalance, F r and F t ; hv and vh crosstalk, C 1 and C 2 ; Faraday rotation, Ω. Masaharu Fujita, ‘Polarimetric Calibration of Space SAR Data Subject to Faraday Rotation – A Three-Target Approach’, IEEE 2005

17. Fujita Three Target Approach System model: M = R T FSFT + N where are transmit and receive distortion matrices, is Faraday rotation matrix. N is additive noise.

18. Fujita Three Target Approach Targets used Trihedral Dihedral PARCp

19. Fujita Three Target Approach Simulation parameters used (taken from original paper) F r = 0.9; F t = 0.9; C 1 = 0.1; C 2 = -0.1; Ω from 0 to 360°; Additive noise -20dB in power; No error in scattering matrices. Two estimators for each quantity, denoted a and b

20. Fujita Three Target Approach Faraday Rotation, Ω

21. Fujita Three Target Approach Receive channel imbalance, F R

22. Fujita Three Target Approach Transmit channel imbalance, F T

23. Fujita Three Target Approach vh crosstalk, C 1

24. Fujita Three Target Approach hv crosstalk, C 2

25. Fujita Three Target Approach vh crosstalk, C 1 hv crosstalk, C 2

26. Fujita Three Target Approach Introduce phase into single quantity, C 1 F r = 0.9; F t = 0.9; C 1 = 0.1 exp (iπ/12); C 2 = -0.1; Ω from 0 to 360°; Additive noise -20dB in power; No error in scattering matrices.

27. Fujita Three Target Approach Amplitude Phase vh crosstalk, C 1

28. Fujita Three Target Approach Amplitude Phase Transmit channel imbalance, F t

29. Fujita Three Target Approach Limitations of Fujita approach: Assumes equivalence of transmit and receive crosstalk Poor phase results Discounts all quadratic terms Does not consider large TEC/FR Incorporate GNSS TEC estimator

30. Chen-Quegan Approach Good results for compact polarisation System model: transmit in right-circular polarised, receive in linear H,V.

31. Chen-Quegan Approach Five parameters: - Circular crosstalk on transmit, δ c ; - Crosstalk on receive, δ 2 (hv) and δ 1 (vh); - Channel imbalance on receive, f; - Faraday rotation, Ω. δ c and f calculated first, ignoring quadratic terms All other quantities derived from these including quadratic terms Estimators optimised

32. Chen-Quegan Approach Uses at most 4 calibrators from - PARCx - PARCy Active - PARCp - Dihedral - Trihedral Passive - Gridded trihedral (x 2)

33. Chen-Quegan Approach Amplitude Phase Channel imbalance, f |  1 | = |  2 |= 0.1, |  c | =0.32, arg{  1 }= arg{  2 }= arg{  c }=0,  =  /4 arg{f} =  /3 |f | = 1.5

34. Chen-Quegan Approach Amplitude Phase Crosstalk,  1 |f | = 1.5, arg{f} =  /3, |  2 |= 0.1, |  c | =0.32, arg{  2 }= arg{  c }=0,  =  /4 arg{   }=0   |=0.1

35. Chen-Quegan Approach Excellent results for compact polarisation - Amplitude - Phase Reduced set of assumptions - Quadratic terms only initially discarded Need to extend to full HV polarimetric case

36. Full Polarisation Scheme System model: M = R T FSFT + N Where now are transmit and receive distortion matrices, is Faraday rotation matrix. N is additive noise.

37. Full Polarisation Scheme Reduced assumptions discount some quadratic terms; zero-mean complex Gaussian noise. Seven parameters (6 complex, 1 real): transmit and receive channel imbalance, F r and F t ; receive crosstalk, C r1 and C r2 ; transmit crosstalk, C t1 and C t2 ; Faraday rotation, Ω. Preliminary results with same values as previously, using PARCx, PARCy, GT 1 and GT 2 calibrators.

38. Full Polarisation Scheme Faraday rotation, Ω

39. Full Polarisation Scheme Amplitude Phase Transmit channel imbalance, F t

40. Full Polarisation Scheme Amplitude Phase Receive channel imbalance, F r

41. Full Polarisation Scheme Preliminary results show better (in)sensitivity to phase in crosstalk Future work Full analysis needs to be performed using method of Chen & Quegan. Incorporate GNSS FR estimator Consider suitability of calibrators