1. ElectroScience Lab A Study of Polarization Features in Bistatic Scattering from Rough Surfaces IGARSS 2011 Joel T. Johnson Department of Electrical and Computer Engineering ElectroScience Laboratory The Ohio State University Vancouver, Canada 26th July 2011

2. ElectroScience Lab Motivation l Increasing interest in bistatic microwave sensing (including out-of-plane geometries) motivates renewed examination of scattering effects l Full hemisphere integration of NRCS required for brightness temperature studies also motivates understanding bistatic properties l Out-of-plane geometries in particular have received little consideration in the literature with a few exceptions: Papa et al, IEEE Trans. Ant. Prop, Oct 1986, Hauck et al, IEEE Ant. Prop. Mag, Feb ’98, Hsieh& Chang, J. Marine Sci. Tech, vol. 12, 2004, Nashashibi & Ulaby, IEEE TGRS, June 2007, Pierdicca et al, TGRS, Oct 2008, Brogioni et al, Int’l J. Rem Sens, Aug 2010 l Pierdicca et al suggest some bistatic configurations for sensing soil moisture l Scattering effects that differ with polarization can be useful l Basic properties of scattering features investigated here analytically l Approach: investigate polarization properties of complete hemisphere bistatic pattern vs. incidence angle/surface roughness/permittivity l Rough surface only considered here: expand in the future to include volume scattering media

3. ElectroScience Lab Outline l Bistatic pattern properties from analytical methods –SPM –PO –SSA/RLCA l Comparison of analytical and numerical models l Further investigation of pattern properties l Summary

4. ElectroScience Lab Bistatic Pattern Properties with Analytical Methods: SPM l The Small Perturbation Method (SPM) is applicable for scattering from surfaces of small rms height compared to the EM wavelength and small slopes l Produces a perturbation series for scattered fields: first order only most typical Fields at first order have the form (incident pol , scattered pol  ): Kernel functions capture all polarization effects for slight roughness; explore as function of scattered polar (  s ) and azimuth (  s ) angles (0 inc. azimuth angle) Field scattered in direction Bragg Fourier Coefficient from surface roughness SPM kernel function: depends only on polarization, incident-scattering angle, and surface permittivity (not roughness)

5. ElectroScience Lab Bistatic Pattern Properties with Analytical Methods: SPM l Things to Notice: –HH always vanishes in the cross-plane (i.e.  s =90 o ) –VH/HV always vanish in plane (i.e.  s =0 o or 180 o ) –VV has a more complicated dependence on  s l Writing with it can be shown that has a minimum in azimuth at and that at the minimum is proportional to l Consequences: –VV goes to zero if A is real: real valued permittivities or approximately for large permittivity amplitude –Does not go to zero for A complex, but has a minimum vs. azimuth –“Null” locations trace out a curve in (k xs,k ys ) space that depends on incidence angle and permittivity l Approximately a shifted circle for large permittivity amplitude

6. ElectroScience Lab SPM Examples  i =20 o,  =10+i0.05, h= /20, L= /2, Gaussian correlation function  i =40 o,  =10+i0.05, h= /20, L= /2, Gaussian correlation function

7. ElectroScience Lab SPM Examples  i =40 o h= /20, L= /2, Gaussian correlation function, vs permittivity Same case, cuts vs. azimuth at  S =40 o  =3  =10+i0.05  =50+i40 VV min location and depth vary with  HH min location and depth fixed with 

8. ElectroScience Lab Bistatic Pattern Properties with Analytical Methods: PO l PO applicable for larger heights so long as slopes small (i.e. large scale features in surface), better near specular l PO polarization and permittivity dependence approximated at stationary phase point; NRCS then decouples roughness and polarization/permittivity effects in a product form l Influence of permittivity through reflection coefficients makes determination of minima in PO NRCS difficult; differs from SPM l In limit of large permittivity amplitude, HH and VV returns become identical –NRCS vanishes for both pols on contour in (k xs,k ys ) plane: Final term differs from SPM VV large |  | limit Same shifted circle as in SPM VV large |  | limit

9. ElectroScience Lab Bistatic Pattern Properties with Analytical Methods: SSA or RLCA l Small Slope Approximation (SSA) or Reduced Local Curvature Approximation (RLCA) reduce to SPM and PO in appropriate limits –Here using two field series terms (3 NRCS terms) from these methods –RLCA/SSA generally similar so only SSA shown in what follows –Analytic forms not simple; require numerical evaluation to examine l Should expect similar bistatic pol behaviors as SPM at small rms height that presumably will approach PO behaviors at larger heights l Differences between PO and SPM imply that “minimum” regions should depend on roughness –e.g. SPM null in HH at  s =90 o apparently “fills in” to no null in PO at larger roughness l All analytical methods considered in what follows are limited to “smoother” surfaces (h/L<~ 1/5) and non-grazing incident/scattering angles

10. ElectroScience Lab Numerical Method l Since higher order scattering effects may dominate when single scattering is weak (i.e. in “null” regions), important to compare with any more “exact” scattering method to verify predictions l Method of moments (MOM) used for this purpose in Monte Carlo simulation –3-D surface scattering problem, 64 realizations –32 x 32 wavelength surface, 512 x 512 points, 1 million unknowns –Point matching solution, iterative solver, Canonical grid acceleration –Run using supercomputing resources at Maui High Performance Computing Center –Use new approach by Saillard and Soriano, Waves Random Complex Media, 2011 to illuminate surface with plane wave without edge diffraction concerns –Isotropic Gaussian correlation function surfaces

11. ElectroScience Lab Comparison of MOM and SSA:  i =20 o,  =10+i0.05, h= /20, L= /2 l MOM predictions show “minimum” regions similar to SPM l Ratio of MOM to SSA NRCS values shows SSA provides good match l ;i

12. ElectroScience Lab Comparison of MOM and SSA:  i =20 o,  =10+i0.05, h= /20, L= /2 Zoom around “null” region for  S =40 o In plane versus  S to examine x-pol “null” region

13. ElectroScience Lab Comparison of MOM and SSA:  i =40 o,  =10+i0.05, h= /20, L= /2 l MOM predictions again show “minimum” regions similar to SPM l Ratio again shows SSA provides good match

14. ElectroScience Lab Comparison of MOM and SSA:  i =20 o,  =10+i0.05, h=0.1, L=1 l Locations of minimum regions coming closer to PO for HH l Larger differences with SSA but minimum regions still similar

15. ElectroScience Lab Comparison of MOM and SSA:  i =20 o,  =10+i0.05, h= /10, L= Zoom around “null” region for  S =40 o In plane versus  S to examine x-pol “null” region

16. ElectroScience Lab Comparison of MOM and SSA:  i =20 o,  =10+i0.05, h=0.3, L=2 l Locations of minimum regions coming closer to PO for HH l Larger differences with SSA but minimum regions still similar

17. ElectroScience Lab Variation with roughness from SSA:  i =20 o,  =3, L=  h varies from  to  Cuts in azimuth at  S =20 o l SSA captures “filling in” of minima as roughness increases, also transition from SPM-like to PO-like minima locations Increasing rms height

18. ElectroScience Lab Potential Applications l Previous bistatic soil moisture sensing study (Pierdicca et al, 2008) used AIEM with a “brute force” approach to study soil moisture sensitivity –Insights from this work may motivate renewed examination? l Since VV minimum region varies with permittivity, some sensitivity to permittivity should be expected l Different effects of surface scattering on polarizations may be useful for separating surface and volume effects –Like co-pol vs. cross-pol for backscatter but again with permittivity dependent minimum location

19. ElectroScience Lab Conclusions l Analytical properties of “null” regions in bistatic cross sections derived –SPM at first order: l HH vanishes in cross-plane, cross-pol vanishes in-plane l VV has a minimum in a curve in (kxs,kys) space, vanishes on this curve if permittivity is real or large amplitude –PO difficult to derive minima locations, but for large permittivity amplitude both HH and VV vanish on a (kxs,kys) curve distinct from that of SPM –SSA/RLCA capture transition between SPM/PO predictions and “filling in” of minima as roughness increases l MOM comparisons indicate that SSA captures these behaviors accurately at least for “smooth” surfaces l Insight into these behaviors may be useful in designing bistatic remote sensing systems (or interpreting insights from previous studies) l Bistatic polarimetry has also been explored (not discussed here)