1. A Novel Approach to the Automatic Detection of Subsurface Features in Planetary Radar Sounder Signals Adamo Ferro Lorenzo Bruzzone E-mail: adamo.ferro@disi.unitn.it Web page: http://rslab.disi.unitn.it

2. 2 A. Ferro, L. Bruzzone Outline Introduction 1 Aim of the Work 2 Statistical Analysis of Radar Sounder Signals 3 Automatic Detection of Basal Returns 4 Conclusions and Future Work 5

4. Main instruments:

5. Moon: ALSE and LRS

6. Mars: MARSIS and SHARAD

7. Their effectiveness lead to the proposal of new orbiting radar sounders, also for Earth science:

8. IPR and SSR for the Jovian Moons[1]

9. GLACIES proposal for the Earth[2]

10. Radar sounder data have been analyzed mostly by means of manual investigations.v Platform height Nadir Across track Range (depth) [1] L. Bruzzone, G. Alberti, C. Catallo, A. Ferro, W. Kofman, and R. Orosei, “Sub-surface radar sounding of the Jovian moon Ganymede,” Proceedings of the IEEE, 2011. [2] L. Bruzzone et al., “GLACiers and Icy Environments Sounding ,” response to ESA’s EE-8 call, 2010. Example of radargram (SHARAD)

12. Different frequency ranges.

13. Better spatial resolution.

14. Detection of buried objects (e.g., mines, pipes) which show specific signatures (e.g., hyperbolas).

15. Investigation of local targets vs. regional and global mapping.

16. Planetary radar sounding missions are providing a very large amount of data.

19. Statistical analysis of radar sounder signals.

20. Characterization of subsurface features.

21. Basis for the development of automatic techniques for the detection of subsurface features.

22. Automatic information extraction from radargrams.

23. First return.

24. Basal returns.

25. Subsurface layering.

26. Discrimination of surface clutter.6 A. Ferro, L. Bruzzone

28. Statistical analysis of radar sounder signals.

29. Characterization of subsurface features.

30. Basis for the development of automatic techniques for the detection of subsurface features.

31. Automatic information extraction from radargrams.

32. First return.

33. Basal returns.

34. Subsurface layering.

35. Discrimination of surface clutter.7 A. Ferro, L. Bruzzone

37. Number of radargrams: 7

38. Area of interest: North Polar Layered Deposits (NPLD) of Mars

39. Resolution: 300 × 3000 × 15 m (along-track × across-track × range)8 A. Ferro, L. Bruzzone -2500 m SHARAD radargram 1319502 -5500 m

41. NT: no target

42. SL: strong layers

43. WL: weak layers

44. LR: low returns

47. Rayleigh: simplest model, scattering from a large set of scatterers with the same size.

48. Nakagami: amplitude version of the Gamma distribution, has the Rayleigh has a particular case.

49. K: models the scattering from scatterers not homogeneously distributed in space, which number is a negative binomial random variable.

50. Distribution fitting performed via a Maximum Likelihood approach.

51. Goodness of fit tested by calculating the RMSE and the Kullback-Leibler distance (KL) between the target histogram and the fitted distribution.10 A. Ferro, L. Bruzzone Mean power Amplitude Shapeparameter Shapeparameter

52. Proposed Approach: Statistical Analysis, Fitting 11 A. Ferro, L. Bruzzone SHARAD radargram 1319502 No target Weak layers Strong layers Low returns Basal returns Summary

54. The parameters of the distribution describe statistically the characteristics of the target.

55. Noise can be modeled with a simple Rayleigh distribution.12 A. Ferro, L. Bruzzone

56. 13 A. Ferro, L. Bruzzone First return detection Calculation of KLHN Thresholding BR seed selection Region growing for m=2 to M Estimation of BR statistics Thresholding BR seed selection Region growing Region selection BR map generation Proposed Approach: Automatic Detection of BR Inputradargram KLHN map Initial BR map KL1 KLm BR map

58. Map of the KLHN:

59. Calculated for the subsurface area using a sliding window approach.

60. It represents a meta-level between the amplitude data and the final product.Inputradargram Local histogram Estimatednoisedistribution KLHN map Initial BR map KL1 KLm SHARAD radargram 1319502 BR map

62. Map of the KLHN:

63. Calculated for the subsurface area using a sliding window approach.

64. It represents a meta-level between the amplitude data and the final product.Inputradargram Local histogram Estimatednoisedistribution KLHN map Initial BR map KL1 KLm SHARAD radargram 1319502 BR map

66. Map of the KLHN:

67. Calculated for the subsurface area using a sliding window approach.

68. It represents a meta-level between the amplitude data and the final product.Inputradargram Local histogram Estimatednoisedistribution KLHN map Initial BR map KL1 KLm SHARAD radargram 1319502 KLHN map BR map

70. The initial BR map is created using a region growing approach based on level sets which starts from the seeds and moves on the KLHN map.Inputradargram Propagation Curvature Level set function KLHN map Initial BR map KL1 KLm KLHN map Initial BR map BR map

72. The procedure is repeated iteratively using lower threshold ranges for the KLHN map.

73. The new regions created during the iterations which are not statistically similar to the estimated BR distribution are deleted.Inputradargram KLHN map Initial BR map KL1 KLm Initial BR map BR map

75. The procedure is repeated iteratively using lower threshold ranges for the KLHN map.

76. The new regions created during the iterations which are not statistically similar to the estimated BR distribution are deleted.Inputradargram KLHN map Initial BR map KL1 KLm Step 2 BR map

78. The procedure is repeated iteratively using lower threshold ranges for the KLHN map.

79. The new regions created during the iterations which are not statistically similar to the estimated BR distribution are deleted.Inputradargram KLHN map Initial BR map KL1 KLm Step 3 BR map

80. Results: Automatic Detection of BR 21 A. Ferro, L. Bruzzone SHARAD radargram 1319502 SHARAD radargram 0371502 SHARAD radargram 1292401 SHARAD radargram 1312901

82. Selection of 3000 reference samples randomly taken in areas of the radargram where BR returns are (or are not) visible.

85. Statistical analysis of radar sounder signals.

86. It can support the analysis of the radargrams.

87. Different statistics / different targets.

88. Generation of statisticalmapsusefulto drive detectionalgorithms.

89. Novel technique for the automatic detection of the basal returns from radar sounder data using statistical techniques.

90. Effectively tested on SHARAD radargrams.

92. Estimation of local statistics using context-sensitive techniques for the adaptive determination of the local parcel size.

93. Develop a procedure for the automatic and adaptive definition of the parameters of the proposed technique.

94. Adapt the algorithm to airborne acquisitions on Earth’s Poles.

95. Other possible developments:

96. Integration of the automatic detection of linear interfaces and basal returns to higher level products.

99. AutomaticDetection of SurfaceClutter, Example 28 A. Ferro, L. Bruzzone SHARAD radargram 1386001 Coregistered surface clutter simulation Detected surface clutter map

100. Automatic Detection of the NPLD BR, Results 29 A. Ferro, L. Bruzzone Example of application to a large number of tracks -2300 20 -4000 0 Depth of detected BR fromdetected surface return [µs] Coverage of selected 45 tracks Mars North Pole topography [m] 180º 90º 270º 88º 86º 84º 82º 0º

101. Results: Automatic Detection of BR 30 A. Ferro, L. Bruzzone SHARAD radargram 1319502 SHARAD radargram 0371502 SHARAD radargram 1292401 SHARAD radargram 1312901

102. Modelparameters 31 A. Ferro, L. Bruzzone