Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Acoustic and topographic sediment classification using high-resolution multibeam echosounder: examples from the Colorado River in Marble Canyon

Acoustic and topographic sediment classification using high-resolution multibeam echosounder: examples from the Colorado River in Marble Canyon

2nd Multibeam in Rivers Workshop, March 2015

D612e176cb18a5190a722dc0313cb583?s=128

Daniel Buscombe

March 25, 2015
Tweet

Transcript

  1. None
  2. Multibeam echosounder measures depth & echo strengths High-resolution soundings ...

    Submerged aquatic vegetation Definition of physical habitats Bed substrate classification Appeal of backscatter = hardness + roughness (?) Topography, Acoustic Backscatter, or both? Bathymetry at RM 61, Marble Canyon, May 2014
  3. Multibeam echosounder measures depth & echo strengths High-resolution soundings ...

    Submerged aquatic vegetation Definition of physical habitats Bed substrate classification Appeal of backscatter = hardness + roughness (?) Topography, Acoustic Backscatter, or both?
  4. Multibeam echosounder measures depth & echo strengths High-resolution soundings ...

    Submerged aquatic vegetation Definition of physical habitats Bed substrate classification Appeal of backscatter = hardness + roughness (?) Topography, Acoustic Backscatter, or both?
  5. Multibeam echosounder measures depth & echo strengths High-resolution soundings ...

    Submerged aquatic vegetation Definition of physical habitats Bed substrate classification Appeal of backscatter = hardness + roughness (?) Topography, Acoustic Backscatter, or both?
  6. RM30, Marble Canyon, August 2013 11 million soundings 25 cm

    grid
  7. RM30, Marble Canyon, August 2013 11 million soundings 25 cm

    grid
  8. Median Backscatter

  9. Stochastic geometry. 1) Spectral Strength P1(K) = ω1 (h0|K|)γ1 measure

    of power at low frequencies
  10. Stochastic geometry. 2) Spectral Width P1(K) = ω1 (h0|K|) γ1

    measure of the rate of decay in power as a function of increasing frequency. The range of frequencies necessary to describe the data
  11. Stochastic geometry. 3) Spectral Variance σ2 1 = 2 K0

    P1(K)dK overall power in the spectrum over all frequencies
  12. Sediment Classification using Backscatter

  13. Sediment Classification using Backscatter

  14. Sediment Classification using Backscatter

  15. Summary so far ... Power law: no significant concentration of

    spectral power in any particular frequency band Slopes range between -0.5 and -3.5. The scaling relationship between amplitude and wavelength does not hold over all wavelengths Rate at which backscatter amplitude declines relative to wavelength is smaller for hard, rough surfaces compared with relatively smooth and soft surfaces. Hypothesis 1: Backscatter responds to grain size which is more rapidly varying than surface topography. Hypothesis 2: These measures of backscatter respond to both grain size and surface topography.
  16. Summary so far ... Power law: no significant concentration of

    spectral power in any particular frequency band Slopes range between -0.5 and -3.5. The scaling relationship between amplitude and wavelength does not hold over all wavelengths Rate at which backscatter amplitude declines relative to wavelength is smaller for hard, rough surfaces compared with relatively smooth and soft surfaces. Hypothesis 1: Backscatter responds to grain size which is more rapidly varying than surface topography. Hypothesis 2: These measures of backscatter respond to both grain size and surface topography.
  17. Summary so far ... Power law: no significant concentration of

    spectral power in any particular frequency band Slopes range between -0.5 and -3.5. The scaling relationship between amplitude and wavelength does not hold over all wavelengths Rate at which backscatter amplitude declines relative to wavelength is smaller for hard, rough surfaces compared with relatively smooth and soft surfaces. Hypothesis 1: Backscatter responds to grain size which is more rapidly varying than surface topography. Hypothesis 2: These measures of backscatter respond to both grain size and surface topography.
  18. Summary so far ... Power law: no significant concentration of

    spectral power in any particular frequency band Slopes range between -0.5 and -3.5. The scaling relationship between amplitude and wavelength does not hold over all wavelengths Rate at which backscatter amplitude declines relative to wavelength is smaller for hard, rough surfaces compared with relatively smooth and soft surfaces. Hypothesis 1: Backscatter responds to grain size which is more rapidly varying than surface topography. Hypothesis 2: These measures of backscatter respond to both grain size and surface topography.
  19. Rest of this talk What’s the relationship between stochastic and

    deterministic measures of backscatter and topography? What’s their relationship to sediment type? Can we use this information to get at what scaling relationships exist between amplitude and lengthscales of riverbed patchiness?
  20. Topography: deterministic geometry (amplitudes) Standard deviation of locally detrended elevations

    Well known to be a function of window size
  21. RM30, Marble Canyon, May 2012

  22. Classification using Backscatter and/or Topography

  23. Classification using Backscatter and/or Topography

  24. Classification using Backscatter and/or Topography

  25. Topography power spectrum: ‘global’ versus ‘local’ detrend

  26. Stochastic geometry. 1) Spectral Strength P1(K) = ω1 (h0|K|)γ1

  27. Stochastic geometry. 2) Spectral Width P1(K) = ω1 (h0|K|) γ1

  28. Stochastic geometry. 3) Spectral Variance σ2 1 = 2 K0

    P1(K)dK
  29. RM30, Aug. 2013. Wentworth sediment type

  30. Relationship between sediment type & ‘roughness’ Variance in fluctuating part

    of both the topographic and backscatter signal increase with increasing clast size
  31. Relationship between sediment type & spectral strength Low-frequency component increases

    with increasing clast size Strong relationships with roughness
  32. Linear relationships on a continuum

  33. Classification using Backscatter and/or Topography

  34. Choice of detrending technique Pollyea and Fairley, 2011. Geology 39,

    623 - 626. ODR - any benefit of smaller residuals?
  35. Summary High-resolution MBES data from a non-cohesive riverbed Issue will

    be how to estimate sediment type when topography and backscatter are unrelated? Is it a problem? Lots of options for acoustic bed sediment classification Relative proportions of sand and gravel in mixtures? How would this change with silt/clay, or vegetated bottoms ?
  36. Summary High-resolution MBES data from a non-cohesive riverbed Issue will

    be how to estimate sediment type when topography and backscatter are unrelated? Is it a problem? Lots of options for acoustic bed sediment classification Relative proportions of sand and gravel in mixtures? How would this change with silt/clay, or vegetated bottoms ?
  37. Summary High-resolution MBES data from a non-cohesive riverbed Issue will

    be how to estimate sediment type when topography and backscatter are unrelated? Is it a problem? Lots of options for acoustic bed sediment classification Relative proportions of sand and gravel in mixtures? How would this change with silt/clay, or vegetated bottoms ?
  38. Summary High-resolution MBES data from a non-cohesive riverbed Issue will

    be how to estimate sediment type when topography and backscatter are unrelated? Is it a problem? Lots of options for acoustic bed sediment classification Relative proportions of sand and gravel in mixtures? How would this change with silt/clay, or vegetated bottoms ? Sandy gravels in Glen Canyon, Dec 2014
  39. Summary High-resolution MBES data from a non-cohesive riverbed Issue will

    be how to estimate sediment type when topography and backscatter are unrelated? Is it a problem? Lots of options for acoustic bed sediment classification Relative proportions of sand and gravel in mixtures? How would this change with silt/clay, or vegetated bottoms ?
  40. Last word. Sand grain size and acoustic roughness

  41. Thanks for listening