Remote Sensing of the Rochester Embayment

1
Remote Sensing of the Rochester Embayment

Rolando Raqueño Adam Goodenough Michael Bellandi
Don Taylor Jason Hamel John Schott

• Model-based Exploitation
• Algorithms Over Water

2
Overview

• Review of Big Picture and Model-based
  Exploitation
• First Attempt at Hyperspectral Water Constituent
  Mapping
• Ongoing Refinement to Water Constituent Mapping
• IOP Modeling
• Atmospheric Compensation
• Modeling Complex Scenarios
• Future Directions Recommendations

3
Overview Big Picture
Concentrations
Model Inherent Optical Properties
Reflectance, r(l)
Model Atmosphere
Radiance, L
Digital Counts
4
Signal Sources
Atmosphere to Sensor
80
10
10
Air/Water Transition
Water/Air Transition
In Water
5
Case 1 (Ocean) vs Case 2 (Inland) Waters

• Chlorophyll Constituent
• Small Concentration Range
• Negligible water leaving radiance in NIR

• Chlorophyll Constituent
• Suspended Sediment
• Colored Dissolved Organic Matter (CDOM)
• Orders of Magnitude Concentration Range
• Significant water leaving radiance in NIR

6
Why use physics-based models?

• Calibrate remote sensing signals to meaningful
  physical parameters
• Whats in the water, atmosphere, etc. ?
• Quantify the impact of various environmental
  parameters on remote sensing signals
• Whens the best time to image a scene?
• Understanding of the image formation
  phenomenology
• We see this image feature because

7
(No Transcript)
8
May 20, 19991145 AM
N
solar glint
Heading 279 degrees
AVIRIS Flightlines
E
Solar Position Zenith 31 degrees Azimuth 120
degrees
Heading 45 degrees
Digital Imaging and Remote Sensing Laboratory
9
Rochester Embayment Sites
10
Example of an underflight ground truth effort
superimposed on a portion of an AVIRIS image of
the shore of Lake Ontario.
Water Quality Samples
MISI flight Boston Whaler Canoe kayak pier/br
idge panels truth panels
4
4
ASD
Pier Team
11
Atmospheric Compensation

• Physics model-based approach
• extremely difficult
• Merging of radiative transfer models
• non-trivial

12
Taking the easy way out

• Empirical Line Method

13
Calibrating AVIRIS Images
Low Signal Pixel
High Signal Pixel
14
After ELM Calibration
AMOEBA FIT
AMOEBA FIT
AMOEBA FIT
AMOEBA FIT
15
Basic Hydrolight World
MODTRAN Generated Sky
Detector
Random Surface (Spatially uncorrelated)
Slabs of homogeneous inherent optical properties
(IOPs) Absorption and Scattering Cross sections,
Scattering Phase functions
Output is a single point
Flat, constant bottom type
16
Example LUT Entries
C0 SM0 CDOM0
C13 SM0 CDOM0
C0 SM0 CDOM50
17
Simple Fitting
ST truth data
?
TRUE
min (ST - SP)2
FALSE
Final CHL CDOM TSS
SQ Error
CHL TSS CDOM
LUT
CDOM
C
SM
Sp predicted
CHL
CDOM
TSS
18
Hyperspectral Concentration Maps
AVIRIS Image Cube Lake Ontario Shoreline

• Comparison to Ground Truth
• Percantage of Concentration Range
• Chlorophyll 14
• Sediments 14
• CDOM 12

Dr. Rolando Raqueno
19
Empirical Line Method Cons

• Assumes Spatial Atmospheric Homogeneity
• Knowledge of Concentrations at Points
• Needed to Revisit Model-based Atmospheric
  Compensation

20
Current Case 1Assumptions
VIS Region 400-700 nm
NIR Region 700-950 nm
21
Current Case 1 (Oceanic) Assumptions
VIS Region 400-700 nm
NIR Region 700-950 nm
Negligible water leaving - All radiance reaching
sensor due to atmosphere
22
Current Case 1 (Oceanic)Assumptions
VIS Region 400-700 nm
NIR Region 700-950 nm
Negligible water leaving - All radiance reaching
sensor due to atmosphere
Compensate in other wavelengths based on
atmospheric effects in NIR
23
Realistic Conditions in Case II
VIS Region 400-700 nm
NIR Region 700-950 nm
Water leaving radiance due to suspended sediments
24
Realistic Conditions in Case II
VIS Region 400-700 nm
NIR Region 700-950 nm
Overestimation of atmospheric effects using Case
1 assumptions
25
Realistic Conditions in Case II
VIS Region 400-700 nm
NIR Region 700-950 nm
Overestimation of atmospheric effects using Case
1 assumptions
Negative reflectance, concentrations, etc.
26
How Can Models Help?
NIR Region 700-950 nm
VIS Region 400-700 nm
27
Realistic Conditions in Case II
VIS Region 400-700 nm
NIR Region 700-950 nm
No Measurements
28
IOP Modeling in Near IR Region

• Model-based estimation of Suspended Mineral IOPs
• Jason Hamel

29
Objective

• Effects of suspended solids on water leaving
  radiance
• Composition
• Particle size
• Concentration
• Tools
• OOPS (Ocean Optical Property Simulator)
• Minsu Kim, Cornell University
• Hydrolight

30
Process Summary
Composition
Refractive index
Particle size distribution
31
Process Summary
Composition
Quartz Albite Kaolinite Calcite Opal
Refractive index
Particle size distribution
32
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
33
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
34
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
OOPS
Mie and T-Matrix Scattering
35
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
OOPS
Scattering Phase function (extremely difficult to
measure)
36
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
OOPS
37
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
Concentration
CHL TSS CDOM
OOPS
38
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
Concentration
CHL TSS CDOM 0 10 0 0.76 0.57 0.57 62.96 22.44
6.12 6.51 10.37 2.14 4.28 10.00 2.75
OOPS
39
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
Concentration
CHL TSS CDOM 0 10 0 0.76 0.57 0.57 62.96 22.44
6.12 6.51 10.37 2.14 4.28 10.00 2.75
OOPS
40
Process Summary
Composition
Quartz Albite Kaolinite Calcite
Opal 1.544 1.527 1.549 1.486/1.658/Spectral
1.44
Refractive index
Particle size distribution
14 Junge 2 Gaussian 7 Log-Normal
Concentration
CHL TSS CDOM 0 10 0 0.76 0.57 0.57 62.96 22.44
6.12 6.51 10.37 2.14 4.28 10.00 2.75
OOPS
41
Original Hydrolight IOPSimulation (Lake Ontario
IOPs)
42
Oceanic Particle Size Distribution(Junge)
43
Hydrolight Modeled Reflectances (Junge)
44
Log-Normal PSDs
45
Effect of Composition (Log-Normal PSD)
46
Suspended Sediment SpectraMeasurements and
Simulation
Bale et al Measured
OOPS Modeled
47
Comparison to Sampled Data
Modeled
Sampled
Shinnecock Canal (Long Island)
http//disc.gsfc.nasa.gov/oceancolor/scifocus/ocea
nColor/turbid_2.shtml
48
Realistic Conditions in Case II
VIS Region 400-700 nm
NIR Region 700-950 nm
49
Combined aerosol/water constituent determination
(Don Taylor)

• Working in the Near Infrared 700-1000 nm

BEST FIT FOR SEDIMENT AND AEROSOL TYPE
Suspended Sediment Reflectance Spectra
Aerosol Types,
50
COMPASS Flight Lines June 7, 2004 June 8, 2004
with ground truth
51
COMPASS Flight Lines June 7, 2004 June 8, 2004
52
Long Pond
June 7
June 8
53
Long Pond(enhanced)
June 7
June 8
54
Basic Hydrolight World
MODTRAN Generated Sky
Detector
Random Surface (Spatially uncorrelated)
Output is a single point
Slabs of homogeneous optical properties
Flat, constant bottom type
55
A More Complex WorldRationale for Photon Mapping
MODTRAN Generated Sky
Detector
Object interaction
Surface with spatial structure
Underwater Plumes
Continuous/Arbitrary distribution of optical
properties
Variable, rough bottom types
56
Megascene ( Tile 5 )
Fused Image
57
DIRSIG MegaScene

• DIRSIG
• Physics based model developed at RIT to simulate
  remotely sensed data
• Various platforms
• Line scanner, framing array, pushbroom scanner

MegaScene DIRSIG simulations (Tiles 1-5). Main
experiment area is labeled CE. (7 sq. km)
58
DIRSIG MegaScene
MegaScene DIRSIG simulations (Tiles 1-5). Main
experiment area is labeled CE. (7 sq. km)
59
DIRSIG Simulation of Tile 5
60
Topography Bathymetry Merging
Nina Raqueño
Bathymetry
DEM
61
Current product merges disparate data sets (DEM
with Bathymetry) Topobathymetry to be flown in
2006-2007 NOAA/Army Corps
62
CAD Models
Air Force Resolution Target
Frogman 21,577 facets
Mantamine 3,672 facets
Boat 305 facets
Skimmer 2,659 facets
Submarine 3,402 facets
63
Simulations Embedding Targets
64
Megascene ( shoreline )
Michael Bellandi
65
Underwater Tree (Adam Goodenough)
66
Spectral Photon Mapping Simulation(Adam
Goodenough)
67
Modeling Input Needs from the Measurement
community

• Spectral Optical Measurements to augment
  traditional water quality measurements (IOPs)
• Absorption and Scattering Coefficients
• Scattering phase functions
• Adoption of Modeling tools
• Particle Size Distribution Measurements
• Both organic and inorganic suspended materials
• Cheaper Technology and Equipment monitoring these
  parameters
• Repackage and miniaturize traditional ocean
  equipment
• Distributed Computing

68
Future Direction

• Use of hyperspectral and thermal model
  simulations to study archive of past remote
  sensing imagery
• Model at hyperspectral resolutions and convolve
  to past sensor capabilities
• Establish design requirements for future sensors
  (environmental and homeland security)
• Airborne Sensors
• Ground based Sensors
• Link to Biological and Ecological Risk Models
• Aquatox (EPA)

69
Questions?
70
CHL Ground Truth Comparison
RMS 9.0 mg/m3 14 of CHL range
71
http//www.epa.gov/waterscience/models/aquatox/SC1
_Overview_Setup20Notes.pdf
72
Case 1 vs. Case 2 WatersENVI N-D Visualizer Tool
640 modeled cases
Case 1
Can we make meaningful matches?
Case 2
Jason Hamel
73
Photon Map ( sinusoid )
Zaneveld, et al. 2001
74
Photon Map ( sinusoid )
Zaneveld, et al. 2001
75
Photon Map ( sinusoid )
Zaneveld, et al. 2001
76
Photon Map ( superposition )
77
Photon Map ( obscuration )
78
Photon Map ( obscuration )
Carter Costello 2003 (2005)
79
CHL Ground Truth Comparison
RMS 9.0 mg/m3 14 of CHL range
80
TSS Ground Truth Comparison
RMS 3.2 g/m3 14 of TSS range
81
Spatial Pattern Comparison of Chlorophyll vs.
Suspended Solids
82
CDOM Ground Truth Comparison
Glint Area
RMS 1.3 1/m_at_350nm 12 of CDOM range
83
(No Transcript)
84
(No Transcript)
85
(No Transcript)