demonstrating landsat's new potential to monitor coastal and inland waters
DESCRIPTION
Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters. by Aaron Gerace Advisor: Dr. John R. Schott Sponsor: United States Geological Survey (USGS) . Research Motivation. Desire to monitor the Earth’s coastal and fresh water supply. - PowerPoint PPT PresentationTRANSCRIPT
![Page 1: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/1.jpg)
Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters
by
Aaron Gerace
Advisor: Dr. John R. Schott
Sponsor: United States Geological Survey (USGS)
![Page 2: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/2.jpg)
2
Research Motivation• Desire to monitor the Earth’s coastal and fresh water supply.• No environmental satellite to date has the necessary
characteristics…– High spatial resolution.– High radiometric fidelity.– Repeat Coverage.– Data are readily accessible.
Rochester Embayment Landsat
Rochester Embayment MODIS
![Page 3: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/3.jpg)
3
46 vs. 68 units of chlorophyll
Monitoring Fresh and Coastal Waters
• Spatial Resolution:
• Repeat Coverage:
• Accessible Data
• Radiometric Fidelity: Significant changes in constituent concentrations often lead to small changes in water-leaving reflectance.
![Page 4: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/4.jpg)
4
Monitoring Fresh Waters
• Case 2 Waters:– Inland and coastal waters.– Optically complex case 2
waters contain significant levels of...
• Chlorophyll-a– phytoplankton
• Suspended Materials– runoff
• Colored dissolved organic matter (CDOM)
– Decaying organic matter
• Issues:- Determine condition of water through constituent retrieval process.-Trophic status and trends- Characterize sedimentation in river plumes.
- Predict beach closings.- Impact of sediment on the surrounding environment.
![Page 5: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/5.jpg)
5
Overview of Research
1
1. Demonstrate that Landsat’s new OLI sensor is suitable for studying optically complex case 2 waters.– Radiometric fidelity.
46 vs. 68 units of chlorophyll
Landsat 5: July 13th, 2009
![Page 6: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/6.jpg)
6
Overview of Research
Atmospheric Compensation 2 2. Develop an over-water atmospheric compensation algorithm for the OLI sensor.
– OLI does not have the appropriate spectral coverage to utilize traditional water-based algorithms.
![Page 7: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/7.jpg)
Landsat 5: July 13th, 2009
Objective 1:Model the improved features of the OLI sensor
and demonstrate its improved radiometric fidelity.
![Page 8: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/8.jpg)
8
OLI Features: Enhanced Spectral Coverage
ETM+ Response
OLI Response
![Page 9: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/9.jpg)
9
OLI Features: Quantization
OLI (12-bit)
ETM+ (8-bit)
![Page 10: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/10.jpg)
10
OLI Features: Signal to Noise
ETM+ OLI
• About a factor of 5 improvement in SNR.
![Page 11: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/11.jpg)
11
Water IOPs-Absorb-Scatter
Modeling the Constituent Retrieval Process: Hydrolight
Wind Speed
Sensor locationSolar location
CHL SM CDOM
x 2000
![Page 12: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/12.jpg)
12
Resample QuantizeAVIRIS
ETM+
OLI
Add Noise
Modeling the Constituent Retrieval Process: At the Sensor
![Page 13: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/13.jpg)
13
Modeling the Constituent Retrieval Process: CRA
CD
OM
CHLSM
Top of Atmosphere
Air/Water Interface
CHL=3SM=4
CDOM=7
CHL(g/L)
SM(mg/L)
CDOM
0 0 0.5 .5 .51 1 .753 2 15 4 27 8 4
12 10 724 14 1046 20 1268 24 14
![Page 14: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/14.jpg)
14
Resample Quantize CRAAVIRIS
ETM+
OLI
Add Noise
Modeling the Constituent Retrieval Process: Summary
– Average residuals can be expressed as a percent of the total range of constituents. CHL [0 – 68], SM [0 – 24], CDOM [0 – 14]
– 10% error is our target for this experiment.
![Page 15: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/15.jpg)
15
Results
![Page 16: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/16.jpg)
16
SNR Margins
![Page 17: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/17.jpg)
17
Results
![Page 18: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/18.jpg)
18
Atmospheric Compensation Objective 2:Develop an over-water atmospheric
compensation algorithm specifically for the OLI sensor.
![Page 19: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/19.jpg)
19
• Purpose: Convert TOA radiances to water-leaving reflectances.• Issue: OLI doesn’t have 2 NIR bands which are required by
traditional water-based algorithms.– Gordon’s method (SeaWiFS).
• 2 methods developed:– Blue Band method.– NIR/SWIR band ratio method.
OLI ApproachCase 2 Waters
![Page 20: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/20.jpg)
20
OLI ApproachCase 2 Waters
Atmosphere
Water
)()()()()()( warart T
Atmosphere WaterImage
![Page 21: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/21.jpg)
21
OLI ApproachCase 2 Waters
Atmosphere
Dark Water
)()()()()()( warart T
Atmosphere Dark WaterImage
![Page 22: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/22.jpg)
22
• Incorporate dark water component into an atmospheric LUT.
OLI ApproachCase 2 Waters
)()()()()()( warart T
Atmosphere Dark Water
• Mid-latitude Summer profile.• May 20th, 1999.• Standard gases• Rural aerosols
• Varied visibility between 5 and 60 kilometers.
10km
15km20km
AVIRIS: May 20th, 1999
![Page 23: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/23.jpg)
23
10km
15km20km
• Incorporate dark water constant into an atmospheric LUT.
OLI ApproachCase 2 Waters
Atmosphere Dark Water
)()()()()()( warart T
AVIRIS: May 20th, 1999
![Page 24: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/24.jpg)
24
OLI Atmospheric Compensation Experiment 1: Simulated Data
• Use same 2000 water pixels described in first experiment.– Propagate to the top of 23 kilometer visibility modeled atmosphere.– Signals are then spectrally sampled to OLI, half margin noise is added, and
quantization effects included.– Average darkest 5% of signals in band 5 to determine atmosphere (22.97km).– Chosen atmosphere removed spectrally from all modeled pixels.
![Page 25: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/25.jpg)
25
• 15% is our target error when atmospheric effects are included.• A typical scene contains hundreds of thousands of water pixels!
OLI Atmospheric Compensation Experiment 1: Simulated Data
![Page 26: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/26.jpg)
26
• Simulated Image from Landsat 5 data– Lake Ontario (Dark Water) ROI used to determine atmosphere.– Chosen atmosphere removed globally from image.– Constituent retrieval algorithm implemented for 6 ROIs.
OLI Atmospheric Compensation Experiment 2: Simulated Scene
Landsat 5: May 16th, 1999 Simulated Image
![Page 27: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/27.jpg)
27
OLI Atmospheric Compensation Experiment 2: Simulated Scene
46 vs. 68 units of chlorophyll
![Page 28: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/28.jpg)
28
• Develop tool to spatially sharpen TIRS data With OLI– Done
• Use TIRS thermal data to calibrate flow field of hydrodynamic model.– Alge outpus surface temperature– Proof of concept complete
• Use Landsat reflective data to calibrate color of hydrodynamic model.
– ALGE outputs sediment profiles.– Initial tool under test
• Continue to investigate OLI atmospheric compensation.– 3 band method, perhaps.
Ongoing
![Page 29: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/29.jpg)
29
Back up
![Page 30: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/30.jpg)
30
• AVIRIS data (May 20th,1999) spectrally sampled to OLI’s sensor response function.
OLI Atmospheric Compensation Experiment 3: Real Data
AVIRIS: May 20th, 1999 Simulated OLI Data: May 20th, 1999
![Page 31: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/31.jpg)
31
• OLI atmospheric compensation method tested over Cranberry Pond and Long Pond.
– Deglint image.– 200 darkest values in bands 5 and 6 were used to determine atmosphere.– Atmospheric effects removed and constituent retrieval process performed.
• Empirical Line Method.
OLI Atmospheric Compensation Experiment 3: Real Data
“OLI” Data
Cranberry Pond
Long Pond
![Page 32: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/32.jpg)
32
• Initial errors are discouraging.– For Cranberry pond, CDOM retrieval is over 20%.– For Long Pond, retrieval errors for 2 constituents are greater than 30%.
OLI Atmospheric Compensation Experiment 3: Real Data
![Page 33: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/33.jpg)
33
OLI Atmospheric Compensation Experiment 3: Real Data
Cra
nber
ry P
ond
Long
Pon
dCompensated Data Retrieved Reflectances Expected Reflectances
Spectral Average: Long Pond ROI
![Page 34: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/34.jpg)
34
• More reasonable errors obtained.– For Cranberry pond, only suspended materials is over 15% retrieval error.– For Long Pond, retrieval errors for all constituents are less than 15%.
OLI Atmospheric Compensation Experiment 3: Real Data
![Page 35: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/35.jpg)
35
OLI Atmospheric Compensation Experiment 3: Real Data
Cra
nber
ry P
ond
Long
Pon
dRetrieved Reflectances: Biased Retrieved Reflectances: Bias Corrected Expected Reflectances
![Page 36: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/36.jpg)
Landsat 5: July 13th, 2009
ALGE Hydrodynamic Model
Develop techniques that will enable Landsat data to be used to calibrate a hydrodynamic model.
Objective 3:
![Page 37: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/37.jpg)
37
Hydrodynamic Modeling: Inputs
• To model the Genesee River plume, ALGE requires information from the scene of interest…
– Static: Land/Water, latitude/longitude, bathymetry, voxel size, DOY, etc.– Dynamic: Environmental Measurements (hourly)
• River data (flow rate, initial temperature)• Surface data (Pier / Rochester airport)• Upper air (Bufkit model)
Landsat 5: July 13th, 2009Northeastern United States
![Page 38: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/38.jpg)
38
Hydrodynamic Modeling: Inputs
ALGE Hydrodynamic Model
• To model the Genesee River plume, ALGE requires information from the scene of interest…
– Nudging Vectors (hourly).• Whole lake simulation provides nudging vectors for small scale simulation.
Landsat 5: July 13th, 2009Lake Ontario simulation: Surface Currents
![Page 39: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/39.jpg)
39
Hydrodynamic Modeling: Outputs
ALGE Hydrodynamic Model
• To model the Genesee River plume, ALGE requires information from the scene of interest…
– Land/Water, latitude/longitude, bathymetry, voxel size, DOY, etc.– Environmental Measurements (hourly)
• River data (flow rate, initial temperature)• Surface data (Pier / Rochester airport)• Upper air (Bufkit model)
– Nudging Vectors (hourly)
Landsat 5: July 13th, 2009
![Page 40: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/40.jpg)
40
Hydrodynamic Modeling: Outputs
ALGE Hydrodynamic Model
• To model the Genesee River plume, ALGE requires information from the scene of interest…
– Land/Water, latitude/longitude, bathymetry, voxel size, DOY, etc.– Environmental Measurements (hourly)
• River data (flow rate, initial temperature)• Surface data (Pier / Rochester airport)• Upper air (Bufkit model)
– Nudging Vectors (hourly)
Landsat 5: July 13th, 2009
![Page 41: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/41.jpg)
41
Hydrodynamic Modeling: Steady State
• Run ALGE until model reaches a steady state.• Satellite data will not match model data…
– Inaccurate inputs.– Model error.
July 2nd, 2009 July 13th, 2009
Start ModelStop Model
Landsat 5: July 13th, 2009
![Page 42: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/42.jpg)
42
Hydrodynamic Modeling: Calibration
• 24 hours prior to obtaining satellite data, the model is stopped and a calibration LUT created
– Vary environmental parameters (about their nominal values) that will affect a plume’s shape.
……10 day ALGE run (steady state period)
July 2nd, 2009 July 12th, 2009
1 day (calibration period)
July 13th, 2009
…Start Model
![Page 43: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/43.jpg)
43
Hydrodynamic Modeling: Calibration
• Develop a calibration LUT whose domain is made up of parameter variations and whose range is made up of ALGE runs.
![Page 44: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/44.jpg)
44
Hydrodynamic Modeling: Calibration
• Nonlinear, least-squares optimizer is used to search the LUT.– Landsat data must be registered and atmospherically compensated.– The point in space that provides the best match contains the model that best
describes the state of the environment.
![Page 45: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/45.jpg)
45
Hydrodynamic Modeling: Results
– RMS-error of 0.28 Kelvin.
Expected OptimizedWind Speed 100% 88.9%
Wind Direction 0.0 6.1
Flow speed 100% 61.8%River Temperature 19C 19.5C
Landsat 5: July 13th, 2009 Optimal Model
![Page 46: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/46.jpg)
46
• OLI exhibits enormous potential to be used for monitoring case 2 waters.
• The ability of the OLI atmospheric compensation algorithms were successfully demonstrated on simulated data, a simulated image, and a real image.
– The sensor must be well calibrated.– Adequate SNR must be achieved.
• Techniques were developed which enable Landsat data to be used to calibrate a hydrodynamic model.
Conclusions
![Page 47: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/47.jpg)
47
• Dr. Schott• My Committee
– Dr. Vodacek– Dr. Salvaggio– Dr. Messinger– Dr. Sciremammano
• DIRS Staff– Cindy Schultz
• Dr. Schott’s students• Mom & Dad• The Outlaws• Matt and Prudhvi• Trisha
Acknowledgements
![Page 48: Demonstrating Landsat's New Potential to Monitor Coastal and Inland Waters](https://reader035.vdocuments.us/reader035/viewer/2022081515/568167b8550346895ddd062f/html5/thumbnails/48.jpg)
48