HyperSpectral Imaging HSI Support for GOESR HESCW Risk Reduction Activities

1
HyperSpectral Imaging (HSI) Support for GOES-R
HES-CW Risk Reduction Activities

• Paul Bissett
• pbissett_at_feriweb.org

2
SAMSONSpectroscopic Aerial Mapping System with
On-board Navigation

• We have developed a low-cost, robust
  HyperSpectral Imager, the Spectroscopic Aerial
  Mapper with On-board Navigation (SAMSON).
• SAMSON provides for a full HSI dataset 256 bands
  in the VNIR (3.5 nm resolution) at 75 frames per
  second, with a SNR, stability, dynamic range, and
  calibration sufficient for dark target
  spectroscopy.

3
HSI/MSI installationon 3-axis stabilized mount
4
SAMSON HSI/MSI installationon 3-axis stabilized
mount
5
Virginia Shore Test FlightsPre Atmospheric
Correction Imagery
6
SAMSON
DSS
7
Product GenerationStep 1. Calibration and
Orthorectification
Putting the photons in the right place Putting
the pixel in the right place
8
Product GenerationStep 2. - Illumination and
Atmosphere Correction
Range in which the GA was bounded Water Vapor
.5, 2.0 Ozone .30, .38 Tau 550 .05,
1.2 Wind Speed 2, 6, 10 Relative
Humidity 50, 70, 80, 90, 98 Aerosol Model
'urban', 'maritime', 'coastal',
'coastal-a', tropospheric' 75,000,000
possible solutions!
Chosen Model (after 59K runs) Water Vapor
1.895 Ozone 0.3792 Tau 550 0.050 Wind
Speed 2 Relative Humidity 50 Aerosol
Model urban
9
HSI Atmospheric Correction
Pre-Atmospheric Correction
Post-Atmospheric Correction
10
ATM Future Directions Issues to
Address-Atmospheric Correction
Northern Monterey Bay
11
ATM Future Directions Issues to
Address-Atmospheric Correction
Northern Monterey Bay
12
Atmospheric Correction from Multiple Views of
Same Sample
3
2
Multi-line Overlap
1
15 overlap
Multi-line Stack
13
Product GenerationStep 3. Geospatial
Intelligence
14
Product GenerationStep 3. Geospatial
Intelligence
FERI LUT 2002
NOAA 1992

• 16 m resolution
• Unknown pixels removed from both data sets.

15
Evidence of Algal Overgrowth in the Florida Keys
Typical algal overgrowth on coral rubble in
Florida. Photo courtesy of U.S. Department of
the Interior, U.S. Geological Survey, Center for
Coastal Geology
Caulerpa brachypus is a nonnative
macroalgae that has invaded Florida's coral
reefs. Photo courtesy of Harbor Branch
Oceanographic Institution, Inc.
16
CSU CICORE HSI ProgramCenter for Integrated
Coastal Observatory for Research and Education
Monterey Peninsula October 2004
MSI (DSS) OrthoPhoto 0.5 m
HSI Kelp Density (10 intervals)
17
Product Generation Kelp Coverage Comparison
2002
2004
18
Product Generation Kelp Coverage Comparison -
Zoom3 - RGB
2002
2004
19
Water Clarity/Quality Inherent Optical
PropertiesSan Francisco Bay 2004
20
Water Clarity/Quality Inherent Optical
Properties San Francisco Bay 2004
MSI (0.5 m) HSI (3 m) HSI Absorption/ Scattering
Product (3 m)
21
Harmful Algal Bloom (Red Tide)Monterey Bay,
October 2002
22
Harmful Algal Bloom (Red Tide)Monterey Bay,
November 2004
PHILLS 2, ATM-corrected, CIR enhanced. J. Ryan,
A.Fischer, MBARI
23
Humboldt Bay Eel Grass IdentificationChaeli
Judd, MS Thesis, Judd et al., 2006 (in prep)
LIDAR/HSI Fused Bathymetric Product
HSI determined eel grass distributions,
previously unknown.
24
High Resolution MSI 0.5 m RGB - Morro Bay 2006
25
High Resolution Digital Elevation Map1.5 m -
Morro Bay 2006
0.25 m contours
Elevation critical in both Sauer et al., and Judd
et al., studies.
26
Fused RGB/Digital Elevation Map0.5/1.5 m - Morro
Bay 2006
Elevation and its rate of change are ecological
determinants in riparian communities.
27
HSI SpectraPre-Flight Calibration
28
HSI SpectraPre-Flight Calibration
Smooth Spectra (7 band boxcar)
29
HSI Classification3.0 m HSI - Morro Bay 2006
Unsupervised classification 16 classes
30
HSI Spectra 3.0 m HSI - Morro Bay 2006
Higher SNR than previous PHILLS, particularly in
NIR.
Much better spatial resolution But is it good
enough?
31
Spatial Resolution StudiesMorro Bay 2006
High Res HSI 1 m
Low Res HSI 3 m
Higher Res MSI 0.5 m also
have 0.15 m
32
Mission History/Data Availability

• Deployments
• 2000 West Florida Shelf
• 2001 WFS and New Jersey Coast
• 2002 San Diego, Mobile, AL, Monterey, CA, and
  Looe Key, FL
• 2003 Monterey, CA, Fort Lauderdale, FL, and
  Sarasota, FL
• 2004 Humboldt Bay, San Francisco, Monterey, Big
  Sur, San Luis Bay, Santa Barbara Channel, LA
  Harbor, San Diego, CA
• 2006 Morro Bay, San Luis Bay, Santa Barbara,
  San Diego, CA Pearl River, MS

33
Nominal Risk Reduction Flight Plans

• Altitude 27,000 ft.
• 6 m cross-track GSD, 1.5 m along track GSD.
• 4 lines, 20 km long, combined 23 km x 23 km
  scene.
• 28 minute re-access time.
• Spectral resolution 3.5 nm from 380 1050 nm.
• 14 day flight window, 5 days of flight
  operations, 6 hours per day.
• Data rate 60 GB/hour SAMSON, 60 GB/hour DSS
• Bin by 14 spatial, bin by 3 spectral 10 m GSD,
  10 nm spectral, 6.5X increase in SNR.
• Each 10 nm, 10 m HSI scene 320 MB.

34
Pre-Deployment Planning

• Google Earth flight line review needs to be
  completed by first week in August.
• See example

35
Deployment Operations
36
Deployment Data Products

• Mission flight logs are generated during
  processing.
• See example.
• We will add other data as it is sent to us.
• Issues to address
• Data product desires.
• Algorithms for data products and processing
  schedule.
• Requires estimate of spectral response function
  for each algorithm, i.e. MODIS algorithms need to
  apply MODIS spectral response function, SeaWiFS
  algorithms need to apply SeaWiFS algorithms, etc.

37
Post-Deployment Schedule

• HyDRO posting of PHASE I data products will occur
  by end of October.
• Possibly sooner depending on demand. We are
  schedule for other flight operations in CA and FL
  through mid-October.
• Post-deployment calibration during November.
• PHASE II data products posted by end of December.

38
Summary

• Need to have flight line planning completed by
  1st week in August.
• We can have multiple scenarios planned for
  adaptive scheduling. We would prefer not to
  generating flight lines on-the-fly.
• In field data product generation is initially
  delayed by 8 hours for transfer of flight data
  and backup.
• If faster products are desired, we need a have a
  risk/reward discussion.
• We can assemble other metadata or data products
  for GE display and distribution.

39
HyDRO
40
What is HyDRO?

• A prototype web-based data distribution site for
  remotely sensed data.
• HyDRO is an acronym for HyperSpectral Data
  Repository Online.

41
Rationale for HyDRO

• HyperSpectral data can be define as a continuous
  spectral coverage of data at less than 10nm
  resolution (Chang et al.,2004)
• FERI collects HyperSpectral data at the rate of
  approximately 60 Gb/flight hour (raw and
  non-geocorrected)
• Facilitate effective distribution of the data
  collected.

42
Original Method of Distribution

• Tape and optical media.
• Drawbacks
• Time and expense involved in writing and sending
  the media.
• Increased turnaround time (5 to 7 days including
  sending the media).
• Version control issues.
• Time consuming and computationally intensive to
  navigate through the datasets to extract areas of
  interest.

43
HyDRO

• Features
• Dynamic web pages to handle user product
  requests, privileges and imagery access.
• Image and band sub-selection through the web
  pages.
• Database used to store information pertaining to
  users, job history and experiments.
• IDL-based back-end program generates the
  requested products.
• Availability of data alerted to the user via
  email notification.
• Data download through a user-specific ftp
  account.
• Imitate Sensor. (still in alpha)

44
Advantages of HyDRO

• Faster turnaround time. (Typically 6-24 hours)
• Ability to choose desired region of interest and
  bands.
• Ability to imitate response functions of other
  sensors.
• Access to the latest versions of the experiments.
• Ability to view job history.
• Easier administration.

45
Using HyDRO The Registration Page

• Prior to accessing HyDRO, a user needs to
  register.
• Upon registration, the information is reviewed by
  the HyDRO administrator and appropriate
  permissions are allotted.
• Note A demo_user account has been set up for
  people wishing to have a demonstration of HyDRO
  without registering. While a job can be submitted
  with this account, it will not be processed.

46
Using HyDRO The Login Page

• The credentials supplied during the login are
  verified against the database.
• After the verification process, the user is taken
  to the experiments page.

47
Using HyDRO The Experiments Page

• The experiments page is a personalized page
  consisting of the list of the experiments the
  user is allowed access to.
• The dots on the United States map indicate the
  locations for the corresponding experiments.
• Clicking on the links under the Experiments
  heading of the table takes the user the console
  page pertaining to that experiment.
• The links in the Flt Log column takes the user
  to external pages detailing the specific
  conditions surrounding the experiment.

48
Using HyDRO The Console Page

• Data selection is made on this page based on the
  mouse controlled interactive zoom window (for
  ROI), wavelengths and spatial resolution.
• The data selection is limited by the file size
  allotted to the user.

49
Using HyDRO The Bands Selection Page

• Bands can be selected picking predetermined band
  templates or manually selecting bands of
  interest.

50
Using HyDRO The Console Page After Bands
Selection

• The arrow and the circles indicate the updates
  that occur once bands are selected.
• Note that the Submit Query button will be
  disabled if the job size exceeds the maximum file
  size allotted to the user.

51
Using HyDRO Imitate Sensor Console Page

• The imitate sensor console page has the option
  selecting wavelengths in addition to choosing the
  sensor model.

52
Using HyDRO The Sensor Model Page

• Users can choose the sensor model and
  corresponding algorithm from the dropdown menus.
  When the sensor model is chosen, the algorithms
  pertaining only to the sensor model are loaded.
• Currently the models have been implemented for
  Landsat4, Landsat5, Landsat7 and SeaWiFS sensors.
• The algorithms that have been implemented are
  k490, oc2 and oc4 for the SeaWiFS sensor.

53
Using HyDRO Imitate Sensor Console Page After
Selection (Sensor Model Only)

• The arrow and the circle indicate the updates
  that occur once the sensor model and algorithm
  are selected.
• Note that the Submit Query button will be
  disabled if the job size exceeds the maximum file
  size allotted to the user.

54
Using HyDRO Imitate Sensor Console Page After
Selection

• Note that the Number of Output Bands are
  displayed according to the sensor algorithm
  chosen. If no algorithm is chosen the number of
  output bands will correspond to the sensor model
  bands.

55
Using HyDRO Imitate Sensor Console Page After
Selection (Sensor Model and Algorithm)

• The arrow and the circle indicate the updates
  that occur once the sensor model and algorithm
  are selected.
• Note that the Submit Query button will be
  disabled if the job size exceeds the maximum file
  size allotted to the user.

56
Using HyDRO The Terms and Conditions Page
57
Using HyDRO For The demo_user Account Only
58
Using HyDRO The Final Page
59
Using HyDRO Automated Notification
60
HSI Data Cube Distribution
61
Global and Regional Navigation
62
Search and Sort
63
Metadata Review
64
Data Selection
65
Packaging and Review
66
Summary

• HyDRO provides for web-based distribution of TB
  data sets.
• Allows for effective version control.
• Security control for project specific access.
• Minimizes server and archive hardware
  requirements.
• Browse facility allows user to select only
  required data.
• SQL data bases provides communication to users
  based on job submission for easy resubmission.
• Alpha imitate sensor/product page allows for
  sensor comparison and user defined products
  delivery.
• Mirror site at OSU in 2008.