Hyperspectral Remote Sensing - PowerPoint PPT Presentation

1 / 60
About This Presentation
Title:

Hyperspectral Remote Sensing

Description:

Hyperspectral Remote Sensing Lecture 12 prepared by R. Lathrop 4/06 How plant leaves reflect light Reflectance from green plant leaves Chlorophyll absorbs in 430-450 ... – PowerPoint PPT presentation

Number of Views:766
Avg rating:3.0/5.0
Slides: 61
Provided by: RickLa4
Category:

less

Transcript and Presenter's Notes

Title: Hyperspectral Remote Sensing


1
Hyperspectral Remote Sensing
  • Lecture 12
  • prepared by R. Lathrop 4/06

2
How plant leaves reflect light
Graphics from http//landsat7.usgs.gov/resources/r
emote_sensing/radiation.php
3
Reflectance from green plant leaves
  • Chlorophyll absorbs in 430-450 and 650-680nm
    region. The blue region overlaps with carotenoid
    absorption, so focus is on red region.
  • Peak reflectance in leaves in near infrared
    (.7-1.2um) up to 60 of infrared energy per leaf
    is scattered up or down due to cell wall size,
    shape, leaf condition (age, stress, disease),
    etc.
  • Reflectance in Mid IR (2-4um) influenced by water
    content-water absorbs IR energy, so live leaves
    reduce mid IR return

4
Hyperspectral Sensing
  • Multiple channels (50) at fine spectral
    resolution (e.g., 5 nm in width) across the full
    spectrum from VIS-NIR-MIR to capture full
    reflectance spectrum and distinguish narrow
    absorption features

5
Hand-held Spectroradiometer
  • Calibrated vs dark vs. bright reference
    standard provided (spectralon white panel - 6
    in image)
  • Can use passive sensor to record reflected
    sunlight or active illuminated sensor clip (4)

6
AVIRISAirborne Visible InfraRed Imaging
Spectrometer
7
Hyperspectral sensing AVIRIS
8
Compact Airborne Spectrographic Imager (CASI)
  • Hyperspectral 288 channels between 0.4-0.9 mm
    each channel 0.018mm wide
  • Spatial resolution depends on flying height of
    aircraft and number of channels acquired

CASI 550
For more info www.itres.com
9
EO-1 Hyperion
  • The Hyperion collects 220 unique spectral
    channels ranging from 0.357 to 2.576 micrometers
    with a 10-nm bandwidth.
  • The instrument operates in a pushbroom fashion,
    with a spatial resolution of 30 meters for all
    bands.
  • The standard scene width is 7.7 kilometers.
    Standard scene length is 42 kilometers, with an
    optional increased scene length of 185 kilometers
  • More info http//eo1.usgs.gov/hyperion.php

10
EO-1
  • ALI Hyperion designed to work in tandem

11
Hyperion over New Jersey
EO1H0140312004120110PY_PF1_01 2004/04/29, 0 to 9
Cloud Cover
EO1H0140312004120110PY_PF1_01 2004/04/29, 0 to 9
Cloud Cover
EO1H0140312004120110PY_PF1_01 2004/04/29, 0 to 9
Cloud Cover
EO1H0140312004184110PX_SGS_012004/07/02, 10 to
19 Cloud Cover
EO1H0140312004184110PX_SGS_012004/07/02, 10 to
19 Cloud Cover
12
Hyperion Image EO1H0140312004120110PY
2004/04/29
R 800- G 650- B 550
Fallow field
Active crop
13
Hyperion Image EO1H0140312004184110PX
2004/07/02
R 800- G 650- B 550
Conifer forest
Deciduous forest
14
Hyperspectral Sensing Analytical Techniques
  • Data Dimensionality and Noise Reduction MNF
  • Ratio Indices
  • Derivative Spectroscopy
  • Spectral Angle or Spectroscopic Library Matching
  • Subpixel (linear spectral unmixing) analysis

15
Minimum Noise Fraction (MNF) Transform
  • MNF 2 cascaded PCA transformations to separate
    out the noise from image data for improved
    spectral processing especially useful in
    hyperspectral image analysis
  • 1st is based on an estimated noise covariance
    matrix to de-correlate and rescale the noise in
    the data such that the noise has unit variance
    and no band-to-band correlation
  • 2nd create separate a) spatially coherent MNF
    eigenimage with large eigenvalues (high
    information content, lgt1) and b) noise-dominated
    eigenimages
  • (l close to 1)

16
MNF Transform example 1
Plot of eigenvalue number vs. eigenvalue MNF 6
noise
Original TM image using ENVI software
17
MNF Transform example 1
MNF 1
MNF 2
MNF 3
MNF 5
MNF 6
MNF 4
18
MNF Transform example 2
Plot of eigenvalue number vs. eigenvalue MNF 5,6
7 noise
Tm_oceanco_95sep04.img Original TM image using
ENVI software
19
MNF Transform example 2
MNF 1
MNF 2
MNF 3
MNF 7
MNF 5
MNF 6
MNF 4
20
Plant Absorption Spectrum
Image adapted from http//fig.cox.miami.edu/cmal
lery/150/phts/spectra.gif
21
Hyperspectral Vegetation Indices
  • NDVI (R800 R680) / (R800 R680) at 680
  • (R800 R705) / (R800 R705)
    at 705
  • Where 680nm and 705nm are chlorophyll absorption
    maxima and 800 is NIR reference wavelength. 705nm
    may be more sensitive to red edge shifts

22
Hyperspectral Vegetation Indices
  • Photochemical Reflectance Index (PRI) designed to
    monitor the diurnal activity of xanthophyll cycle
    pigments and the diurnal photosynthetic
    efficiency of leaves
  • PRI (R531 R570) / (R531 R570)
    where 531nm is the xanthophyll
    cycle wavelength and 570nm is a reference
    wavelength
  • (Gamon et al., 1990, Oecologia 851-7)

23
Hyperspectral Water Stress Indices
  • Water Band Index (WBI) designed to monitor the
    vegetation canopy water status (Penuelas et al.,
    1997, IJRS 182863-2868)
  • WBI R970 / R900
    where 970nm is the trough in the
    reflectance spectrum of green vegetation due to
    water absorption (trough tends to disappear as
    canopy water content declines) and
    900nm is a reference wavelength

24
Hyperspectral Water Stress Indices
  • Moisture Stress Index (MSI) contrast water
    absorption in the MIR with vegetation reflectance
    (leaf internal structure) in the NIR
  • MSI MIR / NIR or R1600/R820
  • Normalized Difference Water Index
    NDWI (R860-R1240) / (R860R1240)

25
Detection of Xylella fastidiosa Infection
ofAmenity Trees Using Hyperspectral
ReflectanceGH Cook project by Bernie Isaacson
Cook 2006
26
Hyperspectral reflectance curves
Green not scorched yellow scorching
brown - senesced
27
(No Transcript)
28
Hyperspectral Indices Applied
  • Normalized Difference Vegetation Index at 705nm
  • Red wavelengths to Green wavelengths
  • Photosynthetic Active Radiation
  • modified Water Band Index
  • Water Band Index
  • Normalized Difference Vegetation Index at 680nm
  • modified Photochemical Reflectance Index
  • Photochemical Reflectance Index
  • Simple Ratio

29
  • Negative vs. Symptomatic Positive (Margins and
    Bases)

Date PRI mPRI NDVI680 NDVI705 WBI mWBI SR PAR redgrn
14-Jul x x l x x l l l x
18-Jul x x l x l x l x x
22-Jul x l l x l x l l l
28-Jul x x l x l l l x x
11-Aug x l l x l l l l x
22-Aug x x l l x x l x l
26-Aug x x l x l l l x x
X - denotes significant difference
l - denotes significant difference not detected
Red text - denotes Nlt10
Negative vs. Symptomatic Positive (Margins Only)
Date PRI mPRI NDVI680 NDVI705 WBI mWBI SR PAR redgrn
14-Jul x x l x l l l l x
18-Jul l x l x l x l l x
22-Jul x l l x l x l l l
28-Jul x x l x l l l x x
11-Aug x l l x l l l l l

30
Pre-Visual Stress Detection?
  • Hypothesis Change in reflectance detectable
    before visual symptoms
  • Where can symptoms be detected?
  • Infected vs. Uninfected
  • Symptomatic (showing scorch) vs. Asymptomatic
    (tree infected but no symptoms)

Slide adapted from B. Isaacson
31
Scorch Timeline Datapoints
32
Derivative Spectroscopy
  • First order quantify slope, the rate of change
    in spectra curve
  • Second order identify slope inflection points
  • Third order identify maximum or minimum
  • Pros can be insensitive to illumination
    intensity variations
  • Con sensitive to noise

33
Derivative Spectroscopy Blue Shift of
Red Edge
As chlorophyll degrades, less absorption in the
red. Leads to a shift in the Red Edge (i.e.,
between 690 and 740nm) towards the blue
Stressed plant
R
Blue Shift
Red Edge inflection point
Normal plant
Spectral wavelength
34
Original spectral reflectance profile
1st derivative
The derivative is the slope of the signal
Derivative positive () ? signal slope
increasing Derivative 0 ? slope 0 Derivative
negative (-) ? signal slope decreasing
2nd derivative
Graphic from http//www.wam.umd.edu/toh/spectrum/
Differentiation.html
35
Derivative spectroscopy
  • Red Edge inflection point (point where the slope
    is maximum) at the center of the 690-740nm range
  • Corresponds to the maximum in the 1st derivative
  • Corresponds to the zero-crossing (point where the
    signal crosses the y 0 line going either from
    positive to negative or vice versa) in the second
    derivative

36
Spectra Matching
  • Spectra Matching takes an atmospherically
    corrected unknown pixel and compares it to
    reference spectra
  • Reference spectra determined from
  • In situ or lab spectro-radiometer measurements
  • Spectral image end-member analysis
  • Theoretical calculations
  • Number of different matching algorithms

37
Spectra Matching spectral libraries
USGS Digital Spectral Library covers the UV to
the NIR and includes samples of mineral, rocks,
soils, vegetations, microorganism and man-made
materials http//speclab.cr.usgs.gov/spectral-lib
.html

Nicolet spectrometer
38
ERDAS Spectral Analysis
39
From http//speclab.cr.usgs.gov
Reference spectra used in the mapping of
vegetation species. The field calibration
spectrum is from a sample measured on a
laboratory spectrometer, all others are averages
of several spectra extracted from the AVIRIS
data. Each curve has been offset from the one
below it by 0.05.
40
The continuum-removed chlorophyll absorption
spectra from Figure 1 are compared. Note the
subtle changes in the shapes of the absorption
between species.
From http//speclab.cr.usgs.gov
41
Spectral matching Spectral Angle Mapper
Material 1
Band Y
Reference material
Material 2
Band X
Spectral Angle Mapper computes similarity
between unknown and reference spectra as an angle
between 0 and 90 (or as cosine of the angle).
The lower the angle the better the match.
42
Subpixel Analysis Unmixing mixed pixels
Spectral endmembers signature of pureland
cover class
endmember1
Unknown pixel represents some proportion of
endmembers based on a linear weighting of
spectral distance. For example 60 endmember
2 20 endmember 1 20 endmember 1
Band j
unknown pixel
endmember3
endmember2
Band i
43
Water Colora function of organic and inorganic
constituents
  • Suspended sediment/mineral brought into water
    body by erosion and transport or wind-driven
    resuspension of bottom sediments
  • Phytoplankton single-celled plants also
    cyanobacteria
  • Dissolved organic matter (DOM) due to
    decomposition of phytoplankton/bacteria and
    terrestrially-derived tannins and humic substances

44
Ocean Color Spectra
Open Ocean Coastal Ocean
Red Algae bloom
45
Water Colora function of organic and inorganic
constituents
  • Phytoplankton contain photosynthetically active
    pigments including chlorophyll a which absorbs in
    the blue (400-500nm) and red (approx. 675nm)
    spectral regions increase in green and NIR
    reflectance
  • Suspended sediment and DOM will confound the
    chlorophyll signal. Typical occurrence in
    coastal or Case II waters as compared to CASE I
    mid-ocean waters

46
Ocean Color function of chlorophyll and other
phytoplankton pigments
Typical reflectance curve for CASE 1 waters where
phytoplankton dominant ocean color signal. Arrow
shows increasing chlorophyll concentration,
dashed line clear water spectrum. Adapted from
Robinson, 1985. Satellite Oceanography
100
10
R
1
0
400 500 600
700 nm
47
Water Colora function of organic and inorganic
constituents
  • Suspended sediment/minerals increases
    volumetric scattering and peak reflectance shifts
    toward longer wavelengths as more suspended
    sediments are added
  • Near IR reflectance also increases
  • Size and color of sediments may also affect the
    relative scattering in the visible

48
Suspended Sediment Plume
49
Water Colora function of organic and inorganic
constituents
  • Dissolved organic matter DOM strongly absorbs
    shorter wavelengths (e.g., blue)
  • High DOM concentrations change the color of water
    to a tea-stained yellow-brown color

50
Ocean Color RS Sensors CZCS, SeaWiFS MODIS
Higher spectral resolution bands across the
visible, with concentration in blue and green

Example CZCS wavebands
Band Center Wavelength (nm) Primary Use
1 412 (violet) Dissolved organic matter (incl. Gelbstoffe)
2 443 (blue) Chlorophyll absorption
3 490 (blue-green) Pigment absorption (Case 2), K(490)
4 510 (blue-green) Chlorophyll absorption
5 555 (green) Pigments, optical properties, sediments
6 670 (red) Atmospheric correction (CZCS heritage)
7 765 (near IR) Atmospheric correction, aerosol radiance
8 865 (near IR) Atmospheric correction, aerosol radiance
Bands 1-6 have 20 nm bandwidth bands 7 and 8
have 40 nm bandwidth.
http//daac.gsfc.nasa.gov/CAMPAIGN_DOCS/OCDST/what
_is_ocean_color.html
51
Ocean Color Indices
CZCS Ocean color image of the Gulf Stream from
May 8, 1981
  • CZCS phytoplankton pigment concentration
  • C Lw,443/Lw,550 for low concentrations
  • C Lw,520/Lw,550 for higher concentrations
  • Where Lw is the water leaving radiance
  • 443 and 520 wavebands should decrease due to
    greater absorption as pigment concentrations
    increase, 550 waveband remains generally stable
  • Note that these ratios are reversed in form from
    the geological indices with the numerator having
    the absorption peak and the denominator
    representing the stable background

52
Sea WiFS
  • Launched Aug 1, 1997.
  • Operated by ORBIMAGE
  • BandWavelength 402-422 433-453 480-500
    500-520 545-565 660-680 745-785 845-885 nm
  • Sun Synchronous, Equatorial crossing Noon
    20min
  • 1 day revisit time
  • 10 bit data
  • Swath width1,500 km 1.1km GRC

53
NOAA CoastWatch http//coastwatch.noaa.gov/
  • NOAA's CoastWatch Program processes and make
    available near real-time oceanographic satellite
    data (both ocean color and SST)

54
MODIS Ocean Color
  • MODIS on Terra and Aqua offers twice-daily
    coverage and simultaneous measurements of Ocean
    Color and SST.
  • 1-km data are available globally, and global
    composites are computed for a variety of spatial
    and temporal resolutions

Terra MODIS Chlorophyll(SeaWiFS-analog
algorithm, QualityAll)February 3, 2003, 0540hrs
GMTWest coast of India
Aqua MODIS Chlorophyll(SeaWiFS-analog algorithm,
QualityAll)February 3, 2003, 0840hrs GMTWest
coast of India
55
Water-leaving radiance Atmospherically-corrected
and normalized to a constant sun angle
Level 3Terra MODIS Normalized Water-leaving
Radiance at 443 nm (H. Gordon)Weekly average
March 6 - 13, 2001NASA/GSFC
http//modis-ocean.gsfc.nasa.gov/dataprod.html
56
MODIS/Aqua Ocean Weekly Productivity Indices
8-Day L4 Global 4km http//daac.gsfc.nasa.gov/MOD
IS/Aqua/ocean/MYD27W.shtml
57
EO-1 Hyperion
  • The Hyperion collects 220 unique spectral
    channels ranging from 0.357 to 2.576 micrometers
    with a 10-nm bandwidth.
  • The instrument operates in a pushbroom fashion,
    with a spatial resolution of 30 meters for all
    bands.
  • The standard scene width is 7.7 kilometers.
    Standard scene length is 42 kilometers, with an
    optional increased scene length of 185 kilometers
  • More info http//eo1.usgs.gov/hyperion.php

58
Hyperion eo1h0140342004241110ky
59
Hyperion Image EO1H0140312004184110PX
2004/07/02
R 800 G 650 B 550
60
Hyperion Image EO1H0140312004184110PX
2004/07/02
R 560 G 490 B 450
Write a Comment
User Comments (0)
About PowerShow.com