Remote Sensing Hyperspectral Imagery April 1 st , 2004 Stefan A. Robila robilas@mail.montclair

Remote Sensing

Hyperspectral ImageryApril 1st, 2004

Stefan A. Robila

robilas@mail.montclair.edu

www.csam.montclair.edu/~robila/RSL/

Source: http://nis-www.lanl.gov/~borel/

Hyperspectral Remote Sensing

• a remote sensing technology• “seeing” characteristics not recognized by the human eye

Electromagnetic Spectrum

Increasing Wavelength (in meters)

10 -8

Ultraviolet

Microwaves

10 -2

10 -6

Infrared

10 -11

Gamma Rays

10

Radio

X-Rays

10 -9

Visible

10 -7

Hyperspectral Remote Sensing

(example: FieldSpec Hand Held Spectroradiometer) • sensor obtains data (amount of light per wavelength) • computer software displays recorded spectrum • analyze spectral signature

Non-Imaging Instruments

Scanning radiometers• Passive system• Produces digital images

Imaging systems

Scanning radiometersMirror scans across-track (swath)

Imaging systems

Scanning radiometers2-D image formed by platform forward motion

Imaging Systems

CCD arrays• Passive system• Line or block of CCDs instead of scanning mirror• Senses entire swath (or block) simultaneously

CCD

Hyperspectral Remote Sensing

Multispectral – Many spectra (bands)

Hyperspectral – Huge numbers of continuous bands

Hyperspectral remote sensing provides a continuous, essentially complete record of spectral responses of materials over the wavelengths considered.

Hyperspectral Platforms

First hyperspectral scanners:

1982: AIS (Airborne Imaging Spectrometer)

1987: AVIRIS (Airborne Visible/infrared Imaging Spectrometer)

1995: Hyperspectral Digital Imagery Collection Experiment (HYDICE)

2000: Hyperion (EO-1)

AVIRIS

AVIRIS Specifications•224 individual CCD (charge coupled device) detectors

• Spectral resolution of 10 nanometers

• Spatial resolution of 20 meters (at typical flight altitude)

• Flight platform: NASA ER-2 (modified U-2)

• Flight altitude: from 20,000 to 60,000, but usually flown at 60,000

• Typical swath width is 11 km.

• Dispersion of the spectrum against the detector array is accomplished with a diffraction grating.

• The total interval reaches from 380 to 2500 nanometers (roughly the same as TM band range).

• image, pushbroom-like, succession of lines, each containing 664 pixels.

• shows the volume of data returned by imaging instruments

• illustrates how data from imaging instruments is geo-referenced

• data from different wavelengths can be used to create a “map” (in either true color or false color infrared formats)

Hyperspectral Cube

Hyperspectral Remote SensingHyperspectral images can be analyzed in ways that multispectral images cannot

In the Visible-NIR range, water ice and dry ice give characteristic spectral curves, as shown here:

Hyperspectral Data Analysis

General Approach:

• Develop Spectral Library

• Construct spectral curve for relatively "pure" materials

• Specific reflectance peaks and absorption troughs are read from these curves.

• Compare to lab spectra (mixture analysis)

• Mixtures of two or even three different materials can be identified as the components of the compound spectral curve.

Hyperspectral Data Analysis

Spectral Libraries:

Sets of hundreds of measured spectra for components likely to be encountered in the study area.

Spectral Angle

For two pixel vectors x and y, the spectral angle is computed as:

21

1

221

1

2

11

22

1 coscos/n

ii

/n

i

n

iii

- ,)α(

yx

yx

yx

yxyx,

i

(x,y)

x

y

Band 1Ba

nd 2

The distance measure used for spectral screening.

Hyperspectral Data Analysis

Pure Pixel Analysis

• Find relatively “pure” pixels

• Pixel Purity Index (PPI)

• “Pure” spectra are spectral endmembers

Endmembers

• Spectral characteristics of an image that represent classes of interest

• Usually assigned based on lab spectra

• Can be done manually

Hyperspectral Data Analysis

Spectral Mixture Analysis (SMA)

• Also called “unmixing”

• Assumes that the reflectance spectrum derived from sensor can be deconvolved into a linear mixture of the spectra of ground components

• Linear / Non-linear

• Linear SMA assumes linear relationship between reflectance and area

Linear Mixture Model•Each pixel vector x can be described as:

where S is the nxm matrix of spectra (s1, .., sm) of the individual materials (also called endmembers), a is an m-dimensional abundance vector and w is the additive noise vector.

•The abundances of the endmembers have the restrictions:

•The ICA performs endmember unmixing; the resulting components correspond to the abundances of the endmembers, the columns in the mixing matrix correspond to the endmembers.

m

iiia

1

wSawsx

m..,iai ,,..10

m

iia

1

1

Future Hyperspectral Sensors

Spaceborne rather than airborne

Success:

• Hyperion, is part of NASA’s EO-1 - launched in December, 2000.

• Co-orbiting with Landsat 7

• 220 channels from 400 to 2500 nm

• Ground resolution 30 meters.

Future Hyperspectral Sensors

Hyperion

Future Hyperspectral Sensors

Off-the shelf (reduce costs)

Success: SOC 700 (Surface Optics)

•Spectral Band: 0.43 –to 0.9 microns •Number of Bands: 120, 240 or 480 (configurable) •Dynamic Range: 12-bit •Line Rate: Up to 100 lines/second (120 bands) •Pixels per line: 640 •Exposure Time: 10 -> 10^7 microsecond

Hyperspectral Problems

• Data volume

• Cost

• Difficulty of analysis

• Spectral Libraries

• More complex