Basic Hyperspectral Analysis Tutorial: Files Used in This Tutorial
Basic Hyperspectral Analysis Tutorial: Files Used in This Tutorial
Basic Hyperspectral Analysis Tutorial: Files Used in This Tutorial
You will also use some mineral spectral library files that are included with your ENVI
installation.
The image used in this exercise was collected by the Airborne Visible Infrared
Imaging Spectrometer (AVIRIS) sensor. AVIRISdata files are courtesy of NASA/JPL-
Caltech. The sample image covers the Cuprite Hills area of southern Nevada, an area
with diverse mineral types. The scene was collected from an ER-2 aircraft on August
8, 2011.
The full radiance scene is available from NASA/JPL's AVIRIS website in the Free Data
section. This image was processed as follows, resulting in the file
CupriteReflectance.dat:
Page 1 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
n Spatially subsetted
n Processed with FLAASH to remove atmospheric effects and to create a
surface reflectance image in ENVI format
n 53 bands marked as "bad" in the ENVI header file. These are primarily water
vapor bands that cause spikes in the reflectance curve.
n 1260 to 1560 nm (Bands 98-128)
n 1760 to 1960 nm (Bands 148-170)
See the "Preprocessing AVIRISTutorial" for the steps used to create the reflectance
image.
Background
Hyperspectral sensorsmore commonly known as imaging spectrometerscollect
spectral information across a continuous spectrum by dividing the spectrum into
many narrow spectral bands. Airborne and satellite imaging spectrometers can have
up to several hundred bands with a spectral resolution of 10 nanometers (nm) or
narrower. Compare this to broadband multispectral sensors such as Landsat 8 OLI,
which has nine spectral bands and a spectral resolution of 106 nm.
Spectral resolution refers to the width of each band within the captured spectrum.
More specifically, it refers to the width of an instrument response (band pass) at half
of the band depth. This is known as the full-width-half-maximum (FWHM). Spectral
sampling is a separate concept that refers to the band spacing, or the quantization of
the spectrum at discrete steps. Quality spectrometers are usually designed so that
the band spacing is approximately equal to the FWHM, which explains why spectral
sampling is often used interchangeably with spectral resolution.
The following figure shows how a given pixel from a multispectral image only covers
discrete spectral bands. Although it can have many bands covering the visible to
thermal spectrum, it cannot produce a continuous spectrum of an object like a
hyperspectral sensor can.
Page 2 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
The spectral resolution required for a specific sensor depends on the spectral
characteristics of the material you are trying to identify. Each material exhibits a
unique spectral signature across the electromagnetic spectrum. Factors that
influence a material's spectrum include composition, structure (crystallinity), grain
size, viewing geometry, and mixture.
The high spectral resolution from an imaging spectrometer allows you to identify
materials, whereas broadband sensors such as Landsat TM only allow you to
discriminate between materials. We will demonstrate these concepts in the steps
that follow.
Page 3 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
Page 4 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
3. Click anywhere inside of the image to select a different pixel. The Spectral
Profile updates to show the spectrum for that pixel.
4. Click the Show arrow on the right side of the Spectral Profile. A plot key
appears with the spectrum name and pixel location: for example,
CupriteReflectance (261, 393). You can also use this panel to set the plot
Page 5 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
properties. For more tips on working with spectral profile plot windows, see the
Spectral Profile topic in ENVI Help.
5. Click the Remove All button above the plot key to clear the Spectral
Profile.
6. Click the Crosshairs icon in the ENVI toolbar. This step is not required to
view image spectra, but it helps to better visualize the selected pixel location.
7. Close the Cursor Value dialog that appears.
8. In the Go To field of the ENVI toolbar, type pixel coordinates 325, 444 and
press the Enter key. The display centers over a pink area in the lower part of
the image. The Spectral Profile updates with the spectrum of that pixel location.
This spectrum represents an unknown material. Next, you will compare this
spectrum to a library spectrum that represents ground-truth data.
Page 6 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
Page 7 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
Included with your installation of ENVI are several groups of laboratory spectra from
the NASAJet Propulsion Laboratory (JPL), Johns Hopkins University (JHU), and the
U.S. Geological Survey (USGS). Data provided courtesy of courtesy of the Jet
Propulsion Laboratory, California Institute of Technology, Pasadena, California.
Copyright 1999, California Institute of Technology. All Rights Reserved.
The aster folder contains the ASTER Spectral Library (version 2), which includes
thousands of laboratory spectra from natural and man-made materials covering 400
to 1540 nm (Baldridge et al., 2009). This folder contains the JPL and JHU spectra.
Reference: Baldridge, A. M., S. J. Hook, C. I. Grove, and G. Rivera, 2009. The
ASTER Spectral Library Version 2.0. Remote Sensing of Environment, Vol. 113,
pp. 711-715.
Spectral libraries are stored in ENVI spectral library (.sli) format, with each line of
the image corresponding to an individual spectrum and each sample corresponding to
an individual spectral measurement at a specific wavelength.
1. From the menu bar, select Display >Spectral Library Viewer.
2. On the left side of the Spectral Library Viewer, expand the igcp264 folder.
3. Expand the igcp_1.sli collection.
4. Click on the KAOLINITE_KL500 spectrum and note the shape and detail of the
reflectance curve. Also note how the X and Y units of the spectral library plot are
scaled differently from those of the image spectra in the Spectral Profile.
5. Click the Show button in the Spectral Library Viewer to see the plot key for the
kaolinite spectrum that you selected.
6. Select the KAOLINITE_KL500 spectrum name from the plot key of the
Spectral Library Viewer, and drag it to the Spectral Profile dialog. The Spectral
Profile automatically scales the Y-axis to accomodate the two spectra, for direct
comparison.
7. Select the KAOLINITE_KL500 spectrum in the plot key of the Spectral Profile,
and select the Curve tab. Change its color to blue.
8. Continuum removal is a powerful visualization tool for spectral analysis,
especially for absorption features. It normalizes the spectral signature so that
you can compare individual absorption features from a common baseline. Select
the Y: Data Value drop-down list and select Continuum Removed. The
Spectral Profile should look like the following figure:
Page 8 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
9. Zoom into the wavelength range of 2000 to 2500 nm shown in the figure below.
If your mouse has a scroll wheel, click inside the plot window and roll the wheel
to zoom in. Or, click-and-drag the scroll wheel to draw a box around the area
you want to zoom into.
The double absorption features near 2200 nm offers a good reference point for
comparing image spectra to library spectra. In the example above, the shape of the
reflectance curve of the mineral kaolinite closely matches that of the image
spectrum.
Page 9 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
You could plot the spectra of other known materials from the spectral library, but with
hundreds of spectra available, that process could be time-consuming. ENVI has
advanced spectral tools such as Spectral Feature Fitting (SFF), Spectral Angle
Mapper (SAM), and Spectral Information Divergence (SID) that can measure the
similarity between an unknown material and a reference material. Please refer to the
ENVI Help for more information on advanced spectral tools.
Another way to extract spectral profiles from an image is to create regions of interest
(ROIs). Up to this point, you have extracted image spectra from a single pixel. With
an ROI, you are extracting the mean spectra from all the pixels that comprise the
ROI.
Close the Spectral Profile and Spectral Library Viewer in preparation for the next
exercise.
Page 10 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
n Minimum and maximum envelope containing all of the spectra in the ROI
(red)
6. Close the ROIStatistics Results dialog.
7. Right-click on the Regions of Interest folder in the Layer Manager and select
Statistics for All ROIs. The plot window shows the mean spectra for all ROIs.
8. Right-click inside the plot window and select Stack Plots. This offsets the
spectra for comparison.
9. Right-click again inside the plot window and select Legend.
10. Zoom into the wavelength range of 2000 to 2500 nm. If your mouse has a scroll
wheel, click inside the plot window and roll the wheel to zoom in. Or, click-and-
drag the scroll wheel to draw a box around the area you want to zoom into. The
plot should look similar to the following figure:
11. Compare the spectral features of each spectrum, and note unique
characteristics that might allow mineral identification.
12. When you have finished, close the ROIStatistics View dialog. Keep the ROI Tool
dialog open for the next exercise.
Page 11 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
You can click and drag the colored bars to change them to other bands. This is a
good way to enhance specific minerals by centering one color bar in an
absorption feature and the other two on opposite shoulders of the feature.
5. Zoom into the wavelength range of 2000 to 2500 nm. If your mouse has a scroll
wheel, click inside the plot window and roll the wheel to zoom in. Or, click-and-
drag the scroll wheel to draw a box around the area you want to zoom into.
6. Disable the Regions of Interest layer in the Layer Manager to hide the ROIs.
Page 12 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
7. Click and drag the red, green, and blue bars to the location shown in the figure
below.
8. Right-click in the Spectral Profile and select Load New Band Combination.
Because the green channel falls within the absorption feature, green values will
be low. The red and blue channels have a higher (and nearly equal) reflectance,
resulting in a purple display in areas where kaolinite is abundant.
Page 13 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
Page 14 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.
Basic Hyperspectral Analysis Tutorial
Experiment with other pixel locations, using the mineral ROIs as a guideline, to
highlight those features with new color combinations. You should begin to understand
how the RGB colors correspond to the spectral signature. You can also predict how
certain spectra will look, given a particular pixel's color in the RGB image. Then you
can design and test specific RGB band selections that maximize your ability to map
certain minerals.
This is a crude way of discriminating mineralogy based on different color composites.
ENVI has more advanced visualization tools such as scatter plots and the n-D
Visualizer. ENVI also provides a rich set of advanced whole-pixel and sub-pixel
spectral analysis algorithms to help you perform endmember extraction, spectral
unmixing to estimate the abundances of materials, and target detection. Refer to
ENVIHelp for further details on these topics.
Copyright Notice:
ENVIis a registered trademark of Exelis Inc., a subsidiary of Harris Corporation.
Page 15 of 16
2015 Exelis Visual Information Solutions, Inc., a subsidiary of Harris Corporation. All Rights Reserved. This
information is not subject to the controls of the International Traffic in Arms Regulations (ITAR) or the Export
Administration Regulations (EAR). However, this information may be restricted from transfer to various embargoed
countries under U.S. laws and regulations.