by Barnaby W. Rockwell¹, Roger N. Clark, K. Eric Livo, Robert R. McDougal, and Raymond F. Kokaly

Open-File Report 02-199
 

2002
 
 
 
 
 

This report is preliminary and has not been reviewed for conformity with U.S. Geological Survey editorial standards or with the North American Stratigraphic Code.  Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

U.S. DEPARTMENT OF THE INTERIOR
U.S. GEOLOGICAL SURVEY

¹barnabyr@usgs.gov, Denver, Colorado

This report contains information regarding the reflectance calibration of spectroscopic imagery acquired over the Wasatch Mountains and Park City region, Utah, by the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor on August 5, 1998. This information was used by the USGS Spectroscopy Laboratory to calibrate the Park City AVIRIS imagery to unitless reflectance prior to spectral analysis.  The Utah AVIRIS data were analyzed as a part of the USEPA-USGS Utah Abandoned Mine Lands Imaging Spectroscopy Project.

The following is a summary of the procedures used by the USGS Spectroscopy Lab to calibrate spectroscopic image data to reflectance.  These procedures will result in the generation of multiplier and offset spectra which can be used to convert to reflectance an imaging spectrometer data cube which has been previously atmospherically corrected using published or COTS MODTRAN-based software such as ATREM (Gao and Goetz, 1990; Gao et al., 1992).

• 1.  Using field spectrometer with spectral range and bandpass similar to that of the imaging spectroscopy data being calibrated, characterize a ground calibration site covered by the image data.  An optimal calibration site is large, spectrally bland, and homogenous (e.g. tarps, gravel parking lots, beaches, salt flats, playas, gravel pits, dam faces, intersections of dirt roads, etc.)
• 2.  View field spectra and delete ones with obvious errors.
• 3.  Average remaining field spectra.
• 4.  Use spectral editing program to remove small glitches in average spectrum related to residual atmospheric absorptions and/or sensor noise or artifacts.
• 5.  Multiply glitch-free average spectrum by Spectralon correction spectrum (spectrum of white plate used to calibrate field spectrometer to reflectance units).
• 6.  Convolve result of step #5 to the sampling interval and bandpass of the imaging spectroscopy data being calibrated.
• 7.  PATH RADIANCE CORRECTION.  Locate a dark, shadowed, and vegetated area in your ATREM-corrected image data set (scaled by 20,000) and extract representative spectra.  Average these spectra together.
• 8.  Use spectral editing program to visually estimate a "true" shape for the average of shadowed vegetation pixels generated in step #7.  ATREM over-corrects for path radiance, resulting in a drastic reflectance fall-off below ~0.55 microns (sometimes the fall-off starts at as high as 0.7 microns).  This fall-off must be removed.
• 9.  CALULATE OFFSET SPECTRUM.  Subtract edited, "corrected" spectrum of dark vegetation (result of step #8) from original, un-edited spectrum.
• 10.  Extract spectra of field calibration site from ATREM-corrected image data cube.  Average these spectra together.
• 11.  Subtract OFFSET spectrum (result of step #9) from average image spectrum of calibration site generated in step #10.
• 12.  CALCULATE MULTIPLIER SPECTRUM.  Divide edited field spectrum of calibration site (result of step #6) by the image spectrum of calibration site from which ATREM path radiance overcorrections have been removed (result of step #11).
• 13.  Apply MULTIPLIER and OFFSET to ATREM-corrected imaging spectrometer data (see flowchart):

• 13a.  Apply radiative transfer corrections to the radiance data supplied by JPL by running the ATREM program, using the ATREM command file shown below.
• 13b.  Download the offset and multiplier spectra provided below.
• 13c.  Subtract the offset spectrum from the ATREM-corrected cube.
• 13d.  Multiply the results of Step 13c by the multiplier spectrum.
• 13e.  Multiply the results of Step 13d by a scaling factor of 20,000.  This step allows the reflectance data to be stored with full dynamic range in unsigned integer format (2 bytes per pixel) rather than floating point format (4-bytes per pixel), reducing data storage requirements.

• Photo of calibration site:  Deer Creek Reservoir dam face
• Photo showing dam face surface and spectral measurement technique
• Average spectrum of calibration site acquired by ASD field spectrometer
• Channel number, wavelength, and FWHM values for 1998 AVIRIS data
• ATREM command file used to apply atmospheric radiative transfer corrections to radiance data
• AVIRIS image of calibration site showing pixels selected for averaging to characterize site
• Ascii list of AVIRIS pixels averaged to characterize calibration site
• ATREM-corrected AVIRIS spectrum of calibration site
• AVIRIS image of dark target showing pixels selected to estimate ATREM path radiance over-correction
• Ascii list of AVIRIS pixels averaged to characterize dark target
• Derived MULTIPLIER spectrum
• New derived MULTIPLIER spectrum
• Derived OFFSET spectrum

• All pixel locations are given in line, sample (y, x) format where the origin (pixel 1,1) is the upper left pixel of the image.
• Pixel locations apply to the two-scene blocks shown in this image, which also shows the calibration site.

Larger 12 KB image

• Click HERE for edited average ASD spectrum of the Deer Creek reservoir dam face calibration site in ascii format (shown in green above).

Click HERE for an ascii listing of information regarding 1998 AVIRIS channels.  This information comprises the "waves.um.98" file required to run ATREM using the command file shown below.  Column descriptions are as follows:
 

Channel Number

Wavelength center  position (microns)

FWHM (microns)

Wavelength uncertainty

FWHM uncertainty

Channel Number

AVIRIS
20.00                           ! plane altitude (in km, above sea level)
08 05 1998 18 40 30             ! date/time (month day year hour minute second)
40 35 48                        ! latitude (degrees minutes seconds)
N                               ! earth hemisphere (N or S)
111 25 42                       ! longitude (degrees minutes seconds)
W                               ! earth hemisphere (E or W)
0.                              ! average spectral resolution (nm) or "0"
waves.um.98                     ! 1998 AVIRIS wavelength file (microns)
1                               ! channel ratio parameters (0 or 1)
0.9140 3 0.9815 3 0.943 3       ! .94 um water vapor band ratio parameters
1.0870 3 1.173 3 1.125 3        ! 1.14 um water vapor band ratio parameters

2                               ! atmospheric model
1 1 1 1 1 1 1                   ! gas selectors
0.34                            ! total column ozone amount (atm-cm)
1 200                           ! aerosol model and visibility (km)
1.792                           ! average elevation of scene surface (in km)
pc98_r4s4-3_rad.v               ! input image cube in radiance units
0                               ! read AVIRIS input data dims from vicar hdr
pc98_r4s4-3_atrem               ! output image cube in scaled reflectance units
0.                              ! spectral resolution (nm) of output data
20000.                          ! scale factor of output reflectance data
pc98_r4s4-3_atrem.vap           ! output water vapor image

Calibration Site: Deer Creek reservoir dam face.

The area shown in red corresponds to the area characterized with the field spectrometer.  See below for an ascii list of the pixels shown in red.

Click HERE for an ascii list of coordinates of the 57 AVIRIS pixels averaged to characterize the calibration site. These coordinates are relative to the upper left (NW) corner of AVIRIS scene pc98_r2s2-1_rtgc.

Larger 15 KB image

• Click HERE for an ATREM-corrected average AVIRIS spectrum of the Deer Creek reservoir dam face calibration site in ascii format.
• The ATREM-corrected spectra of the pixels shown in red above were averaged to produce the spectrum shown here.

• The dark target pixels of shadowed vegetation shown above were selected from AVIRIS scene pc98_r4s4-3_rtgc.
• See below for an ascii list of these pixels.

Click HERE for an ascii list of coordinates of the AVIRIS pixels averaged to characterize the dark target. These coordinates are relative to the upper left (NW) corner of AVIRIS scene pc98_r4s4-3_rtgc.

Larger 14 KB image

• Click HERE for the multiplier spectrum in ascii format.

Larger 16 KB image

 
• This new multiplier spectrum corrects for a small residual positive feature, or "hump" centered near 1.45 microns. This region is important for the mapping of alunites and other sulfates.
• Click HERE for the multiplier spectrum in ascii format.

Larger 15 KB image

• Click HERE for the offset spectrum in ascii format.

References:

Gao, B.C., and Goetz, A.F.H., 1990, Column atmospheric water vapor and vegetation liquid water retrievals from airborne imaging spectrometer data:  J. Geophys Res. 95, pp. 3549-3564.

Gao, B.C., Heidebrecht, K.B., Goetz, A.F.H., ATmospheric REMoval Program (ATREM) User's Guide, 1992, Center for the Study of Earth From Space document, University of Colorado, version 1.1, 24pp.

U.S. Geological Survey, a bureau of the U.S. Department of the Interior
Last modified May 9, 2002.