Figure 6. Optical layout of the Airborne Imaging 
Spectrometer (Goetz et al 1985). 
dedicated column of spectral detector elements for 
each crosstrack pixel in the scene. 
A sensor called the Airborne Imaging Spectrometer 
(AIS) has been built to test the imaging spectrometer 
concept with infrared area arrays (Figure 6). 
This instrument operates in the mode shown in Figure 
5d. The spectral coverage of the instrument is 
1.2-2.4 |lm in contiguous bands that are 9.3 nm wide. 
This sampling interval is sufficient to completely 
describe absorption features for solids in this 
wavelength region. Continuous strip images, 32 
pixels wide and 128 spectral bands are acquired from 
the NASA C-130 aircraft. The 128 spectral bands are 
acquired by stepping the spectrometer grating through 
four positions during the time it takes to fly 
forward 1 pixel width on the ground. The area array 
is read out between each grating position, and the 
data are recorded on the aircraft with a high density 
analog tape recorder. The IFOV of AIS is 1.9 
milliradians, which produces a ground pixel size of 
approximately 8 x 8m for a typical operating altitude 
of 4200m. To aid in locating the AIS ground track, a 
boresighted 35 mm camera acquires black and white, 
wide field of view photography. 
In order to test the capability of imaging 
spectrometry for mineral identification with the AIS, 
the Cuprite mining district of Nevada was chosen for 
study. The Cuprite area contains both hydrothermally 
altered and unaltered rocks, well exposed and nearly 
devoid of vegetation. The altered rocks overflown in 
this study contain secondary quartz, opal and clay 
minerals, and the area has been subjected to 
extensive study with broad-band multispectral images 
in the visible reflective and emissive infrared 
(Abrams et al. 1977; Kahle and Goetz 1983). Several 
minerals whose reflectance spectra are shown in 
Figure 2 occur in the Cuprite area and have 
diagnostic absorption features in the 2.0-2.4 (im 
region. The narrow spectral band sampling possible 
with an imaging spectrometer should, therefore, allow 
these minerals to be identified. 
Figure 7a shows the central region of the Cuprite 
district overflown with the AIS. The bright areas 
are the result of trenching operations that break 
through the dark stained surface crust and expose 
materials consisting of almost pure silica. The 
bottom curve in Figure 7b shows a raw 128 channel 
spectrum of a 5 x 5 pixel area in the AIS coverage 
outlined in Figure 7a. The major features are the 
broad atmospheric water bands centered at 1.4 and 1.9 
Urn and the solar irradiance curve, which exhibits a 
rapid falloff toward longer wavelengths. A shortcut 
to modellino the atmospheric and insolation effects 
can be made if one normalizes the data to an area in 
the image having little or no topographic relief and 
uniform, known spectral reflectance characteristics. 
The top curve in Figure 7b is the result of this 
normalization procedure. Spectral features in the 
surface material become more apparent in the 
normalized spectrum because the removal of systematic 
effects makes it possible to display the data at 
their full radiometric resolution. Enhanced 
32-channel images covering the region from 2.03-2.32 
Hm and acquired in the area outlined in Figure 7a are 
Figure 7. (a) air photo of a portion of the Cuprite 
mining district in Nevada, superimposed with the AIS 
coverage. (b) spectra (128 channels) derived from a 
5x5 pixel area in the AIS image of the Cuprite 
mining district. The normalized spectrum 
approximates the ground spectral reflectance. (c) a 
set of 32 AIS spectral images over Cuprite taken at 
9.3 nm intervals between 2.03 and 2.32 |lm in which 
each pixel spectrum has been normalized to produce an 
equal area under the reflectance curve. The 
differing reflectance characteristics as a function 
of wavelength are clearly visible (location a)(Goetz 
et al. 1985). 
shown in Figure 7c. Surface features having 
absorption bands can be recognized by changes in 
contrast with respect to their surroundings. 
Spectra can be derived on a pixel by pixel basis from 
any point on the AIS image. Figure 8 shows a single 
9.3 nm wide channel image at 2.03 |lm taken from 
Figure 7c and the reflectance spectra derived from 
averages of 3 x 3 pixel areas from all 32 images. 
Three general spectral classes are seen. The spectra 
taken from the hill in the lower portion contain an 
absorption doublet which matches that of kaolinite. 
A laboratory spectrum of a field sample from a 
spectral reference library is shown. The spectra 
show a single broad absorption feature centered at 
2.17 (im indicative of alunite, a mineral associated 
with altered feldspathic rocks. A laboratory 
reflectance spectrum of a field sample is 
superimposed. Spectra obtained from the top of the 
image are devoid of absorption features since this 
surface material consists of almost pure quartz, 
which does not exhibit spectral absorption features 
in this wavelength region except for a weak Si-OH 
feature at 2.25 pm. 
The Airborne Visible and Infrared Spectrometer 
(AVIRIS) is a second generation remote sensing 
instrument under development at the Jet Propulsion 
Laboratory as part of NASA's imaging spectrometer 
program. It is expected to be operational during 
1987. In contrast to the AIS, it is an 
optomechanical scanner which uses line arrays of 
detectors to image a 550 pixel-wide swath in 224 
contiguous bands from 0.4-2.4 pm. The instrument 
will be flown at high altitude in either a NASA U-2 
or ER-2 to provide broad coverage with resolution 
equal or better than that proposed for future space 
missions. By flying above 95% of the Earth's 
atmosphere, AVIRIS will provide representative data 
for the evaluation and potential correction of 
atmospheric effects. The functional parameters of 
the instrument are given in Table 2. 
The swath is overscanned by 3° to permit onboard roll 
correction by the instrument's gyroscope. The 224 
spectral bands are acquired with four separate 
spectrometers and line array detectors, connected to 
the scanner for optics by optical fibers (Figure 9). 
Figure 8. AIS 
pixel spectra < 
Direct identifi 
area can be rr 
spectral re 
field-collecte 
lines), and the 
Figure 9. Opti' 
infrared imagir 
SPECTROMETEI 
PRIMARY MIRR 
VNIR 
CORRECTOR- 
Figure 10. Opti 
Each line arraj 
dewar to prov 
calibration soi 
by optical fib 
for determinine 
holmium oxide 
monitoring spec 
Three of the 
indium antimon 
infrared; one < 
silicon line 
region. 
Detailed moc 
indicates that 
visible and 91 
These studies 
features in mi 
operating condi