IEEE Geoscience and Remote Sensing Magazine - June 2019 - 43

over the ROI. Performance related to execution times was
also stated.
In a 2005 experiment by Rochford et al. [75], HyMap
data were acquired over Davis, California, along with
ground truth reflectance measurements of white and black
tarps and aerosol measurements from two shadow-band
radiometers (see Figure 12). FLAASH reflectance retrievals
with different aerosol models were compared with ground
truth and with results from ACORN provided by Hyvista

(a)
Davis, White Tarp

1
0.8
Reflectance

and FFC). The ATREM and FLAASH models were compared
based on operational characteristics, input requirements,
algorithm limitations, solar zenith angle variation, and
computational complexity. The treatment of aerosols was
discussed in some detail, along with a detailed description
of water-vapor and visibility estimates.
In 2003, Matthew et al. [74] compared reflectance spectra retrieved from FLAASH and ACORN (in terms of spectral angle) with ground truth measurements from 1998
AVIRIS data collected at the NASA Stennis Space Center.
Selected results are shown in Figure 11 for a white panel, a
black panel, grass, and soil. The sensor altitude was 3 km,
with high solar angle, high visibility, and moderate water
vapor (according to a radiosonde measurement). MODTRAN was used with 1-cm -1 resolution, a rural-haze model, and the in-scene atmospheric compensation algorithm
for multiple scattering in addition to the spectra being polished. The largest differences were in overall scaling, which
likely reflects non-Lambertian behavior of the materials.
The retrieved visibility was approximately 70 km, and the
average retrieved water-vapor column amount was 1,570
atm-cm (from the 1,130 nm band), well within the measurement error of the radiosonde value of 1,560 atm-cm.
Wavelength recalibration and polishing steps were found
to significantly improve agreement between the retrievals
and measurements.
In a 2004 ar ticle [46], K r use compared AT R EM,
ACORN, and FLAASH using low-altitude AVIRIS data over
Boulder, Colorado. Adequate background was given on all
three algorithms, including a description of spectral polishing. Various combinations produced six reflectance images. Plots of retrieved gravel for all three algorithms were
shown, along with maps of water vapor and cloud cover

0.6
0.4
0.2
0
0.4

Truth
FLAASH
QUAC
ACORN
0.9

0.9
0.8

2.4

Davis, Black Tarp

0.7

0.1

0.6
0.5

Reflectance

1.4
1.9
Wavelength (µm)
(b)

0.4
0.3
0.2

Truth
FLAASH
QUAC
ACORN

0.05

0.1
0.5

1.5
Wavelength (µ m)

Black Panel
Grass

2.5

White Panel
Soil

FIGURE 11. A plot of FLAASH-retrieved reflectances (solid lines)

versus ground truth (dotted lines) for an AVIRIS image over NASA's
Stennis Space Center using 1-cm-1 band-model parameters and
spectral polishing [74].
JUNE 2019

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

0
0.4

0.9

1.4
1.9
Wavelength (µm)
(c)

2.4

FIGURE 12. (a) A Hymap image of Davis, California, showing a
large white tarp located at the university airport and a large black
tarp in a field (red boxes). ACORN, FLAASH, and QUAC retrieved
reflectances for the (b) white and (c) black tarps compared with
ground truth (black line).

IEEE Geoscience and Remote Sensing Magazine - June 2019

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2019