the measured spectral response profiles of the 
ASAS channels. Thus, an attempt was made to define 
an arbitrary but perhaps more realistic bandpass 
profile. The profile spans 0.045 micrometers and 
is 0.015 micrometers across at the half-maximum 
points (Figure 1). The same profile was applied 
to all 9 ASAS channels used in this study. The 
results of the 5S runs are given in Table 2 under 
the heading "5S BAND (PROF)". 
Discussion of Results 
The results given in Table 2 are portrayed in 
Figure 2. In addition to the 5S results, the 
values obtained after atmospheric correction using 
the code of Fraser et al. (1989) monochromatically 
at NASA are also indicated for comparison. The 
general trend of all four of the surface 
reflectance spectra is characteristic of 
vegetation, increasing from the chlorophyll 
absorption minimum at lower wavelengths up to the 
infrared plateau. However, the monochromatic 
calculations from 5S (LAMBDA GRID) and from the 
code of Fraser et al. yield anomalously high 
values of surface reflectance in the vicinity of 
the narrow oxygen absorption feature at 0.762 
micrometers. The bandpass results should be more 
representative of the situation that actually 
obtains during a measurement in the ASAS channels, 
and this is borne out by the improved results in 
the region of the oxygen band. 
There also appears to be a slight over-prediction 
of surface reflectance around 0.722 micrometers 
where there is some absorption due to water 
vapour. If a moister atmosphere were to be used 
in the 5S runs (a mid-latitude summer profile, for 
example), the bump in the spectrum at 0.722 
micrometers would be considerably higher, 
especially in the monochromatic case but also in 
the bandpass cases. The bandpass cases are also 
affected because the water absorption feature is 
less narrow than the oxygen feature and occupies 
a greater proportion of the ASAS bandpass. The 
point to be made here is that the proper 
atmospheric correction of high spectral resolution 
data in absorption regions depends on a good 
knowledge of the relevant gas content at the time 
of data acquisition. This reguirement is more 
critical for the more variable gases such as water 
vapour, and less critical for other gases such as 
oxygen which is less variable. (Note that some 
gas profile data were obtained on the ASAS over 
flight day, but the US62 standard atmosphere 
profile was used for convenience in this study.) 
Finally, it should be noted that the surface 
reflectance retrieval necessarily assumes that the 
wavelength calibration of the sensor is good. The 
effects of spectral shifts on sensor response have 
been addressed by Suits et al. (1988) and Teillet 
(1990). 
FURTHER INSIGHT 
Of all the parameters involved in the atmospheric 
correction process used in surface reflectance 
retrieval, the exoatmospheric solar irradiance, 
E«,(A), and the two-way gas transmittance, x„(A), 
have the greatest spectral variability. More 
specifically, the formulation used to obtain 
surface reflectance from apparent radiance at the 
sensor includes a factor of [E„( A)x a (A) ]~ x in the 
multiplicative term (Teillet, 1989). In spectral 
regions where there is significant gas absorption, 
x„(A) can become very small and hence the multi 
plicative atmospheric correction coefficient can 
become very large. This is necessary in order to 
retrieve the surface reflectance which will have 
been severely attenuated in such absorption 
regions. However, the predicted surface 
reflectance will be rather sensitive to uncertain 
ties in x a (A). Thus, if there is no information 
available on gaseous absorption at the time of 
image data acguisition and standard values are 
used in the atmospheric correction, the retrieved 
surface reflectances can potentially depart 
significantly from normal values in spectral 
regions affected by absorption. 
The guantity [E„( A)x„( A) ]"' was plotted as a 
function of wavelength in order to illustrate 
where problems might occur (Figure 4). The values 
were obtained from special runs of the 5S 
atmospheric code. The gas transmittance includes 
absorption by water vapour, ozone, and oxygen. 
Most of the spikes of higher value are due to 
water vapour and the narrow spike at 0.762 
micrometers is due to oxygen. Figure 4 shows the 
effect of degrading the spectral resolution from 
0.005 micrometers to 0.01, 0.02, 0.03, and 0.04 
micrometers, using the filter profile described in 
Figure 3. Apart from the general smoothing of the 
spectrum, it is interesting to note how the narrow 
oxygen feature is greatly diminished with decreas 
ing resolution. This explains the ASAS results 
described earlier in that a monochromatic calcula 
tion near the oxygen feature will use too large a 
value of [E„( A)x a (A) ]" 1 compared to the value that 
would obtain in a 0.015-micrometer bandpass. 
CONCLUDING REMARKS 
The narrowness of imaging spectrometer bands 
implies greater sensitivity to spectrally- 
selective atmospheric absorption features. To 
study this effect, surface reflectances were 
retrieved from ASAS data using monochromatic and 
bandpass atmospheric computations. It was found 
that anomalous results for surface reflectance can 
be obtained in the vicinity of absorption features 
in the monochromatic case whereas bandpass cal 
culations yield better results. Even with 
bandpass computations, the retrieved surface 
reflectances can depart significantly from normal 
values in spectral regions affected by absorption 
if there is no information available on 
atmospheric conditions at the time of image 
acguisition. 
ACKNOWI.EDGEMENTS 
The author wishes to thank G. Fedosejevs and A. 
Kalil for assistance with the preparation of the 
manuscript. 
REFERENCES 
Fraser, R.S., Ferrare, R.A., Kaufman, Y.J., and 
Mattoo, S. (1989), Algorithm for Atmospheric 
Corrections of Aircraft and Satellite Imagery, 
NASA Technical Memorandum 100751, NASA Goddard 
Space Flight Center, Greenbelt, Maryland. 
Irons, J.R., Ranson, K.J., Williams, D.L. and 
Irish, R.R. (1989), Forest and Grassland Ecosystem 
Studies Using the Advanced Solid-state Array 
Spectroradiometer, Proceedings of the 1989 
International Geoscience and Remote Sensing 
Symposium (IGARSS'89) and the Twelfth Canadian 
Symposium on Remote Sensing, Vancouver, B. C., 
pp.1761-1764. 
580