Lngh et al. 
;s from the 
3d which is 
. An average 
iza, 1975) 
luation (7). 
manner would 
adiance, 
) and (7) 
of L S (A) is 
ulting 
the actual 
is value of 
the above 
parallel to 
tivity 
would be 
cedure is 
reached, 
flectivities 
found to be 
value. The 
) with radi 
al number. 
using ten 
ee or four 
'nly one case 
iquired for 
cally 
(9) 
were perfect 
Lon index (VI) 
retaining the 
a of 
ar residual 
aate 
igle, varying 
ts. 
1 data are 
k of Duggin 
1985, 1986) 
factors which 
w angle 
the larger 
e the larger 
natural sur 
ly sensed 
tian surfaces 
e as seen by 
t necessarily 
cast by 
made). An 
me which has 
applied to a 
(1985, 1986) 
view angle 
d by (a) above 
possible to 
ises (b) and 
which have 
ition were 
)84 at the 
ring station, 
a from about 
i part 
lose pixels 
3 were positive. 
3, and to some 
extent identifies cloudy pixels over the land. Since 
the data acquisition time was local afternoon and 
since there were clouds on the western side of the 
selected scene, there might have been some pixels 
which were contaminated by cloud shadows and identi 
fication of such pixels is not yet possible. 
4 RESULTS AND DISCUSSION 
Raw NDVI were calculated using equation (1). The 
atmospheric correction procedure was implemented and 
spectral reflectances p(A^)and p(^2) were calculated. 
Equation (9) then yields atmospherically corrected 
NDVI values. A relation of the form of equation 
(10) was sought between raw and atmospherically 
corrected NDVI values. 
Y = mX + c 
(10) 
where Y is atmospherically corrected NDVI and X 
stands for raw NDVI. Line by line regression 
analysis was performed on a 512 x 512 scene but 
because of cloud cover and surrounding waters only 
about 80 to 200 pixels per scanline corresponded to 
cloud-free land area. For each scanline raw NDVIs 
were highly correlated to atmospherically corrected 
NDVI values. In fact the squared correlation 
coefficient ranged from 85% to 98%. The parameter c 
in equation (10) varied from about -0.1 to about 
-0.03 whereas the slope (or enhancement or magni 
fication) m ranged from about 2.2 to 3.5. These 
results indicate that the relation (10) is not a 
unique one. Had it been a unique relation then it 
would have been of great value. Therefore, it 
suggests that one has to apply atmospheric correction 
to each scene of interest. It is also clear from the 
values of m found above that the atmospherically 
corrected NDVI imagery should have high contrast 
compared to the contrast present in raw NDVI maps. 
Next a relation similar to equation (10) was sought 
between p (Aq) and atmospherically corrected NDVI. 
There was a large variability in the value of m (2 to 
30) but an important outcome was that the squared 
correlation coefficient ranged from about 30% to 
about 90%. This shows that p(Aq) carries some 
extra information to which NDVI is not sensitive. 
A similar analysis between p(A2) and atmospherically 
corrected NDVI showed a poor correlation between 
these two parameters. Therefore, p(Aq), p(A2) and 
atmospherically corrected NDVI values may be useful 
in improving surface cover type classification and 
further investigations are underway. 
5 CONCLUSIONS 
The primary motivation was to search for more than 
one parameter for land-cover classification. 
Application of atmospheric correction results in an 
increased contrast between too dissimilar surfaces. 
It is shown that the atmospherically corrected NDVIs 
are partially correlated to either channel reflect 
ivity. This means that these three parameters may 
prove to be of importance in improving land cover 
classification. Although atmospherically corrected 
NDVIs and raw NDVIs are highly correlated to each 
other, these preliminary results indicate that there 
is no unique relation between these two quantities. 
Thus, one has to apply atmospheric correction to each 
scene of interest. 
ACKNOWLEDGEMENT 
This work was carried out under the NERC contract 
No. F60/G6/12. 
REFERENCES 
Duggin, M.J., Piwinski, D., Whitehead, V. and 
Ryland, G., 1982, The scan angle dependence of 
radiance recorded by the N0AA-AVHRR. Proceedings 
of the AIAA/SPIE technical meeting, San Diego, 
California, 23-27 Aug., SPIE Vol. 363, p.98. 
Gordon, H.R., 1978, Removal of atmospheric effects 
from satellite imagery of the oceans, Appl. Opt., 
Vol. 13, p.1361. 
Hayes, L., 1985, The current use of Tiros-N series 
of meteorological satellites for land-cover 
studies, Int. J. Rem. Sens., Vol. 6, p.35. 
Hayes, L. and Cracknell, A.P., 1984, A comparison 
of Tiros-N series satellite data and Landsat data 
over Scotland. Proc. of Integrated Approaches in 
Remote Sensing, ESA SP-214, p.63. 
Holben, B.N. and Justice, C.O., 1981, An examination 
of spectral band ratioing to reduce the topo 
graphic effect on remotely sensed data, Int. J. 
Rem. Sens., Vol. 2, p.115. 
Hughes, N.A. and Henderson-Sellers, A., 1982, 
System albedo as sensed by satellites: its 
definition on variability. Int. J. Rem. Sens., 
Vol. 3, p.l. 
Janza, F.J. (editor), 1975, Manual of Remote Sensing, 
Vol. I, American society of photogrammetry, Falls 
Church, Virginia. 
Justice, C.O., Townshend, J.R.G., Holben, B.N. and 
Tucker, C.J., 1985, Analysis of the phenology of 
global vegetation using meteorological satellite 
data, Int. J. Rem. Sens., Vol 6, P.1271. 
Lauritson, L., Nelson, G.J. and Porto, F.W., 1979, 
Data extraction and calibration of Tiros-N/NOAA 
radiometers. N0AA technical memorandum NESS-107, 
N0AA/NESS, Washington D.C. 
Manual of Remote Sensing, 1983, (editor in chief), 
Data interpretation and applications. American 
society of photogrammetry, Falls Church, Va. 
Otterman, J., 1983, Absorption of insolation by land 
surface with sparse vertical protrusions, Tellus, 
Vol. 35B, p.309. 
Singh, S.M., 1986, Vegetation index and possibility 
of complementary parameters from AVHRR/2. Int. J. 
Rem. Sens., Vol. 7, p.295. 
Singh, S.M. and Cracknell, A.P., 1985, Effect of 
shadows cast by vertical protrusions on AVHRR data. 
Int. J. Rem. Sens., Vol. 6, p.1767. 
Singh, S.M. and Cracknell, A.P., 1986, The 
estimation of atmospheric effects for SPOT using 
AVHRR channel-1 data. Int. J. Rem. Sens., Vol. 7, 
P- 
Singh, S.M., Cracknell, A.P. and Spitzer, D., 1985, 
Evaluation of sensitivity decay of Coastal Zone 
Scanner (CZCS) detectors by comparison with in- 
situ near surface radiance measurements. Int. J. 
Rem. Sens., Vol. 6, p.749. 
Thekaekara, M.P., Kruger, R. and Duncan, C.H., 1969, 
Solar irradiance measurements from research air 
craft, Appl. Opt., Vol. 8, p.1713. 
Toll, D.L., 1985, Landsat-4 Thematic Mapper scene 
characteristics of a suburban and rural area, 
Photogram. Eng. and Rem. Sens., Vol. LI, p.1471. 
Townshend, J.R.G. and Tucker, C.J., 1985, 
Continental land cover classification using 
satellite data. Proc. of the 36th Congress of the 
International Astronomical Federation, Stockholm, 
Sweden. Oct. 1985, Peaceful Space and Global 
Problems of Mankind, Pergamon Press: Oxford, New 
York, Toronto, Sydney, Frankfurt. 
Tucker, C.J., Gatlin, A. and Schneider, S.R., 1984, 
Monitoring vegetation in the Nile Delta with 
N0AA-6 and N0AA-7 AVHRR. Photogramm. Eng. and 
Remote Sensing, Vol. 50, p.53. 
Tucker, C.J., Holben, B.N. and Goff, T.E., 1984, 
Intensive forest clearing in Rondonia, Brazil as 
detected by satellite remote sensing. Remote 
Sensing Environ., Vol. 15, p.255.