g the 
data 
ween 
erved 
| data 
> and 
band 
and 
the 
0.69, 
and 
pute 
sky 
DIS 
and 
data into 5 classes, namely, vegetation, urban, cloud & snow, 
sea, and other, by using the maximum likelihood method. The 
band reflectance correlation does not hold for classes of cloud, 
snow, & sea and these classes are used for screening purpose. 
The classification map is shown in Fig.l. The reflectance 
values for three classes of vegetation, urban and other were 
obtained from Terra/MODIS image data by removing the 
atmospheric effects using the observed aerosol optical 
thickness values in Bands 1 and 3. Since the aerosol scattering 
effects are negligible in the band 7(A=2.16um), the surface 
reflectance in Band 7 was obtained by removing both 
molecular attenuation and water vapor absorption effects. We 
found the values of surface reflectance ratio for vegetation, 
urban and other classes in Bands 1 and 3, applicable in Japan, 
as follows: 
C.=054, C. 0.3547 for vegetation (5) 
C =0330, CG 0489 for urban (6) 
C 770.503 C =0417 for other (7) 
We should note a large discrepancy in the value of C, 
between USA and Japan. Seasonal and local variations in 
vegetation may explain this discrepancy. The scatter diagrams 
in band reflectance ratio for vegetation class are shown in Fig.2. 
For given reflectance ratios for these classes, we can retrieve 
distributions of aerosol optical thickness in bands 1 and 3 by 
using LUT(Look Up Tables) in which the theoretical radiances 
at the top of the atmosphere(TOA) are tabulated as a function 
of the surface reflectance and the aerosol optical thickness for 
given bands and angles of the incident and reflection. Such 
examples in band 3 and band 1 are shown in  Fig.3-(a) and 
-(b), respectively. 
3. RETRIEVED RESULTS 
We retrieved the aerosol optical thickness v, over the lands at 
\=0.66um from Terra/ASTER and Landsat-7/ETM+ data, 
using the band reflectance ratios given by eqns. (4)-(6) in 
section 2. The aerosol optical thickness map at A=0.660 pum, 
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004 
retrieved from Terra/ASTER data, is shown in Fig.4 The 
aerosol optical thickness over the sea was estimated by 
assuming measured Ángstróm exponent a at our study site. 
From seen in Fig.4 we can notice a smooth transition between 
the sea and land. The validation results of the retrieved aerosol 
optical thickness v, from Terra/ASTER and Landsat/ETM+ 
image data are shown in Fig 4. Acquired dates of ASTER 
data sets are on April 02, 2002 and June 08, 2003, and that of 
ETM+ on April 15, 2001. As seen in Fig. 5, we can find an 
excellent agreement in t, between the estimation and 
observation. 
4. CONCLUSIONS 
The following conclusions were made by this study: 
(1) We presented new band reflectance ratios for certain land 
classes between the visible and infrared bands using both 
MODIS data sets and simultaneously sky observation data. 
(2) We confirmed that the band correlation method is a useful 
method for retrieving the aerosol parameters over lands. 
(3) We obtained better results in the aerosol optical parameter 
retrieval over Japan using new band reflectance ratios than 
those using Kaufman's band reflectance ratios. 
(4) Seasonal and regional variation in the band reflectance ratio 
should be studied further. 
ACKNOWLEDGEMENTS 
This research was supported partially by CEReS, Chiba Uni., 
Joint Project Research 15-1. MODIS data sets were provided 
by Yasuoka Lab., ISS, Univ. of Tokyo and Frontier Research 
Center, Tokyo Univ. of Infor. Sciences. ASTER data set was 
provided by ERSDAC, Japan. We appreciate very much for 
their kindness. 
REFERENCES 
Kaufman et al., 1997. The MODIS 2.1mm Channel-Correction 
with Visible Reflectance for Use in Remote Sensing of Aerosol, 
IEEE Trans. GRS, vol.35, no.5, pp.1286-1298.