98 
The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences. Vol. XXXVII. Part Bl. Beijing 2008 
Figure 4. Scatter diagrams of mean AOT for match-ups 
4. CONCLUSIONS 
Most AERONET sites lie with certain distances to the sea, 
influenced by the local aerosol (In this paper, only “Gosan 
SNU” and “Shirahama” are near the sea with little factitious 
contamination), which makes it difficult to validate aerosol 
product over ocean. Restricted by the distance to sea and wind 
direction, suitable sampling window may be different for 
different sites and seasons. Two spatio-temporal sampling 
approaches were adopted and the results were not quite the 
same. For the research area, the sampling window of 25km 
radius and ±05h was better than 45km radius and ±1.0h. 
However, using the small window, there were small amount of 
matched data, which was easily affected by data distribution. It 
still needs more data to get reliable validation result. 
Using 25km radius and ±0.5h as the sampling window, there 
were 32 match-ups. The linear fitting results were 
y=0.672075x+0.132479(R 2 =0.68), 
y=0.707179x+0.112427(R 2 =0.65) and 
y=0.650986x+0.089910(R 2 =0.56), and without AOT less than 
0.05, the R 2 was increased to 0.77, 0.74 and 0.67 respectively 
for MODIS 550, 658 and 860nm channels. The RMSE was 0.15, 
0.13 and 0.10 respectively. Without AOT less than 0.05, the 
mean relative errors were 37.8%, 30.8% and 30.0% 
respectively. The percentages of retrievals within the expected 
uncertainties were 62.5%, 46.9% and 46.9% respectively, and 
increased to 51.7% for 658 and 860nm without AOT less than 
0.05. For spring (Mar. to May, 22 match-ups), the linear fitting 
results were y=0.804612x+0.050988(R 2 =0.77), 
y=0.862697x+0.034432(R 2 =0.75) and 
y=0.851294x+0.012003(R 2 =0.71) respectively, and the mean 
relative errors were reduced to 16.4%, 18.2% and 19.3%. 
Using 45km radius and ±1.0h as the sampling window, there 
were 120 match-ups and the validation results were relatively 
poor with large data dispersion, and R 2 of 3 channels were all 
less than 0.3 while exceeding 0.6 without data from “Taichung” 
in the linear fitting. The RMSE was 0.23, 0.20 and 0.17 
respectively for 550, 658 and 860nm with the best result 0.11, 
0.09 and 0.07 at “Hong Kong PolyU” and worst result at “NCU 
Taiwan”. The mean relative errors all exceeded 55% with the 
best result 23.8%, 24.0% and 25.7% at “Hong Kong PolyU”. 
The percentages of retrievals within the expected uncertainties 
were all less than 50% with the best result of 61.5% at “NCU 
Taiwan”. 
According to former researches, the retrieval performance is 
relatively poor in Asia-Pacific region compared with that of 
global region. Although our validation result is not as good as 
that of former similar region, it should be considered that there 
are differences of sites and time span used in these studies. 
Generally speaking, the MODIS AOT product over ocean 
distributed by NSMC can assure certain precision, but the 
validation is not quite satisfying. 
There appeared relatively large errors, dispersion and intercepts 
in linear fitting in the validation. Considering that the aerosol 
algorithm used in NSMC is almost the same as that of NASA, 
we conclude that calibration of LIB data may be the important 
error source besides the influence of turbid water reflectance 
near shore, cloud contamination, aerosol model suitability, and 
the representative ability of aerosols over ocean by nearby land- 
based observation. 
REFERENCES 
Eck, T. F., Hokben, B. N., Reid, J. S., et al, 1999. Wavelength 
dependence of t he optical depth of biomass burning, urban, and 
desert dust aerosols. Journal of Geophysical Research, 104, pp. 
31,333-31,349. 
Choku, C., Chu, D. A., Mattoo, S. M., 2002. A spatio-temporal 
approach for global validation and analysis of MODIS aerosol 
products. Geophysical Research Letters, 29 (12), pp. MOD 121- 
MOD 124. 
Remer, L. A., Tanre, D., Kaufman, Y. J., et al, 2002. Validation 
of MODIS aerosol retrieval over ocean. Geophysical Research 
Letters, 29 (12), pp. MOD321-MOD324. 
Remer, L. A., Tanre, D., Kaufman, Y. J., et al, 2005. The 
MODIS aerosol algorithm, products and validation. Journal of 
the Atmospheric Sciences, Special Issue, 62, pp. 947-973. 
Chen, B., Yang Y., 2005. MODIS aerosol optical thickness 
validation around Taiwan strait and adjacent region. ACTA 
Oceanologica Sinica, 27(6), pp. 170-176. 
Tanre, D., Kaufman, Y. J., Herman, M., et al, 1997. Remote 
sensing of aerosol properties over oceans using the 
MODIS/EOS spectral radiances. Journal of Geophysical 
Research, 102, pp. 16,971-16,988 
ACKNOWLEDGEMENTS 
This research was supported by the National Natural Science 
Foundation of China (project 40606043) and National Basic 
Research Program of China (project 2006CB40370) and New- 
Tech Popularization Program of China Meteorological 
Administration (project CMATG2006Z02).