The presence of clouds affects the radiance observed by optical sensors at the top of the atmosphere, and thus prevents accurate estimation of physical parameters on the land surface through optical satellite data. In order to overcome this problem, composite methods (MVC, MVI, etc.) were developed to reduce the effect of clouds. However, it is still uncertain whether the use of composite data completely eliminates the cloud interference. Therefore, in this research, the White object Index (WI) is proposed to estimate the cloud coverage for each pixel of MODIS data. Here White objects include both clouds and snow, so it is necessary to isolate the snow factor using the Normalized Difference Snow Index (NDSI) .
Using satellite data, the spectral characteristics of cloud, vegetation and soil are analyzed and compared. Then, the ratio of cloud cover, the WI, is obtained from these characteristics. The variation between visible band and shortwave infrared band (2.1μm) has a large effect on the calculation of the WI. In order to test the effectiveness of this system, the results obtained from the WI calculations were thus compared with the cloud ratio from a synthetic mixture model composed of known ratios of cloud, vegetation and soil spectrums. The results showed only a 5% maximum error, indicating that in analyzing composite MODIS data, WI is an effective index for determining the amount of remaining cloud cover. In addition, the WI results were compared with a cloud mask image (MODIS product: MOD35), indicating that the WI produced a more detailed cloud distribution than the MOD35.