Gradation of MODIS Cloud Mask Algorithm Using Simultaneous ASTER Imagery

Usefulness and authenticity of satellite data are strongly related to weather conditions. Dust storm, atmospheric gases and especially the presence of clouds can considerably affect on the reflected energy from surface and encounter the reading of optical sensors with error. Cloud contaminated pixels usually increase the reflection of land covers and show their temperature less than the real one. The clouds smaller than pixels are not observable and cause an increase in pixel reflection and error. Considering that different clouds have diverse interactions with each other, it is possible to determine the amount of cloudiness of pixel by combining the analysis of different spectral bands of MODIS. By using this method, it is possible to determine the the cloud polluted pixels faster. For the first time, MODIS Cloud Mask algorithm was declared by Ackerman and his colleagues in 2006. Firstly the cloudy pixels were identified in MODIS image by the help of the presented five- step model in MODIS Cloud. Than for upgrading the model and determining the present percentage cloud in one pixel for those clouds smaller than pixel dimensions, brightness temperature of the pixels in band 14 of ASTER and band 31 of MODIS was compared and the cloud coverage percentage of every pixel was calculated. Model evaluation demonstrates more than 93% correlation between the real amounts of distracted cloud coverage in ASTER and the measured amount by the model, which is promising regarding local resolution.