Identification and Removal of Clouds from MODIS Data

Application of satellite remote sensing has been very frequent in past few decades. However, the most important constraint in monitoring through satellite sensor is the identification and removal of cloud cover. In this study ability of the Moderate Resolution Imaging Spectroradiometer (MODIS) to monitor surface features is investigated. For a range of wavelength, both snow and cloud has similar reflectance and often a comparable temperature. Snow cover mapping algorithms, based on land cover information, are developed and compared with MODIS snow product for the period of 2000 to 2008 in current study. Algorithms involved the use of Normalized Difference Snow Index (NDSI), water pixel removal, dark pixel removal and cloud masking based on liberal approach. Individual analysis revealed that the algorithm is successful in identifying snow even in the presence of clouds where ice or water at sub-pixel level may falsely indicate cloud. On average, manual assessment classified 2.2% more pixels as snow as compared to the snow product or MODIS snow product. In comparison to the pixels comprising clouds, MODIS snow product has identified 2.9% more pixels as cloudy.