Current GOES cloud climatologies are currently hindered during the winter season due to snow contamination. Bright snow cover can be confused with cloud, especially if it falls and melts within the period covered by the climatology. Even static snow fields, such as those over mountain ranges, can effect the accuracy of cloud detection. Since the regional climatologies are processed monthly, this is a problem for much of the US Midwest during the winter months.
GOES-R will have an additional 1.6 µm channel that has been shown on both MODIS and SEVIRI to aid in snow detection. This project is to investigate whether the addition of these methods to the current visible cloud detection scheme will improve this problem area.
To test this theory, SEVIRI data over the Europe was collected and processed for 1015, 1115, and 1215Z during January 2007. After reviewing the literature, a snow detection scheme using the 0.64 µm, 0.8 µm and 1.6 µm channels on SEVIRI was developed. These match up with future channels on GOES-R. The snow detection method was incorporated with the current cloud detection algorithm used over the US for a new algorithm. Then both the current method and the new method with snow detection was ran over the US data and compared.
Comparison shows that snow previously mistaken for cloud in current method was correctly identified as snow in the new method, both in areas with consistent snow cover (the Alps) and in areas where the snow cover was intermittent (Northern Spain).