P6.11
Multispectral automated snow identification and monitoring algorithm
Peter Romanov, NOAA/NESDIS, Camp Springs, MD; and G. Gutman and I. Csiszar
The current technology for an operational production of global and continental scale snow cover maps implemented at NOAA NESDIS within an Interactive Snow and Ice Mapping System (IMS). The IMS is based on an interactive visual analysis of data from different sources and requires substantial day-to-day human efforts. NOAA analysts manually draw daily snow maps covering Northern Hemisphere, relying mainly on satellite visible imagery from geostationary satellites. The primary objective of the current study was to develop an automated system that could produce a first-guess snow cover map to facilitate the human effort in the NOAA daily visual snow analysis within the IMS. The proposed system will increase the spatial resolution of the snow product and eliminate the IMS product's subjectivity, i.e. dependence of the product on analyst's skill. The increase of the resolution will enable production of snow fraction for each 20-30 km grid cell of numerical weather prediction models. The proposed automated snow mapping system is based on a synergy of microwave, visible, mid-infrared and infrared observations, utilizing the frequent views throughout the day by geostationary satellite and the ability to observe ground surface through clouds with microwave sensors. The devised technique has been applied to process satellite data over North America during the 1998-1999 snow season. Both GOES-East and -West satellite observations have been used to make the full coverage of the North American continent. The NESDIS Special Sensor Microwave Imager (SSM/I) operational snow cover product has been used to complement the GOES observations in cloudy conditions. The results show strong potential for automated daily snow monitoring with a blended GOES-SSM/I product. To evaluate the system performance as compared to the current NOAA analysis within IMS, the derived snow maps were validated against ground snow reports. Validation tests have shown that in 85% of cases the automated GOES-SSM/I blended snow product fits exactly the ground reports. The accuracy of the blended product was found to be superior to the operational SSM/I product and sometimes higher than that of the IMS product.
Poster Session 6, Environmental Applications of Land and Oceanic Remote Sensing
Friday, 14 January 2000, 10:00 AM-12:00 PM
Previous paper Next paper