Operational Land Cover Products From VIIRS in the NPP/NPOESS Era

Accurate and repeatable information about the actual distribution of land surface types on the Earth's surface is critical to many applications ranging from global change studies and weather forecasting to disaster monitoring and resource management. Here we describe the theoretical bases and illustrate the algorithms which retrieve 17 land cover types from data acquired by the Visible/Infrared Imager/ Radiometer Suite (VIIRS) on an operational basis. This instrument is to be put in orbit early in the 21st century as a part of the National Polar-orbiting Operational Environmental Satellite System (NPOESS). The Surface Type Environmental Data Record (EDR) will be produced at the highest spatial resolution common to the VIIRS bands used (approximately 1 km). The operational EDR was developed around the strategy of producing the best possible global 1km land cover classification product possible, called the VIIRS Quarterly Surface Type Intermediate Product (QSTIP), from a full year of gridded, VIIRS data. The QSTIP will be produced every three months from the accumulation of the previous 12 months of VIIRS data, and will continue the heritage of global land cover products from the Advanced Very High Resolution Radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer (MODIS). During each successive three-month period, the VIIRS QSTIP will be re-delivered for every VIIRS orbit in conjunction with the current VIIRS Vegetation Index, Snow Cover, and Active Fires EDRs. The current fraction of green vegetation cover present per cell will also be computed in near real-time as a part of this EDR, further accommodating users who may require instantaneous information about the surface conditions associated with each surface type. The VIIRS Quarterly Surface Types algorithm will be run in a supervised classification mode with a decision tree classifier, using global training data specifically tailored to the 17 surface types, and temporal metrics developed from 12 months of VIIRS visible/infrared, and thermal spectral band information. The reasoning for using a full year of highly processed data was to exploit not only the spectral differences between the various cover types but also phenological differences, which substantially improved the classification. Results from the current Surface Type EDR and QST IP algorithms will be shown using regional scale MODIS data and continental/global scale data from MODIS and/or AVHRR.