Monitoring Daily Snow Cover over the Western Arctic using AVHRR and MODIS Imagery

Snow cover is an essential climate variable that relates changes in air temperature and precipitation to surface albedo and the timing of streamflow. The Global Climate Observing System has called for global daily snow cover maps at better than 1km resolution over periods suitable for climate studies. Recent snow cover maps from satellite imagery suffer limitations related to resolution (e.g. passive microwave based products). cloud cover (e.g. optical sensor based products) or temporal extent (e.g. synthesis products using multiple sensors). We present an approach for producing continuous daily snow cover maps from passive optical sensors such as NOAA-AVHRR and MODIS. The approach relies on an initial cloud screening of single swaths to generate temporal spectral signatures for snow and snow free conditions. Adaptive temporal compositing, based on cloud probabilities, is applied to generate estimates of daily trends in reflectance in visible bands. Snow cover is then estimated using a linear mixture model corrected for observation geometry effects.

Our algorithm is applied to a 25 year time series of 1km NOAA-AVHRR data over the western Arctic as well as a sub-sampled time series of MODIS imagery over the Arctic to map snow cover on a daily basis from 1981 to present. Performance is assessed by comparison to in-situ measurements with a focus on the snow melt season. Plans to extend this approach to a complete arctic coverage using Polar Pathfinder datasets and full resolution MODIS imagery are discussed.