A Comparison between Snow Cover Products Derived from Visible and Microwave Satellite Observation

Abstract

A comparison between two satellite-derived snow cover products demonstrates the strengths and weakness of each procedure. The current NESDIS operational product is subjectively derived from visible satellite imagery. The analysis is performed once a week, using the most recent clear view of the surface. The experimental product is objectively derived from daily microwave measurements observed by polar-orbiting satellites. The operational product uses a high albedo in the visible spectrum to identify snow cover, whereas the experimental product uses a high albedo in the visible spectrum to identify snow cover, whereas the experimental product uses a passive microwave scattering signature. Comparisons between the operational and experimental product show good agreement in the extent and distribution of snow cover during the middle of the winter and summer seasons. However, the agreement weakness in the transition seasons and along the southern edge of the snowpack. The analysis suggests that the operational procedure is better at observing snow under a densely vegetated canopy, whereas the experimental procedure is better over rugged terrain and persistent cloud cover. The experimental product is also better at observing rapid fluctuations in the snowpack, since it has higher temporal resolution and can see through nonprecipitating clouds. An integration of the two products, currently under development by NOAA/Office of Research and Application and the National Meteorological Center, would represent true snow cover better than either single procedure. However, it would probably introduce discontinuity into the 5-yr time series of the current operational product, which is the longest record of snow cover over the Northern Hemisphere. Averaging the experimental product between two consecutive weeks effectively brings the two datasets into closer agreement throughout the global time series. However, this technique does not resolve the regional biases between the two datasets into closer agreement throughout the global time series. However, this technique does not resolve the regional biases between the two datasets. Surface observations would help identify the source of these biases; unfortunately, these reports are severely limited over many areas.