The snow melt onset date represents an important transitional point in the Arctic energy balance, when the surface albedo decreases rapidly and surface energy absorption increases rapidly in response to the appearance of liquid water. An improved understanding of the interannual melt pattern is valuable for climate change detection, climate simulations, and model validation in the Arctic region. Passive microwave satellite data are indispensable in this task because they are relatively unaffected by cloud cover, not reliant on solar illumination, and have relatively high repeat coverage capabilities. In this project, snow melt onset dates are derived from 1979-1998 using horizontal polarizations from 18 and 37 GHz from the Scanning Multichannel Microwave Radiometer (SMMR) and 19 and 37 GHz from the Special Sensor Microwave/Imager (SSM/I) satellites.

Interannual variations in the snow melt onset dates are large, ranging from approximately 10 days in the central Arctic, to greater than 60 days in southern regions such as the Beaufort Sea. Additionally, trends towards earlier melt onset dates are observed in the West Central Arctic Ocean, the Lincoln Sea, the Beaufort Sea, the Canadian Arctic Archipelago, and Baffin Bay.

The underlying cause of interannual variations in the snow melt onset date over sea ice is a combination of high-frequency synoptic events that lie within the larger atmospheric patterns, described by teleconnection indices. In particular, earlier melt onset dates across much of the western Arctic are associated with positive phases of the Arctic Oscillation (AO), while delayed melt is associated with negative phases of the AO. Furthermore, earlier melt in portions of Baffin Bay are linked to positive phases of the North Atlantic Oscillation (NAO) and negative phases of the Pacific-North American Anomaly (PNA) pattern, while earlier melt in the western Arctic is linked to positive phases of the Pacific-North American Anomaly (PNA). In all cases, analysis of the 500 mb height patterns suggests a combination of abnormally low heights and thermal advection are associated with earlier than average melt onset. At a more regional scale, analysis with the Penn State/NCAR mesoscale model illustrates that increased longwave radiation, associated with a low-pressure cell, initiates melt onset.