Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4460290 | Remote Sensing of Environment | 2009 | 19 Pages |
Results from several previously published algorithms for wet snow detection in Antarctica from K-band spaceborne brightness temperature are compared and evaluated vs. estimates of wet snow conditions from ground measurements. In addition, a new physically-driven algorithm, in which the detectable liquid water content is assumed constant, is proposed and assessed. All algorithms are also evaluated by analyzing their results during collapses of ice shelves. Two algorithms are selected for deriving updated trends of melting index (MI, the number of melting days times the area subject to melting) between 1979 and 2008 over the whole Antarctica and at sub-continental scales. In the first algorithm wet snow is identified when brightness temperature exceeds the mean of winter brightness temperature plus 30 K and the second is the new model-based approach described here. Both negative and positive MI trends are obtained, depending on the algorithm used. A high number of melting days (up to 100 days) are detected over the Wilkins ice shelf, the Peninsula and the George VI ice shelf. Over East Antarctica, the West and Amery ice shelves are subject to melting for a maximum of approximately 50 days. Positive trends of number of melting days are detected over most of the West Antarctica, with peak values up to 1.2 days/year over the Larsen C ice shelf, 1.8 days/year over the George VI ice shelf and 0.55 days/year over the Wilkins ice shelf area. The correlation between MI values and December–January (DJ) averaged air/surface temperature over selected locations show values ranging between ∼ 0.8 and ∼ 0.4. Results suggest that a 1 °C increase in the monthly averaged DJ air/surface temperature corresponds to an average MI increase of approximately 2·106 × km2 × day.