科研之友
微信
新浪微博
Facebook
分享链接
摘要
Arctic sea ice has undergone a significant decline in recent years. Previous studies have demonstrated that the annual sea ice cycle has experienced earlier melt and later freeze up, leading to a significant reduction in minimum sea ice extents and the lengthening of the melting season. The Arctic is being transformed into a regime of widespread seasonal ice with a large loss of old and thick multiyear ice in recent years. However, the sea ice change exhibits considerable interannual and regional variability at different spatial and temporal scales. In this study, we present a new method for hypertemporal sea ice data change detection based on the annual sea ice concentration (SIC) profile for the melt months of each year. A decision tree-based classification is adopted to group pixels with similar annual SIC profiles, and a phenology map of each year is generated for visualization. The phenoregion map visualizes the spatial and temporal configurations of ice melt process for a year. The change detection objective is achieved by comparing the phenoregion number of the same pixel in different years. The algorithm further leads to interpretation of anomalies to obtain change maps at the pixel level. Compared to previous sea ice studies that mainly focused on a particular spatial region and commonly use time period averages, the proposed pixel-based approach has the potential to map sea ice data change both temporally and spatially.
- 出版日期2016-7
- 单位华中师范大学
