极地研究

极地研究 ›› 2018, Vol. 30 ›› Issue (1): 67-76.DOI: 10.13679/j.jdyj.20170005

• 研究论文 • 上一篇    下一篇

基于改进FCLS算法的南极海冰密集度估算及算法比较

宋翔宇1  刘婷婷1  王泽民1  刘艳霞2   

  1. 1. 武汉大学中国南极测绘研究中心, 湖北 武汉 430079
    2. 武汉市测绘研究院, 湖北 武汉 430022
  • 收稿日期:2017-01-20 修回日期:2017-04-09 出版日期:2018-03-30 发布日期:2018-03-30
  • 通讯作者: 刘婷婷
  • 基金资助:

    国家自然科学基金(41676179, 41531069)、南北极环境综合考察与评估专项(CHINARE2016-02-04)资助

Estimation of Antarctic sea ice concentration based on an improved FCLS algorithm

Song Xiangyu1, Liu Tingting1, Wang Zemin1, Liu Yanxia2   

  • Received:2017-01-20 Revised:2017-04-09 Online:2018-03-30 Published:2018-03-30
  • Contact: Tingting Liu

PDF (PC)

491

摘要：

海冰具有良好的热力隔绝效应, 它通过影响海洋和大气的热交换进而影响全球的气候变化。海冰密
集度是极区海冰研究的重要指标之一。为实现高空间分辨率多类型海冰密集度的估算, 本文将亮温极化梯
度率和光谱梯度率引入基于全约束最小二乘法(fully constrained least squares, FCLS)的海冰密集度估算方法,
并利用南极海冰过程与气候计划(Antarctic Sea Ice Processes and Climate, ASPeCt)对改进方法的精度进行验
证, 然后与NASA Team2(NT2)算法和ARTIST Sea Ice(ASI)算法获得的海冰密集度结果进行了对比分析。结
果显示, 3 种算法中本研究的方法精度最高, 全年均方差13.8%, 偏差为−0.7%; 改进的方法对多年冰的估算
精度优于一年冰。

关键词: 南极, 被动微波数据, 海冰密集度, FCLS

Abstract:

Sea ice has a heat isolation effect that could influence global climate change via modification of the exchange of heat
between the ocean and the atmosphere. Sea ice concentration is one of the most important indicators in polar sea ice research.
To achieve high spatial resolution in estimations of multitype sea ice concentrations, this study introduced a polarization gradient
rate and a spectral gradient rate into a fully constrained least squares (FCLS) algorithm. Accuracy was validated using
Antarctic Sea Ice Processes and Climate(ASPeCt) data, and the results were compared with NASA Team2(NT2) and Arctic
Radiation and Turbulence Interaction Study Sea Ice(ASI) algorithms. The root mean square error(RMSE) and bias of the improved
FCLS algorithm throughout the year were 13.8% and −0.7%, respectively. The accuracy of the improved algorithm
was found better for multiyear ice than first-year ice.

Key words: Antarctic, passive microwave data, sea ice concentration, FCLS