项目名称: 基于不确定边界建模的南极海冰冰缘动态变化研究
项目编号: No.41301463
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 天文学、地球科学
项目作者: 赵羲
作者单位: 武汉大学
项目金额: 25万元
中文摘要: 从大尺度被动微波遥感影像上提取的海冰密集度反映了南极海冰的空间分布具有从开阔海域到密集海冰逐渐过渡的特点,采用传统单阈值法确定的海冰范围不仅降低了海冰变化分析的准确度,也难以描述海冰冰缘的模糊过渡性特征。本项目针对这一问题,以"不确定边界建模-海冰冰缘提取-冰缘变化与评估"为研究主线,采用随机集方法对海冰进行不确定边界建模;从中小尺度的海冰观测数据获取训练和验证样本,更新随时间变化的不确定边界模型的参数;分析基于不确定目标模型提取的冰缘长度、冰间湖个数、面积等特征和其随时间的变化规律,并对冰缘目标在不同时间和地理区域的的不确定程度做定量化描述,以及评估不确定信息对海冰长期变化分析的影响。本研究可以在不改变海冰密集度数据源的情况下提高提取海冰冰缘位置的准确性,同时可以提高后续的海冰时空变化分析的精度,在海冰监测、气候变化分析、极区考察、生态环境评价等方面具有重要的应用
中文关键词: 海冰;边缘;密集度;南极;随机集
英文摘要: The Antarctic sea ice edge zone observed from passive microwave satellite images (6km resolution) can be either diffuse or compact, and often shows changing transition zones from the open ocean to dense ice concentration zone. Traditional single threshold classification simplifies the extent determination of sea ice as crisp objects and consequently may impacts on the spatial temporal analysis of the sea ice change. The aim of this project is to improve the capability and accuracy of identifying Antarctic sea ice extent, as well as to assess the quality of spatial-temporal change results based on large scale image data. The project consists of three steps: uncertain object modeling - object extraction - spatial-temporal change analysis. Random set theory is chosen to model the uncertain boundary of sea ice objects as observed from passive microwave satellite images and their sea ice concentration data. Secondly, satellite images with a finer resolution are used as training and testing data to update the random set model using Basysian statistics. Thirdly, the spatotemporal changes of sea ice from the random set modeling results are analyzed to obtain detailed information on dynamics of sea ice edge length, location and number of polynya. Finally, the degree of uncertainty in ice edge is quantifed in space and ti
英文关键词: Sea ice;edge;ice concentration;Antarctic;random set