The methodology of CART decision tree is utilized to retrieve the sea ice area from MODIS satellite remote sensing data, which realizes the automatic classification by multi -bands of visible light, near infrared light and thermal infrared light, and effectively eliminates the misjudgment by traditional threshold methods when retrieving sea ice in the marine environment such as high sediment suspension. The retrieved results are verified by the Small Satellite Constellation for Environment and Disaster Monitoring and Forecasting (HJ-1A/1B) with the high spatial resolution,which suggests that the retrieved data based on the CART decision tree have the higher resolution. The process of sea ice growing and melting in the Liaodong Bay during winter since 2003 is retrieved by this methodology, which provides precise and reliable data for analyzing the relationship between meteorologic factor and sea ice evolution.%采用 CART 决策树方法从 MODIS 卫星遥感数据反演海冰面积，同时对可见光、近红外和热红外多波段进行自动分类计算，有效消除了传统阈值法在反演高悬沙等海洋环境时出现的海冰误判。经较高分辨率的中国环境与灾害监测预报小卫星（HJ-1A/1B）校验，结果表明基于决策树方法所得出的反演数据具有较高精度。利用建立的 CART 决策树方法反演了2003年以来辽东湾冬季海冰面积的生消变化过程，为进一步分析和建立气象因素与海冰演化规律的关系提供了精确可靠的基础资料。
展开▼
