摘要
利用2012年国产FY2-E静止卫星数据,结合地面能见度观测资料,分析北部湾地区晴天海雾、云、洋面的辐射性质、光谱及纹理特征,利用面向对象分类方法对白天海雾进行检测提取。结果表明:FY2-E卫星影像对北部湾地区晴天海雾的区分度较高,其光谱及纹理特性与洋面、中高云多无交叠现象;但大雾与低云区有部分混淆。利用面向对象分类方法,在选择合适分割尺度的基础上根据光谱Spectral_Mean和纹理Texture_Mean两个影像特征可将海雾进行有效分离。精度分析结果表明,2012年全年晴空海雾监测误警率为9.1%,命中率为76.9%,成功指标为71.4%,总体监测效果较理想。
Domestic FY2-E geostationary satellite data, combining with ground visibility observation data are used to analyze radiation, spectrum and texture features of sea fog, cloud and ocean of sunny days. Then, object- oriented classification method is applied to sea fog detection and extraction. The results show that FY2-E satellite images possess a relative strong ability to distinguish of sea fog. Overlap regions of spectrum and texture features between sea fog and middle/high clouds and sea surface are rare but low cloud area and fog are easily to confuse in some regions. With object-oriented classification method and appropriate segmentation scale, sea fog can be effectively separated based on spectrum feature of spectral_mean and texture feature of texture_mean. The overall sea fog monitoring effect for the whole year 2012 is ideal with false alarm ratio 9. 1% , probability of detection 76. 9% , critical success index 71. 4%.
出处
《科学技术与工程》
北大核心
2017年第19期8-11,共4页
Science Technology and Engineering
基金
广西自然学科学基金(2013GXNSFBA019225)
广西科学研究与技术开发计划(桂科攻14124004-4-9)
广西科学研究与技术开发计划(桂科攻1598017-13)资助
广西自然科学基金青年基金(2014GXNSFBA118216)
华南区域气象中心科技攻关项目(GRMC2014Z03)
关键词
海雾
FY2-E
可见光
面向对象
北部湾
sea fog FY2-E visible band object-oriented classification Beibuwan
