基于SAR图像的冰雪覆盖信息提取方法研究
发布时间：2018-12-11 20:20
【摘要】：随着气候变暖和极端天气增多,冰雪制图和雪灾评估成为全球研究的热点。合成孔径雷达(SAR)具有全天候、全天时、高分辨率的特点,在积雪制图研究中有着特殊的优势,弥补光学卫星的不足。本文的主要研究内容为利用SAR图像提取积雪覆盖面积,并对提取结果进行验证分析。合成孔径雷达干涉(InSAR)技术,可用于地表相干性评估。本文选取了不同地区、不同地形、不同时间段的多组ERS-1/2串接干涉数据对,统计分析雪及其它地物的相干性。结果表明,对于时间基线为一天的干涉对,积雪、水域、冰面、森林等地物相干值较低,而其它地物的相干值相对较高。本文还利用ERS-2和ASAR六组时间基线较长的干涉对,分析普莫雍错湖地区地物的相干性。与时间间隔仅一天的相干性对比发现,地物在间隔35天和70天的相干性下降明显。本文根据积雪与其它地物的相干性差异,采取相干性阈值分类的方法,提取了多组实验地区的积雪覆盖面积。将SAR图像积雪提取结果与同时期的光学图像积雪分类结果对比,计算积雪提取的准确度,并分析错误分类的原因。实验表明,InSAR技术可以有效地提取积雪面积信息。时间基线较短的干涉对,有利于提高积雪面积的提取精度。然而这种方法需要实验地区的先验信息,如气候、地形等,同时复杂地表对积雪提取精度影响较大。本文还分析了SAR图像叠掩和阴影对积雪分类的影响,加入了叠掩和阴影区的判别。结果表明,因雪山的叠掩和阴影区也可能有积雪覆盖,若直接排除这类区域,并不能明显地提高分类精度。论文的最后一章,InSAR技术用于城镇地区的积雪检测。利用无雪和有雪时期多组ASAR数据,统计了乌鲁木齐地区的地物相干性。通过分析两组夏季时期的干涉对,发现城镇由于人类活动、农作物生长等因素,35天时间间隔的相干性非常低,但机场、大面积广场等特殊地物的相干性较高。本文还利用四组降雪期的干涉对、以及一组有雪和无雪的干涉对,分析降雪后城镇各类地物的相干性变化。结果表明无雪时相干性较高的机场、郊区荒地,在被积雪覆盖后相干性显著下降。可以利用城镇中这类特殊地物的相干性特征判断是否有雪,进而再进行积雪覆盖监测。
[Abstract]:With the increase of climate warming and extreme weather, ice mapping and snow disaster assessment have become the focus of global research. Synthetic Aperture Radar (SAR) has the characteristics of all-weather, all-day and high-resolution, so it has special advantages in snow mapping and makes up for the shortage of optical satellite. The main research content of this paper is to use SAR image to extract snow cover area and verify the result. Synthetic Aperture Radar (SAR) interferometric (InSAR) technique can be used to evaluate surface coherence. In this paper, the coherence of snow and other ground objects is statistically analyzed by selecting multiple sets of ERS-1/2 interference data pairs in different regions, different terrain and different time periods. The results show that the coherence values of snow, water, ice, forest and other ground objects are lower than those of other ground objects when the time baseline is one day. In this paper, the coherence of the features in Pumoyuncao Lake area is analyzed by using six groups of ERS-2 and ASAR interference pairs with long time baselines. Compared with the coherence of only one day, it was found that the coherence between 35 days and 70 days was obviously decreased. According to the difference of the coherence between snow cover and other ground objects, the snow cover area of many experimental areas was extracted by using the method of coherence threshold classification. The results of snow extraction from SAR images are compared with those of optical images in the same period, the accuracy of snow extraction is calculated, and the causes of false classification are analyzed. Experimental results show that InSAR can effectively extract snow area information. The interference pair with shorter time baseline is helpful to improve the accuracy of snow area extraction. However, this method requires prior information such as climate, topography and so on in the experimental area, and the complex surface has a great influence on the accuracy of snow extraction. In this paper, we also analyze the influence of SAR image overlay and shadow on snow classification, and add the discriminant between overlay and shadow region. The results show that snow cover may also exist in the overlay and shadow areas of the snow-covered mountains, and the classification accuracy can not be improved obviously if these areas are excluded directly. In the last chapter, InSAR technology is used to detect snow cover in urban areas. Based on the ASAR data of non-snow and snow-free periods, the coherence of features in Urumqi was analyzed. By analyzing the interference pairs of the two groups in summer, it is found that due to human activities, crop growth and other factors, the coherence of 35 days interval is very low, but the coherence of special features such as airports and large square is high. In this paper, four sets of interference pairs during snowfall period and one pair of interference pairs with and without snow are used to analyze the coherence changes of various features in cities and towns after snowfall. The results showed that the coherence of the airport and the suburban wasteland decreased significantly after snow cover in the absence of snow. It can be used to judge whether there is snow or not and then to monitor the snow cover by using the coherence characteristics of such special features in cities and towns.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52
本文编号：2373162
[Abstract]:With the increase of climate warming and extreme weather, ice mapping and snow disaster assessment have become the focus of global research. Synthetic Aperture Radar (SAR) has the characteristics of all-weather, all-day and high-resolution, so it has special advantages in snow mapping and makes up for the shortage of optical satellite. The main research content of this paper is to use SAR image to extract snow cover area and verify the result. Synthetic Aperture Radar (SAR) interferometric (InSAR) technique can be used to evaluate surface coherence. In this paper, the coherence of snow and other ground objects is statistically analyzed by selecting multiple sets of ERS-1/2 interference data pairs in different regions, different terrain and different time periods. The results show that the coherence values of snow, water, ice, forest and other ground objects are lower than those of other ground objects when the time baseline is one day. In this paper, the coherence of the features in Pumoyuncao Lake area is analyzed by using six groups of ERS-2 and ASAR interference pairs with long time baselines. Compared with the coherence of only one day, it was found that the coherence between 35 days and 70 days was obviously decreased. According to the difference of the coherence between snow cover and other ground objects, the snow cover area of many experimental areas was extracted by using the method of coherence threshold classification. The results of snow extraction from SAR images are compared with those of optical images in the same period, the accuracy of snow extraction is calculated, and the causes of false classification are analyzed. Experimental results show that InSAR can effectively extract snow area information. The interference pair with shorter time baseline is helpful to improve the accuracy of snow area extraction. However, this method requires prior information such as climate, topography and so on in the experimental area, and the complex surface has a great influence on the accuracy of snow extraction. In this paper, we also analyze the influence of SAR image overlay and shadow on snow classification, and add the discriminant between overlay and shadow region. The results show that snow cover may also exist in the overlay and shadow areas of the snow-covered mountains, and the classification accuracy can not be improved obviously if these areas are excluded directly. In the last chapter, InSAR technology is used to detect snow cover in urban areas. Based on the ASAR data of non-snow and snow-free periods, the coherence of features in Urumqi was analyzed. By analyzing the interference pairs of the two groups in summer, it is found that due to human activities, crop growth and other factors, the coherence of 35 days interval is very low, but the coherence of special features such as airports and large square is high. In this paper, four sets of interference pairs during snowfall period and one pair of interference pairs with and without snow are used to analyze the coherence changes of various features in cities and towns after snowfall. The results showed that the coherence of the airport and the suburban wasteland decreased significantly after snow cover in the absence of snow. It can be used to judge whether there is snow or not and then to monitor the snow cover by using the coherence characteristics of such special features in cities and towns.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52
【参考文献】
相关期刊论文 前1条
1 曾琪明,焦健;合成孔径雷达遥感原理及应用简介(三)[J];遥感信息;1998年02期
，本文编号：2373162
本文链接：https://www.wllwen.com/kejilunwen/wltx/2373162.html
