【作者】 王星东；

【导师】 熊章强； 李新武；

【作者基本信息】 中南大学 ， 地质资源与地质工程， 2013， 博士

【摘要】 南极冰盖对全球热量平衡起着重要作用,控制着大气与地表的热量、水汽和能量交换,直接影响着全球大气环流和气候的变化。冰盖消融将引起冰雪表面湿度的变化,进而改变冰流的运动、加速边缘冰架的崩塌。因此,探测南极冰盖冻融的分布对于研究全球的气候变化至关重要。微波遥感具有全天候、穿透性强以及对地物介电特性敏感等优势,是对地观测中十分重要的前沿领域。其中,微波辐射计的亮度温度和微波散射计的后向散射系数对冰盖液态水含量的变化非常敏感,随着融化的开始和结束,液态水含量的增大和减小,亮度温度和后向散射系数均发生剧烈变化。论文根据冰雪微波的这一特性,针对微波辐射计和微波散射计冰盖冻融探测算法中存在的问题进行研究,同时开展主被动协同的冰盖冻融的研究,具体的内容如下：(1)分析了微波辐射计现有的XPGR算法和简单物理模型算法在冰盖冻融探测中存在的问题,针对这些问题,提出了改进的辐射计简单物理模型算法、XPGR结合小波变换算法和辐射计简单物理模型结合小波变换算法。这些算法解决了原算法的实用性和可操作性差的问题,实现了南极地区冰盖冻融的自动化监测。并在此基础上利用南极自动气象站数据进行了地面验证。地面验证结果表明,这些算法对南极冰盖冻融探测是有效的。(2)微波散射计简单物理模冰盖冻融探测算法确定干湿雪分类阈值时需实测数据,为提高其实用性和可操作性及使用范围,根据干湿雪信号的特征提出了基于广义高斯模型的自动阈值分割的改进的散射计简单物理模型冰盖冻融探测方法。另外,微波散射计的后向散射系数在冰盖融化和冻结时均有剧烈的变化幅度,而小波变换能灵活地进行边缘提取,基于以上两点,提出了小波边缘检测的散射计南极冰盖冻融探测算法；此外,冰盖处于融化状态的情况下后向散射系数易受外界条件的干扰,为了去除这些干扰对小波边缘检测的冰盖冻融探测的影响,提出了数学形态学结合小波边缘检测的南极冰盖冻融算法,这些算法均能自动化探测南极冰盖冻融,为冰盖冻融提供了方法学的支持和补充。(3)在将主被动微波结合进行冰盖冻融探测的研究方面,一些学者已做了一定的工作,但所有做法基本上都是分别进行数据处理,得到各自的冻融分布图,然后对结果进行综合分析,在方法学上没有将主动微波和被动微波有机结合起来。本文基于微波辐射计的可靠性和微波散射计的灵敏性和高空间分辨率,提出了利用小波变换将两者协同起来提高冰盖冻融探测精度的协同算法一,即利用微波散射计灵敏度高能较早地探测到冰盖的表面融化和冻结和微波辐射计可靠性高能准确地得到融化区域。即利用微波辐射计确定融化区域,通过微波散射计确定融化开始和结束时间,充分利用两者的优势从而有效提高融化开始、结束和持续时间的探测精度。协同算法二利用微波辐射计可靠性高,受粗糙度等外界因素影响较小,微波散射计空间分辨率高但受外界因素影响较大。基于微波辐射计和微波散射计的简单融化物理模型,提出了利用微波辐射计的可靠性来修正微波散射计的后向散射系数的主被动协同的冰盖冻融探测方法。实验结果表面这两种协同算法在一定程度上提高了南极冰盖冻融的探测精度。(4)分别从空间和时间上对南极冰盖的冻融结果进行了分析,从空间分布来看,南极冰盖的融化区域主要分布在南极边缘的各个冰架区,融化强度受地物覆盖类型、地理位置和海拔等因素的影响；从时间分布上来看,南极冰盖的融化面积年际变化较大,1991年的融化面积最大为1518750km2,1999年融化面积最小为565000km2,且融化面积的年际变化具有周期性；南极冰盖的融化具有很强的季节性,融化一般集中在11月至次年2月,在1月达到融化顶峰,3-8月份融化面积非常小。图49幅,表11个,参考文献138篇更多还原

【Abstract】 Antarctic ice-sheet contributes significantly to the global heat budget by controlling the exchange of heat, moisture, and momentum at the surface-atmosphere interface, which directly influences the global atmospheric circulation and climate change. Ice-sheet melt will cause snow humidity increase, which will accelerate the disintegration and movement of ice sheet. As a result, detecting Antarctic ice-sheet melt is essential for global climate change research.Microwave remote sensing is an all-weather technology, which has strong penetration capability and it is sensitive to dielectric properties. All these advantages make it an important frontier of earth observing. Furthermore, microwave brightness temperature and microwave backscatter coefficient are so sensitive to the water content variation that, with the onset and end of snowmelt, brightness temperature and backscatter coefficient experience dramatic variation. On the basis of snow microwave characteristics, I provide a review on the principle of detecting snowmelt with passive microwave remote sensing data and active microwave remote sensing data, and then introduced some passive microwave algorithms and active microwave algorithms for Antarctic ice-sheet melt monitoring and for long-time series melting result derivation, at the same time, the synergy of microwave radiometer and scatterometer was researched for ice-sheet melt monitoring, and the specific contents are as follows:(1) The problems of the current XPGR algorithm and the simple physical model algorithm for the ice-sheet melt detection of microwave radiometer are analyzed, in order to solve the problems, the improved radiometer simple physical model, XPGR combined with wavelet transform algorithm and radiometer simple physical model combined with wavelet transform algorithm are proposed to improves the computational efficiency, usability and operability of detecting the ice-sheet melt detection. The algorithms do not rely on the actual melt information and can automatically select many samples. The ground verification is done based on the automatic weather station data. The results show that the methods can be effectively used for the detection of Antarctic ice-sheet melt.(2) On the basis of the simple snowmelt physical model for microwave scatterometer, a new automatic threshold segmentation algorithm of Antarctic ice-sheet freeze-thaw detection was proposed, which did not depend on the field observations. That was the histogram for the data of the physical model by the use of generalized Gaussian model to automatically get Antarctic melt distribution. The algorithm improves the computational efficiency, usability and operability of the freeze-thaw detection because the algorithm does not rely on the actual melt information and can automatically select more samples. In addition, in according to the characteristic of backscattering coefficient having dramatic changes with the event of melt or freeze and the characteristic of wavelet transform edge extraction, wavelet edge detection for Antarctic ice-sheet freeze-thaw detection algorithm is proposed. Moreover, the backscattering coefficient is not stable in the melt state, and the mathematical morphology can filer the signal with the edge preserved. So the mathematical morphology combined with wavelet transform algorithm was proposed, and these algorithms can automatically detect the Antarctic ice-sheet melt, which provide methodological support and supplement for the global ice-sheet freeze-thaw detection.(3) Passive and active microwave measurements have been used in various studies to detect melt based on their sensitivity to liquid water present in snow. The scatterometer is more sensitive to surface melt than passive microwave observations. The radiometer is more reliable than active microwave. In order to effectively carry on the ice-sheet freeze-thaw detection, the active and passive microwave sensors are efficiently combined based on the sensitivity and the high spatial resolution of the scatterometer and the reliability of the radiometer, two new Antarctic freeze-thaw synergistic detection methods were proposed. One was based on the high sensitivity of the scatterometer and the high reliability of the radiometer by the use of edge detection model to automatically extract the edge information to get the distribution of Antarctic melt onset date, melt duration and melt end date. The other was based on the high spatial resolution of the scatterometer and the reliability of the radiometer through the physical models of the active and passive microwave data sets. Both new algorithms make use of the advantages of active and passive microwave sensors. The results show that the algorithms improve the detection accuracy of melt onset date, melt duration, melt end date and melt distribution.(4) The spatial distribution of melt areas shows that the majority of melt areas are located on the edge of Antarctic ice shelf region. It is affected by land cover type, surface elevation and geographic location. The temporal distribution of melt areas shows that the Antarctic ice sheet melt varies with years with some rule. The melt areas in1991are1518750km2,and its areas are largest. The melt areas in1999are565000km2,and its areas are smallest. In addition, the Antarctic ice sheet melt varies with seasons. It is particularly acute in summer, peaking at December and January, staying low in March.更多还原