【作者】 邓睿；

【导师】 黄敬峰；

【作者基本信息】 浙江大学 ， 农业资源利用， 2010， 博士

【摘要】 中国是世界上受台风影响最严重的国家之一,影响中国的台风主要来自西北太平洋。西北太平洋是全球唯一的一年四季都有台风形成,而且发生频数最多、分布范围最广的海域。台风影响范围很广,既能带来狂风、暴雨、风暴潮等灾害,又具有缓解干旱、降低高温等有益的作用。改革开放以来,中国的社会经济得到了飞速的发展,人口和社会财富高度集中,更容易受到台风的影响。研究台风的影响,意义深远。本研究首先利用1971-2008年登陆中国热带气旋的最佳路径资料,结合地理信息系统GIS(Geographic Information System)研究热带气旋在中国的时空分布特征；然后使用MODIS(Moderate Resolution Imaging Spectroradiometer)1B影像,经过一系列预处理,反演地表温度；再利用植被供水指数,结合AMSR-E (Advanced Microwave Scanning Radiometer-Earth Observing System)土壤湿度研究台风缓解干旱,并使用气象站点实测日最高气温和AMSR-E地表温度数据研究台风降低高温影响；最后基于HJ-1影像研究台风带来降雨引起水库水体面积变化和滑坡泥石流灾害,以及台风影响植被生长。研究的主要方法和结论如下：1登陆中国热带气旋的时空分布特征利用1971-2008年登陆中国热带气旋的最佳路径资料,从频数、强度和登陆地点等方面分析登陆中国热带气旋的时空分布特征。结果表明20世纪70年代初是热带气旋登陆中国的高发期,20世纪90年代末是热带气旋登陆中国的低发期。热带气旋登陆主要集中在7-9月,登陆热带气旋个数总体上以8月为中心,成对称减少分布。77%的登陆热带气旋强度都在风力10级以上。热带气旋登陆的地点主要在广东、台湾、海南、福建、浙江和广西等东南沿海省份。统计2004-2008年期间每次热带气旋每6小时的七级和十级风圈半径,分析风圈半径与热带气旋近中心最大风速之间的关系,重建历史热带气旋的风圈半径,并在GIS的支持下,可视化1971-2008年热带气旋在中国的空间分布,可以看出热带气旋对中国的影响从沿海到内陆逐渐减弱,海南、台湾以及其它东南沿海地区影响最为严重。2 MODIS 1B影像预处理及地表温度反演MODIS 1B是MODIS的1级产品,是将探测到的数据信号按一定比例缩放成16位整数值SI(Scaled Integer)保存的文件。利用MODIS 1B数据的SI值反演地表反射率,需要经过辐射校正(辐射定标和大气校正)、几何校正、去边缘重叠(bow-tie效应)等预处理,而且为了获取研究区全覆盖资料,通过检测去除云、多天影像去云后拼接得到研究区的反射率。地表温度通过分裂窗算法反演,该方法使用MODIS 1B影像的可见光、近红外和热红外波段,通过Planck方程由辐亮度计算亮度温度；通过大气水汽含量、温度校正函数和视角校正函数求算大气透过率；基于植被覆盖率来估算地表比辐射率。最后结合估算出来的亮度温度、大气透过率和地表比辐射率三个因素计算地表温度。3台风缓解干旱和降低高温影响的遥感评估台风带来的降雨能缓解干旱,而且由于台风期间云层增厚,减少了地表大气对太阳辐射的吸收,温度降低。本研究利用多源遥感数据评估台风缓解干旱和降低高温影响。(1)首先使用地面实测降雨量资料计算台风“海棠”影响前研究区的降水距平来分析台风影响前研究区的干旱状况；再利用MODIS 1B数据计算的归一化差值植被指数(Normalized Difference Vegetation Index, NDVI)和反演的地表温度(Land Surface Temperature, LST)计算植被供水指数(Vegetation Supply Water Index, VSWI),植被供水指数从植被生长和地面温度两个方面来监测干旱状况。通过对比台风“海棠”前后植被供水指数的变化,分析由于受台风影响,研究区干旱状况的变化；另一方面利用台风“海棠”前后AMSR-E土壤湿度的变化来反映台风缓解干旱。台风后研究区的平均植被供水指数由18.62增大到20.67,而平均土壤湿度由35.70%增大到52.37%。二者都增大,说明研究区的干旱状况得到了缓解。(2)台风“海棠”影响研究区期间,研究区各气象站点的实测日最高气温都大幅下降,平均下降8.6℃。由于气象站点是单点数据,点不能反映面的温度变化,而MODIS等光学影像在云雨较多的台风期间,又很容易受到云的影响,所以本研究使用不受云影响的微波AMSR-E数据。通过对比台风“海棠”前后三天的平均AMSR-E地表温度来反映台风对地表温度的影响。由于AMSR-E地表温度和实测日最高气温之间有一定差距,所以利用实测日最高气温来订正AMSR-E地表温度。台风后研究区订正后的平均AMSR-E地表温度明显下降,由34.51℃降低为29.53℃。4台风引起的灾害遥感评估首先对HJ-1星影像进行了一系列预处理,然后从台风前后台湾曾文水库水体面积的变化、台风导致台湾小林村滑坡泥石流以及台风大风影响植被生长三个方面来反映台风引起的灾害。(1)利用台风“莫拉克”前后多时相的HJ-1影像,通过去相关拉伸光谱增强方法减小波段间的相关性,再利用最大似然分类法分类,提取多时相台湾曾文水库水体信息。经过去相关拉伸的各时相影像总体分类精度达90%以上,Kappa系数也达0.9以上,精度明显高于没有进行去相关拉伸光谱增强的分类精度。台风前曾文水库面积为10.72km2,台风带来的强降雨使曾文水库水位暴涨,台风后第五天影像上提取的水库水体面积为14.81km2。监测台风对水库水体面积的影响,可以为合理地调度泄洪,保证水库下游的安全提供决策依据。(2)利用台风“莫拉克”前后多时相的HJ-1影像,通过基于去相关拉伸光谱增强的最大似然法分类,提取台风前后由于台风暴雨使小林村后山发生滑坡泥石流灾害而引起各种地类的变化。结合TRMM资料分析台风降雨情况,并辅以DEM资料,形成3D视图分析滑坡泥石流灾害。各时相影像的总体分类精度都高于90%,Kappa系数也都大于0.9。台风后小林村后山植被区域变成裸地,呈现较清晰的滑坡边界和形态特征,并在河流处形成明显的泥石流洪积扇沉积区,滑坡泥石流(包括泥石流沉积区)总面积约2.92km2,3D视图更直观的反映小林村滑坡泥石流灾害的发生。(3)由于HJ-1影像会受到云的影响,在台风“莫拉克”移动路径上选择一处多时相HJ-1影像都未被云污染的区域作为研究区,根据MODIS MCD12Q1资料分析研究区的土地覆盖情况。由于研究区内植被的归一化差值植被指数达到饱和,所以通过分析增强型植被指数的变化来反映台风大风对植被生长的影响。植被处于生长期,但台风大风吹落树叶、折断树枝,导致增强型植被指数明显下降。到台风后一个月,植被逐渐恢复,增强型植被指数有所回升。更多还原

【Abstract】 China is one of the countries in the world most affected by the typhoon, and the typhoon affecting China is mainly from Northwest Pacific. Northwest Pacific is the only sea in which typhoon formation all the year round. The number of typhoon is largest, and the distribution of typhoon is most widely. Typhoon has a wide range, and it can bring disasters such as wind, rainstorm and storm surge. On the other hand, it can bring some benefits. For example, it can relieve drought and make temperature drop substantially. With the rapid development of economic, China is more vulnerable to typhoon due to the high concentration of population and social wealth. It is significant to study the impact of typhoons on China.In the study, the spatio-temporal distribution of typhoon was analyzed based on the best track data of tropical cyclones in 1971-2008 and GIS (Geographic Information System). Then, MODIS 1B (Moderate Resolution Imaging Spectroradiometer) image was preprocessed and land surface temperature (LST) was retrieved from MODIS 1B. Next, the benefit of drought relief of typhoon was studied combining vegetation supply water index (VSWI) and AMSR-E (Advanced Microwave Scanning Radiometer-Earth Observing System) soil moisture data, and the benefit of temperature drop of typhoon was studied by daily maximum temperatures of meteorological stations and AMSR-E land surface temperature data. Finally, disasters caused by typhoon including the water body change, landslide, debris flow and vegetation disturbance were assessed by HJ-1 images. The mainly methods and conclusions are as follows:1 Spatio-temporal distribution of tropical cyclones landfalling in ChinaAccording to the best tracks of landfalling tropical cyclones in 1971-2008, landfalling frequency, intension and place were analyzed. The results showed that landfalling tropical cyclone kept high frequency in the early 1970s, and kept low frequency in the late 1990s. Tropical cyclone concentrated in the July to September, and August is as the center. Tropical cyclones with wind level above 10 acounted for 77% of the landfalling tropical cyclones. Tropical cyclones landed mainly in the southeast coast of China including Guangdong, Taiwan, Hainan, Fujian, Zhejiang and Guangxi provinces.Wind radius of historical tropical cyclone was reconstructed by studying the relationship between wind radius and maximum wind speed near the center at 6 hourly intervals of every tropical cyclone in 2004-2008, and the spatial distribution of tropical cyclone was mapped with the support of GIS. It can be seen from the spatial distribution map, the impact reduced when tropical cyclone moving inland, and the southeast coast provinces comprising Hainan, Taiwan and so on suffered the severest impact.2 Preprocess of MODIS 1B image and LST retrievalMODIS 1B is the level 1 product, and it writes a 16-bit scaled integer representation of the calibrated digital signals measured by the MODIS. Reflectance can be retrieved from SI through radiometric correction (radiometric calibration and atmospheric correction), geometric correction, bow-tie removal. Then, cloud detection and mosaic were applied for getting the data of the whole study area. LST was retrieved by split window algorithm combining visible, near infrared with thermal infrared band. Brightness temperature was calculated by radiance through Planck function. Atmospheric transmittance was gotten based on atmospheric water vapor, temperature correction function and perspective correction function. Surface emissivity was computed by vegetation coverage. Finally, LST was calculated by brightness temperature, atmospheric transmittance and surface emissivity.3 Remote sensing assessments on the drought relief and temperature drop cuased by typhoonTyphoon brings rain to relieve drought condition. As the typhoon has thick clouds, and the absorption of solar radiation reduces, temperature decreases. In the study, drought relief and temperature drop caused by typhoon were assessed using multi-source remote sensing data.(1) First, precipitation anomaly calculated on the basis of measured rainfall was adopted to analyze drought condition before typhoon Haitang. Then, vegetation supply water index (VSWI) integrating NDVI (Normalized Difference Vegetation Index) with LST was used to monitor the drought change. Impact of typhoon on drought was studied through comparing VSWI before and after typhoon Haitang. On the other hand, AMSR-E soil moisture change was also used to reflect the drought change. After typhoon Haitang, average VSWI and soil moisture increased from 18.62 to 20.67 and from 35.70% to 52.37%, respectively. It indicated that drought was relieved.(2) During the typhoon Haitang, daily maximum temperature of meteorological stations all dropped, the average is 8.6℃. Due to the daily maximum temperature of meteorological stations is point data, it can not reflect the temperature of a polygon, in addition, optical Images such as MODIS, are vulnerable to contamination by clouds, microwave data of AMSR-E were used in the study. Average AMSR-E LST before and after typhoon Haitang was compared to reflect the impact of typhoon on temperature. Moreover, because there had difference between AMSR-E LST and daily maximum temperature, AMSR-E LST was revised by daily maximum temperature. After typhoon Haitang, revised average AMSR-E LST decreased from 34.51℃to 29.53℃.4 Remote sensing assessments on the disasters caused by typhoonAfter a series of preprocess on HJ-1, disasters caused by typhoon were shown from three aspects, including change of water body, landslide and debris flow, and wind disturbance on vegetation.(1) Using multi-temporal HJ-1 images before and after typhoon Morakot, decorrelation stretch (DS) was applied to decrease the correlations among bands, and water body of the Tseng-Wen reservoir was extracted through maximum likelihood classification (MLC). The overall accuracie of each phase was all above 90%, Kappa coefficient was above 0.9, and they were better than that of direct MLC without DS spectral enhancement. The area of the water body was 10.72km2 before typhoon Morakot, and after the typhoon, the area went up to 14.81km2. It can provide a basis for decision-making in scheduling flood discharge and ensure the safety of reservoir downstream by monitoring water body impacted by typhoon.(2) Using multi-temporal HJ-1 images before and after typhoon Morakot, land types change in Xiaolin village caused by typhoon was extracted by MLC based on DS. TRMM data were adopted to know precipitation brought by typhoon, and 3D view was used to analyze landslide and debris flow with the aid of DEM. The overall accuracie of each phase was all above 90%, Kappa coefficient was above 0.9. After the typhoon, vegetation became bare land, and it appeared clear boundaries and morphological characteristics of landslide. Xiaolin village buried by landslide and debris flow, and clear alluvial fan deposition zone of debris flow was formed at the river. The whole area of landslide and debris flow (including alluvial fan deposition zone of debris flow) was about 2.92km2. It can reflect landslide and debris flow more directly from 3D view.(3) Due to HJ-1 images are vulnerable to contamination by clouds, the study area was selected in a cloudless area along the typhoon track. The NDVI is saturated in the study area, according to land cover type of MODIS MCD12Q1, the change of enhanced vegetation index (EVI) was used to reflect wind disturbance on vegetation through comparing multi-temporal HJ-1 images before and after typhoon Morakot. Vegetation was in the growing season at that time, but wind blew down leaves and broken branches, resulting in EVI decrease. After one month, vegetation gradually recovered and EVI rebounded too.更多还原