【作者】 宋薇；

【导师】 张镭；

【作者基本信息】 兰州大学 ， 大气物理学与大气环境， 2007， 硕士

【摘要】 气溶胶光学厚度是气溶胶最重要的参数之一，也是表征大气浑浊度的重要物理量。近年来利用卫星遥感反演气溶胶光学厚度已取得了很多成果，在植被浓密的地区，红、蓝波段光谱的地表反射率较低，这类区域的气溶胶光学厚度反演已达到较高的精度。但是，对于干旱、半干旱下垫面以及城市下垫面等高反射率地区，气溶胶光学厚度的反演依然是个难题。主要原因是由于上述地区的地面特性复杂，地表的非均一性使得地表反射率的精确确定十分困难。本文重点研究了干旱、半干旱以及城市下垫面条件下的气溶胶反演问题。研究中采用了Terra(Aqua)／MODIS的卫星资料，通过Kaufman提出的扩展暗像元方法和R．C．Levy等人提出的V5．2算法，对兰州及周边地区的气溶胶光学厚度进行了反演。并且利用兰州大学测点的地面光度计观测结果，与反演结果进行了比较。主要的结论有：(1)地面光度计的观测结果显示，气溶胶光学厚度的大值集中在冬、春季节，进入夏季以后，污染减轻，空气质量较好，此时的气溶胶光学厚度相对比较小。(2)利用扩展暗像元方法和新的V5．2算法反演的兰州及周边区域气溶胶光学厚度分布趋势基本一致，从气溶胶光学厚度的分布图来看，在兰州市区反演得到的气溶胶光学厚度相对较大，主要原因是城市污染比较严重，兰州的河谷地形，不利于污染物的扩散。从春季部分资料的反演情况来看，一般地，光学厚度大值区位于西固区，同时兰大所处的城关区也是光学厚度较大的区域。兴隆山地区，森林覆盖率高，但是春季有些日期反演情况与实际情况不是很符合，该区域的光学厚度反演值异常偏大。原因可能是受到兴隆山山顶积雪或者云的影响，使得反演方法不太适用。对于兰州周边的县镇，部分日期的反演出现光学厚度较大值，可能与局地污染有关。(3)将利用卫星资料反演的光学厚度与光度计观测分析结果进行了比较分析。利用扩展暗像元方法得到的14天的反演结果中，反演值与观测值最为接近的是2007年3月27日，二者的相对误差为-0.03％；有8天的反演结果相对误差较小，小于20％，绝对误差的绝对值有8天小于0.100，气溶胶光学厚度反演的平均相对误差为8.49％。而利用V5．2算法所反演的11天个例中，有3天的反演结果相对误差较小，小于20％，绝对误差的绝对值有4天小于0.100，气溶胶光学厚度反演的平均相对误差为24.89％。更多还原

【Abstract】 The aerosol optical depth (AOD) is one of most important parameters of aerosol. It is also an important physical parameter to attribute the atmospheric turbidity. In recent years, it has already been made many progresses in the retrieval of AOD from satellite remote sensing. In the thick vegetation areas, the AOD retrieval has achieved high precision for the low surface reflection in red and blue wave band. But in some areas with high surface reflectance, such as arid, semiarid areas and cities, the retrieval of AOD is still a difficult problem. The main reason is, in these areas, the ground characteristic is very complex. It is extremely difficult to determine the surface index of reflection because of the heterogeneity of surface.The main point in this paper is the retrieval of AOD over land surface of arid, semiarid areas and cities. In this study, we choose the areas around Lanzhou city as the study area, and use the satellite data from Terra (Aqua)/ MODIS. The retrieval methods contain Kaufman extended dark target method and V5.2 algorithm proposed by Robert C. Levy et al. And then, the surface observations of sunphotometer are used to compare with the retrieval result.(1) From the observations of ground-based photometer, we can get the result of the high value of AOD concentrated in winter and spring. In summer, the pollution reduces and air quality becomes good, the AOD becomes relative small.(2) Comparing with the extended dark target method and V5.2 algorithm, we can find that the distribution of AOD is similar. From the distribution charts, the AOD of Lanzhou urban district is relative big. The main cause is the serious pollution in the city. The river-valley topography is disadvantageous to the diffusion of pollutant. The retrieval results in spring season show that there is a relative bigger center of AOD in Xigu district, and at the same time, the AOD of Chengguan district where Lanzhou University located is also big. However the retrieval results are not satisfied in Xinglongshan Mountains with good forest coverage in spring. A few of the retrievals are exceptionally large because the retrieval method is not available on the top of Xinglongshan Mountains with snow or cloud. Over some counties outside Lanzhou, there are high centers of the retrieval AOD corresponding the local high center of air pollution.(3) The comparison analysis between retrievals and sunphotometer observations is carried on. We can find that, in 14-day-retrieval results using the extended dark target method, the retrievals are close to the observations on 27 March 2007, with error -0.03%, and the relative errors are less than 20% in 8 days. The absolute errors are less than 0.10 in the 8 days. The average relative errors are 8.49%. Using V5.2 algorithm in 11-day-retrieval results, the relative errors are less than 20% in 3 days. The absolute errors are less than 0.10 in 4 days. The average relative errors are 24.89%.更多还原