基于MODIS影像空气质量评价及中国区域云量分析研究

发布时间：2018-02-10 13:11

本文关键词： MODIS图像 AOT PM2.5 云量云检测　出处：《安徽师范大学》2015年硕士论文　论文类型：学位论文

【摘要】：对地观测卫星可以对地球进行长时间大范围的监测,为人们提供大量重要的观测资料。利用卫星数据研究云主要从两个方面考虑,一方面,云是遥感应用领域中一个主要干扰因素,影响对地表的监测;另一方面,云在地气系统中起到重要的调节作用,为精确预测气候变化,构建数值预测模型时需要大量的云参数资料。本论文的研究内容有:首先,利用去除云像元的MODIS L1B影像反演北京市上空的气溶胶光学厚度,并与地面监测PM2.5质量浓度数据进行回归拟合出四个经验预测模型,从而完成对城市空气质量评价模型的构建这一遥感应用;其次,通过MODIS云检测产品反演高分辨率的云量资料,并与地面观测云量进行对比分析,从而一方面为气候变化研究提供高分辨率的云量资料,另一方面对云检测产品精度进行分析评价,为改进云检测算法模型提供方向;最后对MODIS云检测方法进行探究。本论文结论主要有以下几点:1.构建的PM2.5质量浓度的四个经验模型的精度分别为R2=0.818,R2=0.750,R2=0.699和R2=0.629。其中,二次模型效果最好,能够提供快速而较经济的PM2.5空间分布信息。2.利用四个模型和2012年10月11日的MODIS影像反演了PM2.5质量浓度,并与地面监测的PM2.5进行对比,有50%,46.4%,46.4%和39.3%的站点分别在四个预测模型的误差范围内。3.利用MODIS云检测产品反演了中国区域近十年的上午星和下午星的高分辨率云量,统计分析其时间序列发现:总云量变化趋势为略有下降,并且下午星云量比上午星要多,与地面观测的日均云量的相关性较好,例如2012年的相关系数为0.878。4.对比卫星反演和地面观测的月平均云量发现:在寒季两者相差较大,可能是植被覆盖较少或冰雪覆盖导致地表反射率较大,被误判为云。5.经过在6个不同下垫面区域的统计分析实验表明:去除寒季数据后,中国北方的几个研究区域的两种云量的相关性增加明显,特别是东北森林区域,而在南方研究区域没有明显变化,可能是积雪造成云的误判。6.通过云检测测试,发现利用亮温差9.37.3BTBT?检测能够有效抑制将荒漠或植被稀疏的亮地表区域误判为云。
[Abstract]:Earth observation satellites can monitor the Earth for a long time and on a large scale, providing people with a large amount of important observation data. The use of satellite data to study clouds is mainly considered from two aspects: on the one hand, Clouds are a major interference factor in remote sensing applications, affecting the monitoring of the surface. On the other hand, clouds play an important role in regulating the earth and atmosphere systems to accurately predict climate change. A large amount of cloud parameter data are needed to construct a numerical prediction model. Firstly, the aerosol optical thickness over Beijing is retrieved by using MODIS L1B image with cloud pixel removal. Four empirical prediction models were fitted by regression with ground monitoring PM2.5 mass concentration data, thus the remote sensing application of urban air quality evaluation model was completed. Secondly, high resolution cloud data were retrieved through MODIS cloud detection products. And compared with the cloud amount observed on the ground, on the one hand to provide high-resolution cloud data for the study of climate change, on the other hand, to analyze and evaluate the accuracy of cloud detection products, so as to provide a direction for improving the cloud detection algorithm model. Finally, the MODIS cloud detection method is explored. The main conclusions of this paper are as follows: 1. The accuracy of four empirical models of PM2.5 mass concentration are R2O0.818R2O0.750R2O0.699 and R2O0.629. among them, the quadratic model has the best effect. It can provide fast and economical spatial distribution information of PM2.5. Using four models and MODIS image of October 11th 2012, the mass concentration of PM2.5 is retrieved and compared with PM2.5 monitored on the ground. 46.4% and 39.3% of the stations are within the error range of four prediction models, respectively. Using the MODIS cloud detection products, the high resolution cloud amounts of the morning and afternoon stars in the last ten years in the Chinese region have been inversed. The statistical analysis of the time series shows that the change trend of the total cloud amount is a slight decrease, and the number of nebula in the afternoon is more than that of the morning star, and the correlation with the daily average cloud amount observed on the ground is better. For example, in 2012, the correlation coefficient was 0.878. 4. By comparing the monthly average cloud cover of satellite inversion and ground observation, it was found that in cold season, the difference between the two is large, which may be caused by less vegetation cover or greater surface reflectivity due to snow and ice cover. The results of statistical analysis in six different underlying areas show that after removing cold season data, the correlation between the two types of cloud cover in several study areas in northern China has increased significantly, especially in the northeast forest region. However, there is no obvious change in the southern study area, which may be caused by snow. 6. Through the cloud detection test, it is found that 9.37.3 BTT BTT can be used as a result of bright temperature difference. The detection can effectively restrain the misjudgment of the desert or vegetation sparse bright surface area into cloud.
【学位授予单位】：安徽师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：X823;P426.5

【相似文献】