摘要京津冀是重工业最密集也是烧煤最多的地区。全国雾霾最严重的的 10 个城 市中有一半都位于京津冀地区。全国曾遭受多次严重雾霾，其中以京津冀地区最 为严重，引起了社会上的广泛讨论。对于城市大气污染，快速而高效地对空气质 量进行实时监测、探究其时空特征和演变规律成为首要解决的问题。69981

本文以北京地区为建模区域，首先 MODIS 气溶胶产品进行了真实性检验， 确定了该产品的精度(R=0.910)。再提取其中的气溶胶光学厚度值与地面实测细 的颗粒物质量浓度进行相关性分析，建立了各个季度的细颗粒物质量浓度与气溶 胶光学厚度之间的最佳拟合模型(R=0.624~0.833)，并结合地面气象数据，进行了 垂直订正和湿度修正，提高了各个季度模型的精度(R=0.742~0.887)。在此基础上 根据各个季度最佳拟合模型反演了京津冀地区的 PM2.5 浓度。并利用北京地区 的地面实测数据对各个季度的 PM2.5 质量浓度的反演精度，进行了验证，结果 发现 3 月和 11 月由于数据量少的缘故出现了较大的偏差，其余月份反演的精度 基本可靠，结果还发现利用冬季估算模型对 11 月 PM2.5 质量浓度进行估算相比 于秋季模型具有更高的精度。

本论文有图 15 幅，表 6 个，参考文献 38 篇。

毕业论文关键词：细颗粒物质量浓度 中等分辨率光谱仪 气溶胶光学厚度 统计建模

Study on PM2.5 concentration estimation from Satellite Remote Sensing Data

Abstract Beijing, Tianjin and Hebei is the most intensive heavy industry is the most burned areas of coal. Half of the country's 10 cities in the country's most serious haze are located in Beijing, Tianjin and Hebei region. The country has suffered several severe haze, which is the most serious in Beijing, Tianjin and Hebei t region has aroused extensive discussion. For urban air pollution, it is the most important problem to timely monitor the air quality in real time, and to explore the temporal and spatial characteristics and the evolution law of air pollution.

In this paper, the Beijing area as the modeling area, the first MODIS aerosol products to carry out the authenticity test, to determine the accuracy of the product (R=0.910). To extract the aerosol optical thickness values and ground measured fine particulate mass concentration correlation analysis, established each quarter of the fine particle mass concentration and aerosol optical thickness between the best fitting model R=0.624~0.833, combined with ground meteorological data according to vertical correction and humidity correction, improves the quarter of each model accuracy R=0.742~0.887. On this basis, the PM2.5 concentration in Beijing, Tianjin and Hebei region was retrieved based on the best fitting model in each quarter. And the Beijing area of ground truth data for each quarter of PM2.5 mass concentration inversion precision, simple verification. It was found in March and November because of less amount of data the large deviation, in the rest of the inversion accuracy is reliable. The results also now use winter estimation model on November PM2.5 mass concentrations were estimated compared to fall model has a higher accuracy.

This paper has 15 figures, 6 tables, 38 references.

Key Words: PM2.5 Concentration MODIS AOD Statistical Modeling

目 录

摘  要 . I 

Abstract . II 

图清单 . V 

表清单 . V 

变量注释表 .. VI 

1 绪论 . 1 

1.1 研究背景和意义