PM₁₀ Concentration Monitoring Based On Aerosol Optical Depth Retrieval From MODIS Data

With rapid development of economy and course of industrialization and urbanization in our country, the environment stress of most areas increases, the pollution of the atmosphere also becomes more and more serious. In most areas, a kind of inhalable particle—PM₁₀ becomes the critical pollutant which affects the air quality. Focus on the existing reality, utilize the satellite remote sensing to monitor the space distribution and variation trend of the pollution has the most important practical significance.The article based on the research of the monitoring function of MODIS data which affects the air pollution and PM₁₀, put forward the utilization of MODIS L1B data invert the AOD (Aerosol Optical Depth). Then proceed the regression analysis by multiple methods for mean value of PM₁₀ and AOD data which appear at the same period, establish relevant relational model. The purpose of the research is expected to discover the model which can reflect the relationship of the two data exactly, and realize the utilization of MODIS data to monitor the air pollution.The main conclusions are as follows:(1) The Aerosol Optical Depth and the inhalable particulate matter (PM₁₀) concentration have better relationship which was inverted by MODIS, this can reflect the condition of air pollution in cities to a certain extent. However, only when the critical pollutant is PM10, we can take a statistic regression analysis between PM₁₀ and AOD. If in a period, the critical pollutants were not inhalable particulate matter but SO₂ or NO₂, we should eliminate the data in these days.(2) When we proceed the aerosol inversion on the MODIS data, the atmosphere model parameter in 6S model is continent pattern, the utilization of this model has been affected by climate condition. This model has limitations in some periods (such as summer), and these limitations would affect the precision of inversion. Under these conditions, we'd better analyze the components of the aerosol in order to decrease errors. (3) Constructed five types of regression model by the utilization of and AOD data which inverted by MODIS L1B data that generated between September and November in 2008. According to analyze these data which verified by coefficient of determination R² and F, we got two better regression models: linear model and quadratic model. When verified them by testing data, their average relative errors are 11.7% and 30.6%, respectively. So we can ensure that the best fitting model which can reflect the AOD and PM₁₀ is linear model.