Estimation of PM2.5 Spatial Distribution Based on Kriging Interpolation

Abstract

The air quality in Beijing City is getting more and more attention. Aimed at analyzing distribution of PM2.5 concentration, this study adopts thirty-two observation sites PM2.5 concentration data in six days from Beijing City Environmental Protection Bureau. By kriging interpolation method, 100 100× unknown points’ concentrations have been predicted and PM2.5 spatial distribution drawings have been plotted. The results show that PM2.5 concentration takes the gradient distribution. South area is higher than north while west area is higher than east. Further, we discussed reasons for this distribution. Introduction Fine particles are called PM2.5, referring to the ambient air particulate matter diameter is less than or equal to 2.5 micron. The smaller the particle diameter is, the deeper the respiratory tract it goes in. It has a direct effect on the lung ventilation function. In 2012 the environmental NGO friends of nature announced air quality rankings in China's major cities, in which Beijing ranked the bottom third[1]. In the air pollution indexes, PM2.5 pollution problem is significant in Beijing. Therefore, our research is to study Beijing PM2.5 spatial distribution in March. We screen concentration information of each PM2.5 monitoring point, plot PM2.5 distribution drawings and conduct some analysis. Data and Methods Selection of Monitoring sites and Data. According to Beijing environmental protection monitoring center website [2], we get thirty-two monitoring spots information, including every spot geographical position, which is shown in Fig.1. We collect PM2.5 24 hours daily average concentration with every monitoring spot above during March 7 and March 13 in 2014. These mean values can be simple and effective to reflect a day in the situation of pollutant concentration. Methods. Since our data is from each monitoring station, PM2.5 concentration is discrete. Spot data need to be interpolated to surface data to characterize the entire region. Commonly used spatial interpolation, such as inverse distance weighted methods, trend surface method and the Tyson polygon are difficult to analyze errors. However, geostatistical interpolation with its solid theoretical foundation in spatial statistics, not only analyzes interpolation error, but also conducts the error of point-wise estimation. As one of main contents of geostatistics, kriging interpolation is a method of unbiased optimal estimation for regionalized variables, which is based on variation function theory and structural analysis [3]. So kriging interpolation is selected to plot PM2.5 distribution drawings Kriging Estimators. In our study area, there are 32 measured points. We assume that these data obey normal distribution. For any unknown point to be estimated, its estimator is expressed as a linear combination of those effective sample values. 32 1 ˆ (s ) (s ) i j j j Z Z λ = =∑ (1) International Conference on Information Sciences, Machinery, Materials and Energy (ICISMME 2015) © 2015. The authors Published by Atlantis Press 1806 Fig.1 Observation sites of PM2.5 concentration in Beijing where i s is predicted position; (s ) i Z is measured value of th i point; j λ is unknown weight of (s ) j Z . j λ depends on the spatial relationship fitting model between i j s s − distances and measured values of i s . In order to ensure the model is unbiased estimation,