Analyzing Temporal, Spatial Characteristics And Seque-nce Pattern Of Atmospheric Compound Pollution With Air Quality Data

Atmospheric compound pollution characterized by PM2.5and O3has been a bottleneck of the economic development in areas like Beijing-Tianjin-Tang-shan region, the Yangtze River Delta, Pearl River Delta and other urban agglomerations. Atmospheric compound pollution is a complex problem China has faced. PM2.5and O3are two most important city air pollutants. Other gases like CO, SO2, NO2and PM10exist in one atmosphere. As from2001to2013, the contradiction becomes acuter and wins focus nowadays, so it is the time to protect our atmospheric environment. The information management will bring us a huge amount of air quality data. New environmental management will form based on air quality data and knowledge from environmental science. Mining massive air quality data, discover hidden knowledge in complex phenomenon can provide scientific support for environmental management decisions. Scientists in and abroad analyze air quality data with traditional statistic method and data mining method, for example, principal component analysis, Spearman rank correlation coefficient, linear correlation ect. And data mining methods are Artificial Neural Networks, Hierarchical Bayesian and Fractal Analysis for classification and prediction and K-means, FLR for cluster analysis. Especailly, association analysis techniques, also a data mining method, can also be applied in the analysis and prediction of environmental information. Based on real air quality monitoring data at945stations in190cities and inspired by principle of association rules and sequence pattern mining in computer science, this paper builds a method to analyze the status of atmospheric compound pollution and its changing process, reveals relation between air pollutions, its spatial, temporal characteristics and sequence patterns of atmospheric pollutants during pollution process, which is helpful to decision in making environmental management and a new way to make full use of the massive air quality data.In this paper, the following work have done:1. Collecting city air quality hour data around the country between September4,2013and December4,2013.Attribute data like spatial data and time data about monitoring stations are added to the main database, and are preprocessed as the data mining technology required.2. Establishing data mining task frames to analyzing temporal, spatial characteristics and process of atmospheric air pollution effectively.3. Writing program in R Language to call the function packages, to compute, to visualize the result rules and sequences, to distinguish strong rules and sequences.The main results are as follows:1. In China, PM2.5of high concentration is often accompanied by multiple pollutants. When the pollution levels of CO、NO2、SO2reach second, that of PM10reaches six, the probability that PM2.5reaches Ⅵ will be93.0%. But when there are only one kind of pollutant exists in the air, the highest probability that PM2.5reaches Ⅵ will only be65.7%.2. In Northeast China, when PM10reaches Ⅵ and NO2reaches Ⅱ or only NO2(Ⅲ) exists in the air, there is high probability that PM2.5reaches Ⅵ. In Beijing-Tianjin-Tangshan region, CO, NO2, SO2, PM10, PM2.5mixed pollution situation has high confidence, around90%. In North, there is strong correlation between SO2and PM2.5, NO2and PM2.5, CO and PM2.5. In Sorth there is strong correlation between PM10and PM2.5.3.14:00-16:00,23:00-6:00are two featured periods with high confidence (bigger than0.9). At the average of country, the ease degree that PM2.5produces in seven time period list like: T=7>T=6>T=5>T=4>T=1>T=3>T=2. It means that in9:00-11:00and12:00-13:00, air is quite clean, in7:00-8:00and14:00-16:00, air is lightly polluted, in17:00-19:00,20:00-22:00and23:00-6:00, air is heavily polluted.There always be more than one pollutant species in the polluted air, like CO、SO2、NO2、PM10. And there only one or two pollutant species in the mor clean air.4. The required time that PM2.5concentration remained stable, progressively higher and progressively lower varies with pollution degree. The longest time that PM2.5remained stable is24hours. As pollution get severer, it changes from12-24hours to3hours to lose stable and progressively heavier. It needs about one hour to change PM2.5Ⅵ to PM2.5V, while other changes all need12-24hours. PM2.5has strong sequence patterns with PM10, CO, NO2and SO2.5. During research period, O3(Ⅱ) maintains longer than O3(Ⅲ), and the concentration tends to stabilize rather than aggravate or mitigate.