Abstract: The air quality monitoring data presents significant spatio-temporal multidimensional attributes.The traditional parallel coordinates are able to reveal multiple attributes,but show some limitations in terms of spatio-temporal analysis of the air quality monitoring data.In this paper,we design a visual analysis system aiming at the exploration of the spatio-temporal characteristics of atmospheric pollution.Firstly,the parallel coordinates are employed to display multiple attributes of a section of air quality monitoring data,which also allow users to interactively exchange the spatio-temporal perspective and specify the axis order.Some metrics are used to measure the difference of data in the parallel coordinates,including angle,area,and correlation.The matrix graph and interactive bar chart are applied to display the difference of data from different perspectives.We then weight the matrix graphs to construct a similarity matrix and use the MDS to reduce the original multidimensional data into 2D coordinates.A hierarchical clustering scheme is used to classify the coordinates.Furthermore,a design of time metaphor and an abstract of space are used to enable users achieve the spatio-temporal features in the air quality monitoring data.Convenient interactions are also integrated in this system to help users deeply investigate atmospheric pollution.A large number of experimental results as well as credible user studies further demonstrate the effectiveness and practicability of the system.