Abstract: With the rapid development of data observation and digital computing technology,the granularity of air quality data tends to be fine,which provides a data basis for visualizing data of air quality and under-standing the transformation law of air pollutants.This paper proposes a visual analysis system for inline fea-tures of fine-grained air quality data.We design a spatial-temporal data dimensional model to enhance the spa-tial-temporal expression of the data.We propose a deep learning based scheme that embeds high-dimensional air quality data into a low-dimensional feature space,which supports interactive visual analysis.We implement a multi-view linkage visualization system which assists discover the fine-grained inline relationship in the spread of air pollution and understand the data distribution characteristics of pollution transmission routes.User cases and user evaluation results show that our system is effective.