Air Pollution Monitoring and Spatial-Temporal Hotspot Pattern Analysis of Sensors Based on Sensor Grid for the Industrial Parks in Taiwan

摘要

To identify sources of pollution and predict future pollution events, the Environmental Protection Administration of Taiwan has deployed dense sensor networks in industrial districts. In face of overwhelming real-time data collected from the Internet of Things (IoT) applications for smart environmental sensing, no standard procedure based on the space-time statistical methods, such as Getis-Ord G* or Moran's I exist for defining and analyzing pollution events. We used raw data generated from microsensors as the data source, adopted spatial statistics to perform hotspot analysis, then define the event base on the result of statistical hypothesis and grid connectivity. This approach was effective in distinguishing independent pollution events when two or more events occurred concurrently in the same region. Finally, spatial and temporal descriptive statistical analysis was performed on the targeted pollution events, including the identity of pollution events through spatial-temporal hotspot analysis integrated with data visualization.