Innovative PM2.5 monitoring data fusion technology published to the top journal of environmental science field by the team of NCU
Posted on: 2021-09-17
Figure1. Spatial distribution of PM2.5 concentration computed by heterogeneous data fusion technology, take 2016-10-16 21:00 as example. (left) Data only from EPA stations; (middle) Data only from Airbox; (right) Data fusion from EPA and Airbox. After fusing the data, the spatial distribution information can be more detailed, and unreasonable information can be avoided too.
Under the guidance of Dr. Lin, Yuan-Chien — an Associate Professor from the Department of Civil Engineering, the team including postgraduates Chi, Wan-Ju and Lin, Yong-Qing from Hydrological and Environmental Informatics Laboratory integrated the application of Internet of Things (IoT) and spatiotemporal big data analysis, and developed novel data fusion technology to provide high spatiotemporal resolution PM2.5 concentration information. The results were published in a top SCI international journal "Environment International" in the field of environmental science.
At present, there are 76 national air quality monitoring stations with high accuracy and reliability established by the Environmental Protection Agency in Taiwan. Because the stations cannot be deployed on a large scale, the low spatiotemporal resolution of these monitoring cannot meet the PM2.5 information requirements with high variability. On the other hand, the number of air quality micro-sensors including airboxes is gradually increasing, and the government is also actively deploying the Internet of Things for urban and rural air quality. However, the micro-sensors are inherently restricted and prone to large spatiotemporal variation and interference. Thus, the team adopted the best linear data fusion theory and the spatial estimation method (Kriging) to calculate the degree of daily variation of the monitoring data and the possible errors in the spatial estimation. As the weight when the data was linearly fused, which could retain the advantages of different sensors. The research results show that the spatiotemporal distribution of PM2.5 concentration can be estimated more reasonably through data fusion technology. The method can not only effectively reduce the cost of monitoring pollutants, but also improve the efficiency of monitoring air quality.
Long-term exposure to high concentrations of PM2.5 will increase the risk of various diseases and, therefore, having accurate and detailed PM2.5 monitoring is of great significance for public health and exposure assessment. The results can provide a strong scientific basis for research on allergies, various types of respiratory and cardiovascular diseases, and even cancer. This study has both academic and practical application values, which can be regarded as pioneering research on cross-domain integration in the fields of civil engineering, environmental science, information technology, and spatial information application.
The team led by Associate Professor Lin has been often recognized for outstanding performance in research on hydrology and environment-related issues, and it is also strongly supported by the "Einstein Project", a young scholar development fellowship of the Ministry of Science and Technology. At present, the team is actively cooperating with scholars in other fields, and it will continue to devote itself to the development of related academic research in the future.