空氣污染會提高罹患呼吸道疾病及死亡之風險，是目前全世界都關注的環境議題，其中PM2.5對人體的健康有極大的威脅而被重視，由於PM2.5的來源辨別困難，進而難以做到防治。然而在PM2.5的資料中能夠觀察到，感測器若有相似波動的數值，其背後的影響因素可能相同，因此本論文提出PM2.5的中尺度劃分研究，清楚的劃分空氣品質區能夠瞭解當地空氣品質特徵，也能幫助污染源判斷、PM2.5數值預測等其他研究。本研究使用臺灣PM2.5的開放資料，以哈爾小波轉換之分群策略，並利用動態時間變形技術提高分群準確度，不需要加入其他額外的地理環境資訊，也能夠清楚劃分中尺度空品區，分群結果符合當地環境且能做出合理解釋，能夠與該地區的生活型態、地形氣候有高相關性。

Air pollution will increase the risk of respiratory diseases and death. It is an environmental issue that is of concern to the whole world. PM2.5 is a great threat to human health and is therefore been extensively studied. Due to the difficulty of identifying the source of PM2.5, effective prevention measures are hard to realized. However, it can be observed in the data of PM2.5 that if the sensor has similar fluctuation values, the influencing factors behind it may be the same. Therefore, this paper proposes a Mesoscale Air Quality Management Areas of PM2.5, in which local characteristics of air quality is studied in each area. Understanding local air quality characteristics can also help with pollution source identification, PM2.5 numerical prediction and other research. This study uses the open data of PM2.5 in Taiwan, the clustering technique of Haar wavelet transform and dynamic time warping techniques to improve the grouping accuracy. Without adding other additional geographical environment information, it can successfully partition the mesoscale air quality management area. The grouping results are in line with the local environment and can be reasonably explained, which can be highly correlated with the life style and topographic climate of the area.

[1] 行政院環境保護署 (2017)，中華民國空氣品質監測報告105年年報。
[2] 行政院環境保護署(2018)，空氣品質指標，取自https://taqm.epa.gov.tw/taqm/tw/b0201.aspx
[3] He, Z., Kim, Y. J., Ogunjobi, K. O., & Hong, C. S. (2003). Characteristics of PM2. 5 species and long-range transport of air masses at Taean background station, South Korea. Atmospheric Environment, 37(2), 219-230.
[4] Perrone, M. R., Becagli, S., Orza, J. G., Vecchi, R., Dinoi, A., Udisti, R., & Cabello, M. (2013). The impact of long-range-transport on PM1 and PM2. 5 at a Central Mediterranean site. Atmospheric environment, 71, 176-186.
[5] Huang, D., Zhang, N., Yu, H., Zhou, H., Ma, Z., Hu, W., & Guo, J. (2015, August). Activation force-based air pollution observation station clustering. In Heterogeneous Networking for Quality, Reliability, Security and Robustness (QSHINE), 2015 11th International Conference on (pp. 115-121). IEEE.
[6] Yu, Y., Zhao, S., Wang, B., Fu, P., & He, J. (2017). Pollution characteristics revealed by size distribution properties of aerosol particles at urban and suburban sites, northwest China. Aerosol and Air Quality Research, 17(7), 1784-1797.
[7] Baker, K. R., & Foley, K. M. (2011). A nonlinear regression model estimating single source concentrations of primary and secondarily formed PM2. 5. Atmospheric Environment, 45(22), 3758-3767.
[8] Bhardwaj, R., & Pruthi, D. (2016, October). Time series and predictability analysis of air pollutants in Delhi. In Next Generation Computing Technologies (NGCT), 2016 2nd International Conference on (pp. 553-560). IEEE.
[9] Mahajan, S., Liu, H. M., Tsai, T. C., & Chen, L. J. (2018). Improving the Accuracy and Efficiency of PM2. 5 Forecast Service Using Cluster-based Hybrid Neural Network Model. IEEE Access.
[10] Lyu, B., Zhang, Y., & Hu, Y. (2017). Improving PM2. 5 Air Quality Model Forecasts in China Using a Bias-Correction Framework. Atmosphere, 8(8), 147.
[11] Liu, J., Li, W., Wu, J., & Liu, Y. (2018). Visualizing the intercity correlation of PM2. 5 time series in the Beijing-Tianjin-Hebei region using ground-based air quality monitoring data. PloS one, 13(2), e0192614.
[12] Zhou, Z., Ye, Z., Liu, Y., Liu, F., Tao, Y., & Su, W. (2017). Visual Analytics for Spatial Clusters of Air-Quality Data. IEEE computer graphics and applications, 37(5), 98-105.
[13] Du, X., & Varde, A. S. (2016, April). Mining PM2. 5 and traffic conditions for air quality. In Information and Communication Systems (ICICS), 2016 7th International Conference on (pp. 33-38). IEEE.
[14] Chang, J. H., Tseng, C. Y., Chiang, H. H., & Hwang, R. H. (2017, November). Analysis of Influential Factors in Secondary PM2. 5 by K-Medoids and Correlation Coefficient. In Cloud and Service Computing (SC2), 2017 IEEE 7th International Symposium on (pp. 177-182). IEEE.
[15] Ding, G., Zhang, B., & Zhao, P. ESDA Based Spatial Correlation Features Analysis of China Urban Regional Haze Pollution. Boletín Técnico, 55(6), 704-713.
[16] LIU, P., YAO, Y., LIANG, H., LIANG, Z., ZHANG, Y., & WANG, H. (2017). Analyzing Spatiotemporal Distribution of PM2. 5 in China by Integrating Kalman Filter and Multi-level Clustering.
[17] Huang, P., Zhang, J., Tang, Y., & Liu, L. (2015). Spatial and temporal distribution of PM2. 5 pollution in Xi’an City, China. International journal of environmental research and public health, 12(6), 6608-6625.
[18] Wu, X., Wang, Q., Liu, Y., & Li, Y. (2017, December). Data-driven regional analysis of urban atmosphere pollution based on density clustering. In Technology, Networking, Electronic and Automation Control Conference (ITNEC), 2017 IEEE 2nd Information (pp. 412-417). IEEE.
[19] Chen, L. J., Ho, Y. H., Lee, H. C., Wu, H. C., Liu, H. M., Hsieh, H. H., ... & Lung, S. C. C. (2017). An Open Framework for Participatory PM2. 5 Monitoring in Smart Cities. IEEE Access, 5, 14441-14454.
[20] Lin, Y., Chiang, Y. Y., Pan, F., Stripelis, D., Ambite, J. L., Eckel, S. P., & Habre, R. (2017, November). Mining public datasets for modeling intra-city PM2. 5 concentrations at a fine spatial resolution. In Proceedings of the 25th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (p. 25). ACM.
[21] West, M. (1997). Time series decomposition. Biometrika, 84(2), 489-494.
[22] Chaovalit, P., Gangopadhyay, A., Karabatis, G., & Chen, Z. (2011). Discrete wavelet transform-based time series analysis and mining. ACM Computing Surveys (CSUR), 43(2), 6.
[23] Mallat, S. G. (1989). A theory for multiresolution signal decomposition: the wavelet representation. IEEE transactions on pattern analysis and machine intelligence, 11(7), 674-693.
[24] Sundararajan, D. (2011). Fundamentals of the Discrete Haar Wavelet Transform.
[25] Müller, M. (2007). Dynamic time warping. Information retrieval for music and motion, 69-84.
[26] Sardá-Espinosa, A. (2017). Comparing Time-Series Clustering Algorithms in R Using the dtwclust Package.
[27] Han, J., Pei, J., & Kamber, M. (2011). Data mining: concepts and techniques (3rd ed.). Elsevier, 443-461.
[28] Jain, A. K., Murty, M. N., & Flynn, P. J. (1999). Data clustering: a review. ACM computing surveys (CSUR), 31(3), 264-323.
[29] Cai, Q., & Zhou, Y. (2016, May). A quadtree-based hierarchical clustering method for visualizing large point dataset. In Information Science and Technology (ICIST), 2016 Sixth International Conference on (pp. 372-375). IEEE.
[30] Vesanto, J., & Alhoniemi, E. (2000). Clustering of the self-organizing map. IEEE Transactions on neural networks, 11(3), 586-600.
[31] Liao, T. W. (2005). Clustering of time series data—a survey. Pattern recognition, 38(11), 1857-1874.
[32] Kassambara, A. (2017). Practical Guide to Cluster Analysis in R: Unsupervised Machine Learning (Vol. 1). STHDA.
[33] 政府資料開放平台，各鄉鎮市區人口密度，取自https://data.gov.tw/dataset/8410。
[34] 政府資料開放平台，臺灣各工業區範圍，取自https://data.gov.tw/dataset/25598。
[35] 風聆，用右手手臂記憶臺中市五大地形分佈，取自http://icecream6.pixnet.net/blog/post/26878808。
[36] 內政部國土測繪中心，國土測繪國資服務雲，取自https://maps.nlsc.gov.tw/。