臭氧(Ozone, O3)暴露會增加人類罹患心血管和呼吸系統疾病的風險。基於此政府立法管制室外O3的濃度，然而近年來全台O3的平均濃度，不降反升，揭示O3的管制仍為當今政府迫切需重視的空污議題。目前了解O3濃度變化的方式，主要是透過環保署空氣品質監測站所量測的濃度資料，然而，有限的定點式監測站，難以反映O3於全台大範圍時空上的變化。除此之外，O3主要是由工廠、汽機車所排放的前驅物，在陽光的作用下發生光化反應而來，然而，O3前驅物的濃度與陽光於一天之中並非定值，使得日夜的O3濃度有明顯的差異。綜上所述，有效且精準地評估O3濃度於時空上的變化，尤其在日夜之間的變化，是為當今重要的挑戰。近年來科技日新月異，機器學習演算法發展日趨成熟，因此本研究室發展出空間混合集成學習模型(Ensemble mixed spatial model, EMSM)，其為一個能截長補短，整合傳統和現今推估空氣污染物濃度各方法之優點，加上考量空間資訊的新穎集成學習模型，因此適合用來作為推估O3濃度於時空變化的工具。基於此，本研究的目的為透過空間混合集成學習模型來推估O3濃度於白天、夜間和全天的時空變化。
本研究使用2000至2019年環保署74個測站的白天、夜間及全天平均空污監測資料以及影響O3濃度的相關變數，以建置O3濃度時空推估模型。首先使用空間內插法處理影響O3濃度的相關變數，接著透過土地利用迴歸模型選出重要變數，進一步基於五種機器學習演算法法建模，這些演算法包含隨機森林(Random forest, RF)、梯度提升(Gradient boosting, GB)、輕量梯度提升(Light gradient boosting, LGBM)、分類特徵提升(Categorical boosting, CatBoost)與極限梯度提升(Extreme gradient boosting, XGB)；其次選擇其中解釋能力最佳且穩定的前三種演算法之推估值，再次進行集成學習建模，本研究將此模型稱之為空間混合集成學習模型；最後比較不同方法建置模型的差異。
結果顯示，空間混合集成學習模型在所有模型中具有最好的解釋能力，其在白天、夜間及全天模型Adjusted R2分別為0.89、0.88及0.88，並且從內部、外部資料驗證及極端高值驗證中，可得知此模型沒有過度擬合的問題，且推估能力在不同時間及地點都穩定可靠，於極端高值時仍有一定的解釋力。觀察模型推估之長期(2000年至2019年)高空間解析度(50m×50m)每日白天、夜間與全天O3分布圖可知，白天O3濃度大於全天及夜間，春季與秋季較高，而夏季與冬季較低；在任何時段濃度較高地區主要在郊區或山區，濃度低地區以市區為主，這是因為O3生成機制複雜，因此與其他污染物呈現不同的分布，而此模型也可以良好的應用於O3高濃度污染事件、COVID-19 流行期間的變化與O3過去二十年的長期趨勢。本研究結果證明，空間混合集成學習模型可以整合空間內插法、土地利用迴歸法、機器學習演算法與集成學習法之優勢，以準確推估台灣本島白天、夜間與全天O3濃度的空間變異。

Ozone (O3) is a secondary pollutant produced from a photochemical reaction involving the precursors like VOCs and NOx in sunlight. Exposure to O3 has been associated with a wide range of adverse health impacts, including cardiovascular and respiratory diseases. Higher O3 concentration usually occurs in the daytime, and conversely, O3 concentration levels are lowest at night. With the growing concerns of a government agency on air pollution reduction, among the pollutants, only O3 was elevating in the last two decades. Thus, it is necessary to simulate the long-term spatiotemporal pattern of surface O3, which is critical for epidemiological studies. Due to the limited number and uneven distribution of air quality stations, it is difficult to represent the air exposure of local people. Therefore, we need a large-scale model to estimate the temporal and spatial variation of O3 concentration.
O3 concentration data were collected from 74 fixed air quality monitoring stations from the Taiwan EPA, including daytime, nighttime, and daily from 2000-2019. Also contained related variables that affect O3 concentration, to build a spatiotemporal estimation model for O3 concentration. First, spatial interpolation was used to deal with the relevant variables affecting O3 concentration, and then important variables were selected through the land use regression model. The modeling is further based on five machine learning algorithms, which include Random Forest (RF), Gradient Boosting (GB), Light Gradient Boosting (LGBM), Categorical boosting (CatBoost), and extreme gradient boosting (XGB). Secondly, the estimates of the top three algorithms with the best explanatory ability and stability are selected, and the ensemble learning modeling is performed again. This study calls this model Ensemble mixed spatial model (EMSM). Finally, the differences between models constructed by different methods are compared.
The values of R2 increased from 0.30 to 0.90 in LUR and EMSM, the improvement is up to 60％, which also confirms that EMSM has better estimation ability than traditional or single machine learning models . Then, in the validation of the internal, external, and extremely high values, it is shown that the explanatory power of the validated model is similar to that of the main model, which means that the model is stable and there is no problem with overfitting. The results of this study demonstrated that the ensemble mixed spatial model can integrate the advantages of spatial interpolation, land use regression, machine learning algorithms, and ensemble learning to accurately estimate the spatial variability during the daytime, nighttime, and daily O3 concentrations in Taiwan.

中華民國交通部. (2022). 機動車輛登記數. 中華民國交通部. https://stat.motc.gov.tw/mocdb/stmain.jsp?sys=100&funid=a3301
台灣環保署. (2022). 臭氧及氮氧化物濃度日變化. 台灣環保署. https://airtw.epa.gov.tw/cht/Encyclopedia/pedia09/pedia9_2.aspx
張艮輝, 簡慧貞, & 呂鴻光. (2002). 臭氧污染控制物種與空氣污染防制策略之分析. 環境保護, 25(2), 95-116.
溫育勇. (2002). NO2 干擾環形氣固分離器之氣相亞硝酸採樣的誤差
葉士鳴. (2002). 大氣中懸浮微粒含碳成分之分佈與來源
Abraham, J. S., & Comrie, A. C. (2004). Real-time ozone mapping using a regression-interpolation hybrid approach, applied to Tucson, Arizona. J Air Waste Manag Assoc, 54(8), 914-925. https://doi.org/10.1080/10473289.2004.10470960
Adam-Poupart, A., Brand, A., Fournier, M., Jerrett, M., & Smargiassi, A. (2014). Spatiotemporal modeling of ozone levels in Quebec (Canada): a comparison of kriging, land-use regression (LUR), and combined Bayesian maximum entropy-LUR approaches. Environ Health Perspect, 122(9), 970-976. https://doi.org/10.1289/ehp.1306566
Ainsworth, E. A., Yendrek, C. R., Sitch, S., Collins, W. J., & Emberson, L. D. (2012). The effects of tropospheric ozone on net primary productivity and implications for climate change. Annu Rev Plant Biol, 63, 637-661. https://doi.org/10.1146/annurev-arplant-042110-103829
Beelen, R., Hoek, G., Pebesma, E., Vienneau, D., de Hoogh, K., & Briggs, D. J. (2009). Mapping of background air pollution at a fine spatial scale across the European Union. Sci Total Environ, 407(6), 1852-1867. https://doi.org/10.1016/j.scitotenv.2008.11.048
Beelen, R., Hoek, G., Vienneau, D., Eeftens, M., Dimakopoulou, K., Pedeli, X., Tsai, M.-Y., Künzli, N., Schikowski, T., & Marcon, A. (2013). Development of NO2 and NOx land use regression models for estimating air pollution exposure in 36 study areas in Europe–The ESCAPE project. Atmospheric environment, 72, 10-23.
Bell, M. L., McDermott, A., Zeger, S. L., Samet, J. M., & Dominici, F. (2004). Ozone and short-term mortality in 95 US urban communities, 1987-2000. Jama, 292(19), 2372-2378. https://doi.org/10.1001/jama.292.19.2372
Breton, C. V., Wang, X., Mack, W. J., Berhane, K., Lopez, M., Islam, T. S., Feng, M., Lurmann, F., McConnell, R., Hodis, H. N., Künzli, N., & Avol, E. (2012). Childhood air pollutant exposure and carotid artery intima-media thickness in young adults. Circulation, 126(13), 1614-1620. https://doi.org/10.1161/circulationaha.112.096164
Briggs, D. J., Collins, S., Elliott, P., Fischer, P., Kingham, S., Lebret, E., Pryl, K., Van Reeuwijk, H., Smallbone, K., & Van Der Veen, A. (1997). Mapping urban air pollution using GIS: a regression-based approach. International Journal of Geographical Information Science, 11(7), 699-718.
Brokamp, C., Jandarov, R., Hossain, M., & Ryan, P. (2018). Predicting daily urban fine particulate matter concentrations using a random forest model. Environmental science & technology, 52(7), 4173-4179.
Chen, K., Peng, Y., Lu, S., Lin, B., & Li, X. (2021). Bagging based ensemble learning approaches for modeling the emission of PCDD/Fs from municipal solid waste incinerators. Chemosphere, 274, 129802. https://doi.org/10.1016/j.chemosphere.2021.129802
Chen, S. P., Liu, W. T., Hsieh, H. C., & Wang, J. L. (2021). Taiwan ozone trend in response to reduced domestic precursors and perennial transboundary influence. Environ Pollut, 289, 117883. https://doi.org/10.1016/j.envpol.2021.117883
Chen, T. H., Hsu, Y. C., Zeng, Y. T., Candice Lung, S. C., Su, H. J., Chao, H. J., & Wu, C. D. (2020). A hybrid kriging/land-use regression model with Asian culture-specific sources to assess NO(2) spatial-temporal variations. Environ Pollut, 259, 113875. https://doi.org/10.1016/j.envpol.2019.113875
Di, Q., Rowland, S., Koutrakis, P., & Schwartz, J. (2017). A hybrid model for spatially and temporally resolved ozone exposures in the continental United States. Journal of the Air & Waste Management Association, 67(1), 39-52.
Eeftens, M., Beelen, R., de Hoogh, K., Bellander, T., Cesaroni, G., Cirach, M., Declercq, C., Dėdelė, A., Dons, E., de Nazelle, A., Dimakopoulou, K., Eriksen, K., Falq, G., Fischer, P., Galassi, C., Gražulevičienė, R., Heinrich, J., Hoffmann, B., Jerrett, M., . . . Hoek, G. (2012). Development of Land Use Regression models for PM(2.5), PM(2.5) absorbance, PM(10) and PM(coarse) in 20 European study areas; results of the ESCAPE project. Environ Sci Technol, 46(20), 11195-11205. https://doi.org/10.1021/es301948k
Gilbert, N. L., Goldberg, M. S., Beckerman, B., Brook, J. R., & Jerrett, M. (2005). Assessing spatial variability of ambient nitrogen dioxide in Montreal, Canada, with a land-use regression model. Journal of the Air & Waste Management Association, 55(8), 1059-1063.
Hoek, G., Beelen, R., De Hoogh, K., Vienneau, D., Gulliver, J., Fischer, P., & Briggs, D. (2008). A review of land-use regression models to assess spatial variation of outdoor air pollution. Atmospheric environment, 42(33), 7561-7578.
Holle, R., Happich, M., Löwel, H., & Wichmann, H. E. (2005). KORA--a research platform for population based health research. Gesundheitswesen, 67 Suppl 1, S19-25. https://doi.org/10.1055/s-2005-858235
Hsu, C. Y., Xie, H. X., Wong, P. Y., Chen, Y. C., Chen, P. C., & Wu, C. D. (2022). A mixed spatial prediction model in estimating spatiotemporal variations in benzene concentrations in Taiwan. Chemosphere, 301, 134758. https://doi.org/10.1016/j.chemosphere.2022.134758
Jerrett, M., Burnett, R. T., Pope III, C. A., Ito, K., Thurston, G., Krewski, D., Shi, Y., Calle, E., & Thun, M. (2009). Long-term ozone exposure and mortality. New England Journal of Medicine, 360(11), 1085-1095.
Ke, H., Gong, S., He, J., Zhang, L., Cui, B., Wang, Y., Mo, J., Zhou, Y., & Zhang, H. (2022). Development and application of an automated air quality forecasting system based on machine learning. Sci Total Environ, 806(Pt 3), 151204. https://doi.org/10.1016/j.scitotenv.2021.151204
Kerckhoffs, J., Wang, M., Meliefste, K., Malmqvist, E., Fischer, P., Janssen, N. A., Beelen, R., & Hoek, G. (2015). A national fine spatial scale land-use regression model for ozone. Environ Res, 140, 440-448. https://doi.org/10.1016/j.envres.2015.04.014
Lee, M., Lin, L., Chen, C. Y., Tsao, Y., Yao, T. H., Fei, M. H., & Fang, S. H. (2020). Forecasting Air Quality in Taiwan by Using Machine Learning. Sci Rep, 10(1), 4153. https://doi.org/10.1038/s41598-020-61151-7
Lei, W., de Foy, B., Zavala, M., Volkamer, R., & Molina, L. (2007). Characterizing ozone production in the Mexico City Metropolitan Area: a case study using a chemical transport model. Atmospheric Chemistry and Physics, 7(5), 1347-1366.
Lim, C. C., Hayes, R. B., Ahn, J., Shao, Y., Silverman, D. T., Jones, R. R., Garcia, C., Bell, M. L., & Thurston, G. D. (2019). Long-Term Exposure to Ozone and Cause-Specific Mortality Risk in the United States. Am J Respir Crit Care Med, 200(8), 1022-1031. https://doi.org/10.1164/rccm.201806-1161OC
Loibi, W., Winiwarter, W., Kopsca, A., Zufger, J., & Baumann, R. (1994). Estimating the spatial distribution of ozone concentrations in complex terrain. Atmospheric environment, 28(16), 2557-2566.
Lundberg, S. M., & Lee, S.-I. (2017). A unified approach to interpreting model predictions. Advances in neural information processing systems, 30.
Ma, J., Zhang, R., Xu, J., & Yu, Z. (2022). MERRA-2 PM(2.5) mass concentration reconstruction in China mainland based on LightGBM machine learning. Sci Total Environ, 827, 154363. https://doi.org/10.1016/j.scitotenv.2022.154363
Ma, R., Ban, J., Wang, Q., & Li, T. (2020). Statistical spatial-temporal modeling of ambient ozone exposure for environmental epidemiology studies: A review. Sci Total Environ, 701, 134463. https://doi.org/10.1016/j.scitotenv.2019.134463
Maji, K. J., Ye, W. F., Arora, M., & Nagendra, S. M. S. (2019). Ozone pollution in Chinese cities: Assessment of seasonal variation, health effects and economic burden. Environ Pollut, 247, 792-801. https://doi.org/10.1016/j.envpol.2019.01.049
Miller, T. H., Gallidabino, M. D., MacRae, J. I., Hogstrand, C., Bury, N. R., Barron, L. P., Snape, J. R., & Owen, S. F. (2018). Machine Learning for Environmental Toxicology: A Call for Integration and Innovation. Environ Sci Technol, 52(22), 12953-12955. https://doi.org/10.1021/acs.est.8b05382
Mills, G., Pleijel, H., Malley, C. S., Sinha, B., Cooper, O. R., Schultz, M. G., Neufeld, H. S., Simpson, D., Sharps, K., & Feng, Z. (2018). Tropospheric Ozone Assessment Report: Present-day tropospheric ozone distribution and trends relevant to vegetation. Elementa: Science of the Anthropocene, 6.
Rai, R., & Agrawal, M. (2012). Impact of tropospheric ozone on crop plants. Proceedings of the National Academy of Sciences, India Section B: Biological Sciences, 82(2), 241-257.
Ren, X., Mi, Z., & Georgopoulos, P. G. (2020). Comparison of Machine Learning and Land Use Regression for fine scale spatiotemporal estimation of ambient air pollution: Modeling ozone concentrations across the contiguous United States. Environ Int, 142, 105827. https://doi.org/10.1016/j.envint.2020.105827
Requia, W. J., Di, Q., Silvern, R., Kelly, J. T., Koutrakis, P., Mickley, L. J., Sulprizio, M. P., Amini, H., Shi, L., & Schwartz, J. (2020). An Ensemble Learning Approach for Estimating High Spatiotemporal Resolution of Ground-Level Ozone in the Contiguous United States. Environ Sci Technol, 54(18), 11037-11047. https://doi.org/10.1021/acs.est.0c01791
Ross, Z., Jerrett, M., Ito, K., Tempalski, B., & Thurston, G. D. (2007). A land use regression for predicting fine particulate matter concentrations in the New York City region. Atmospheric environment, 41(11), 2255-2269.
Sanderson, M. G., Jones, C. D., Collins, W. J., Johnson, C. E., & Derwent, R. G. (2003). Effect of Climate Change on Isoprene Emissions and Surface Ozone Levels. Geophysical Research Letters, 30(18). https://doi.org/https://doi.org/10.1029/2003GL017642
Shahriar, S. A., Kayes, I., Hasan, K., Hasan, M., Islam, R., Awang, N. R., Hamzah, Z., Rak, A. E., & Salam, M. A. (2021). Potential of Arima-ann, Arima-SVM, dt and catboost for atmospheric PM2. 5 forecasting in bangladesh. Atmosphere, 12(1), 100.
Shang, Y., Sun, Z., Cao, J., Wang, X., Zhong, L., Bi, X., Li, H., Liu, W., Zhu, T., & Huang, W. (2013). Systematic review of Chinese studies of short-term exposure to air pollution and daily mortality. Environ Int, 54, 100-111. https://doi.org/10.1016/j.envint.2013.01.010
Shi, X., & Brasseur, G. P. (2020). The Response in Air Quality to the Reduction of Chinese Economic Activities During the COVID-19 Outbreak. Geophysical Research Letters, 47(11), e2020GL088070. https://doi.org/https://doi.org/10.1029/2020GL088070
Son, Y., Osornio-Vargas Á, R., O'Neill, M. S., Hystad, P., Texcalac-Sangrador, J. L., Ohman-Strickland, P., Meng, Q., & Schwander, S. (2018). Land use regression models to assess air pollution exposure in Mexico City using finer spatial and temporal input parameters. Sci Total Environ, 639, 40-48. https://doi.org/10.1016/j.scitotenv.2018.05.144
Tan, Z., Lu, K., Dong, H., Hu, M., Li, X., Liu, Y., Lu, S., Shao, M., Su, R., & Wang, H. (2018). Explicit diagnosis of the local ozone production rate and the ozone-NOx-VOC sensitivities. Science bulletin, 63(16), 1067-1076.
Tranchant, B., & Vincent, A. (2000). Statistical interpolation of ozone measurements from satellite data (TOMS, SBUV and SAGE II) using the kriging method. Annales Geophysicae,
U.S. Environmental Protection Agency. (2022). Ozone source. U.S. Environmental Protection Agency. https://www.epa.gov/ground-level-ozone-pollution
Wang, M., Keller, J. P., Adar, S. D., Kim, S. Y., Larson, T. V., Olives, C., Sampson, P. D., Sheppard, L., Szpiro, A. A., Vedal, S., & Kaufman, J. D. (2015). Development of Long-term Spatiotemporal Models for Ambient Ozone in Six Metropolitan regions of the United States: The MESA Air Study. Atmos Environ (1994), 123(A), 79-87. https://doi.org/10.1016/j.atmosenv.2015.10.042
Watson, G. L., Telesca, D., Reid, C. E., Pfister, G. G., & Jerrett, M. (2019). Machine learning models accurately predict ozone exposure during wildfire events. Environ Pollut, 254(Pt A), 112792. https://doi.org/10.1016/j.envpol.2019.06.088
Wei, J., Li, Z., Li, K., Dickerson, R. R., Pinker, R. T., Wang, J., Liu, X., Sun, L., Xue, W., & Cribb, M. (2022). Full-coverage mapping and spatiotemporal variations of ground-level ozone (O3) pollution from 2013 to 2020 across China. Remote Sensing of Environment, 270, 112775.
Wolpert, D. H. (1992). Stacked generalization. Neural networks, 5(2), 241-259.
Wong, P. Y., Lee, H. Y., Chen, Y. C., Zeng, Y. T., Chern, Y. R., Chen, N. T., Candice Lung, S. C., Su, H. J., & Wu, C. D. (2021). Using a land use regression model with machine learning to estimate ground level PM(2.5). Environ Pollut, 277, 116846. https://doi.org/10.1016/j.envpol.2021.116846
Wu, C. D., Chen, Y. C., Pan, W. C., Zeng, Y. T., Chen, M. J., Guo, Y. L., & Lung, S. C. (2017). Land-use regression with long-term satellite-based greenness index and culture-specific sources to model PM(2.5) spatial-temporal variability. Environ Pollut, 224, 148-157. https://doi.org/10.1016/j.envpol.2017.01.074
Wu, C. D., Zeng, Y. T., & Lung, S. C. (2018). A hybrid kriging/land-use regression model to assess PM(2.5) spatial-temporal variability. Sci Total Environ, 645, 1456-1464. https://doi.org/10.1016/j.scitotenv.2018.07.073
Xu, D., Yuan, Z., Wang, M., Zhao, K., Liu, X., Duan, Y., Fu, Q., Wang, Q., Jing, S., Wang, H., & Zhao, X. (2022). Multi-factor reconciliation of discrepancies in ozone-precursor sensitivity retrieved from observation- and emission-based models. Environ Int, 158, 106952. https://doi.org/10.1016/j.envint.2021.106952
Yin, P., Chen, R., Wang, L., Meng, X., Liu, C., Niu, Y., Lin, Z., Liu, Y., Liu, J., Qi, J., You, J., Zhou, M., & Kan, H. (2017). Ambient Ozone Pollution and Daily Mortality: A Nationwide Study in 272 Chinese Cities. Environ Health Perspect, 125(11), 117006. https://doi.org/10.1289/ehp1849
Young, M. T., Bechle, M. J., Sampson, P. D., Szpiro, A. A., Marshall, J. D., Sheppard, L., & Kaufman, J. D. (2016). Satellite-Based NO2 and Model Validation in a National Prediction Model Based on Universal Kriging and Land-Use Regression. Environ Sci Technol, 50(7), 3686-3694. https://doi.org/10.1021/acs.est.5b05099
Zhan, Y., Luo, Y., Deng, X., Chen, H., Grieneisen, M. L., Shen, X., Zhu, L., & Zhang, M. (2017). Spatiotemporal prediction of continuous daily PM2. 5 concentrations across China using a spatially explicit machine learning algorithm. Atmospheric environment, 155, 129-139.
Zhan, Y., Luo, Y., Deng, X., Grieneisen, M. L., Zhang, M., & Di, B. (2018). Spatiotemporal prediction of daily ambient ozone levels across China using random forest for human exposure assessment. Environ Pollut, 233, 464-473. https://doi.org/10.1016/j.envpol.2017.10.029