利用克利金及來源解析結合人口分布探討健康風險之研究—以臺灣南部大型石化複合工業區VOCs致癌物種為例

Abstract

揮發性有機物(VOCs)因其對人體健康的直接與間接影響而備受關注，其中部分物種更被國際癌症研究機構(IARC)列為致癌物。因此，正確評估其健康風險顯得格外重要。然而，過去健康風險相關研究多採用單站點監測污染物濃度推估健康風險，往往忽略了更廣泛未監測區域的污染濃度與人口分布，導致風險評估結果無法全面反映實際情況。現有具空間分佈的健康風險評估多採「由污染源至環境周界」的推估方法，受限於排放清冊與模型本身物理化學反應過程參數的不確定性。有鑑於此，本研究發展一個替代性的「由環境周界溯源」風險評估架構，利用已反映複雜大氣物化過程的多站監測數據，結合成熟的正矩陣因子法(PMF)進行污染來源解析，採用一般克利金法(ordinary kriging)推估小尺度濃度分佈，再與人口疊圖，估算每年增量致癌人口數，進行空間化、來源導向且人口加權的完整健康風險評估。 本研究以南高雄大型石化複合工業區為例，包含石化、半導體及鋼鐵等多種產業，分析周圍16個監測站資料，結果顯示在監測項目中，當地有8種VOCs致癌物種具有健康風險疑慮，包含甲醛、萘、苯、乙苯、1,2-二氯乙烷、二氯甲烷、氯乙烯及乙醛，研究範圍內有97.9%人口暴露於不可接受的致癌風險(大於1×〖10〗^(-6))。估算每年因這8種 VOCs所導致的增量癌症人口數為1.79人，其中污染物以甲醛貢獻最多，污染源則以石化產業為主。此外，致癌物種濃度分佈圖與增量終生癌症風險(iLCR)分佈圖空間趨勢一致，然而與每年增量致癌人口數分佈圖則差異顯著，後者受人口密度分佈主導。石化產業在濃度與iLCR地圖貢獻約為67.1%及70.9%，顯示石化產業確實對當地居民造成風險威脅，而考量人口後，其於致癌人口地圖中貢獻雖降至41.9%，但仍為最主導的來源。反觀緊鄰人口稠密區的交通污染源對濃度與iLCR貢獻僅佔10.4%及9.2%，但在致癌人口地圖中卻上升至19.2%，顯示人口加權分析有助避免誤判污染源重要性，對公衛健康風險評估具有高度價值，若能在公衛政策上結合個體風險iLCR及總體風險CI兩大指標，可以更好的保護公眾健康。據我們目前所知，將「由環境周界溯源」觀測基礎的複合方法整合進健康風險評估流程，在現有文獻中尚缺乏相關前例，本研究可望成為發展此領域方法的初探；利用小尺度捕捉濃度與人口異質性，並進行污染溯源，提升管制策略的可參採性，同時考量人口權重對健康風險的影響，有助於政策制定者推動貼近實際且具執行效益之環境與公衛管理措施。

Volatile organic compounds (VOCs) have gained significant attention due to their direct and indirect impacts on human health, with some species classified as carcinogens by International Agency for Research on Cancer (IARC). Accurate health risk assessment (HRA) of VOCs is therefore of critical importance. However, traditional health risk studies have relied on station-based health risks, neglecting broader unsampled areas and their populations. Consequently, such assessments may fail to fully reflect real-world conditions. While existing spatially distributed HRAs commonly adopt top-down approaches, which are constrained by uncertainties in emission inventories and the physicochemical processes adopted by models. In response to these limitations, this study develops an alternative bottom-up HRA framework. This framework utilizes multiple-site VOC monitoring data, which already reflect complex atmospheric processes, and integrates several mature methods, including source apportionment using Positive Matrix Factorization (PMF), ordinary kriging for fine-scale spatial interpolation, and population overlay mapping to estimate annual incremental cancer incidence (CI). This framework was applied to a large petrochemical complex in southern Kaohsiung, Taiwan, encompassing petrochemical, semi-conductor, steel, and various other industries. Data from 16 surrounding monitoring stations were analyzed. Results indicate that eight carcinogenic VOCs, namely formaldehyde, naphthalene, benzene, ethylbenzene, 1,2-dichloroethane, methylene chloride, vinyl chloride, and acetaldehyde, posed health concerns, with 97.9% of the local population exposed to unacceptable cancer risks (greater than 1×〖10〗^(-6)). These VOCs contribute to an estimated 1.79 cancer cases per year, with formaldehyde identified as the most dominant species and the petrochemical industry as the leading source. The spatial patterns of both concentration and inhalation incremental lifetime cancer risk (iLCR) maps appeared consistent. However, these patterns differed markedly from those in the CI map, which was strongly shaped by population distribution. The petrochemical industry contributed approximately 67.1% and 70.9% to the concentration and iLCR maps, respectively, indicating a significant health threat to local residents. Although its contribution to CI dropped to 41.9% after considering population weighting, it remained the most dominant source. In contrast, traffic sources, which are concentrated in densely populated areas, exhibited a relative importance from 10.4% and 9.2% in concentration and iLCR to 19.2% in the CI map. These findings underscore the critical importance of incorporating population distribution into HRA to avoid misinterpretation and support public health policies. Moreover, adopting both individual risk indicator iLCR and population risk indicator CI into public health policy can better protect public health. To the best of our knowledge, there is a lack of precedent in the current literature for incorporating such bottom-up hybrid approach into the HRA process. This study may serve as an initial attempt to develop a framework in this field. By capturing fine-scale spatial heterogeneity in both pollutant concentrations and population distribution, conducting source apportionment, and incorporating population weighting into the risk estimation, this approach can support policymakers in developing more practical and effective environmental and public health management strategies.