桃園市為臺灣重要之工業與都市發展區域，區域內河川水與地下水除供應民生、農業及產業用水外，亦為維持水環境與生態系統之重要資源。然而，隨著都市化、工業活動及農業利用持續發展，河川水與地下水水質同時受到多重污染來源及自然環境因子影響，使污染來源辨識與管理工作更加複雜。本研究以桃園地區河川水與地下水為研究對象，利用正矩陣因子法(Positive Matrix Factorization， PMF)解析主要污染來源及其相對貢獻，並結合土地利用與空間分布分析，探討污染來源與環境背景之關聯性。 本研究蒐集2020年至2024年間桃園市河川水及2021年至2024年間地下水水質監測資料，首先進行水質統計分析，再建立PMF模式進行污染來源解析，並透過Q值、殘差分布、觀測值與預測值相關性及Bootstrap等模式診斷結果，評估模式之合理性與穩定性。最後結合監測點位周邊土地利用及空間分布特徵，綜合判釋各污染因子之環境意義。 研究結果顯示，河川水共解析出四類主要因子，包括工業製程排放、季節水文、工業金屬污染，以及生活污水／含氮有機污染等來源；地下水則解析出水岩作用、鹽化與離子濃縮、農業施肥／生活污水含氮污染，以及還原環境下地質釋出等主要因子。河川水污染來源主要受工業活動、都市發展及生活污水等人為活動影響，且具有明顯空間差異；地下水則除人為污染外，更受到含水層地質、水岩作用、地下水停留時間及氧化還原環境等水文地球化學作用控制。土地利用分析結果亦顯示，各污染因子之空間分布與工業區、都市建成區、農業用地及區域地質環境多數呈現一定程度之空間一致性，但仍受河川上下游關係與分析尺度限制。 綜合而言，河川水較能反映短期地表污染輸入與人為活動影響，而地下水則主要反映長期水文地球化學作用及污染累積特性。本研究證實PMF結合土地利用與空間分析可有效辨識桃園地區河川水與地下水之污染來源及其貢獻特性，研究成果可作為污染熱區辨識、水質監測規劃及污染防治策略擬定之參考。;Taoyuan City is one of the major industrial and urbanized regions in Taiwan. River water and groundwater in this area not only support domestic， agricultural， and industrial water demands， but also play important roles in maintaining regional water resources and aquatic environments. However， rapid urbanization， intensive industrial activities， and agricultural land use have increased the complexity of water quality issues. Both river water and groundwater may be affected by multiple pollution sources as well as natural environmental processes， making source identification and water quality management more challenging. Therefore， this study applied Positive Matrix Factorization (PMF) to identify potential pollution sources and estimate their relative contributions in river water and groundwater in Taoyuan City. Land-use information and spatial distribution analysis were further integrated to interpret the environmental significance of the identified factors. In this study， river water quality monitoring data from 2020 to 2024 and groundwater quality monitoring data from 2021 to 2024 were collected and analyzed. Basic statistical analyses were first conducted to characterize the water quality conditions of the two water bodies. PMF receptor models were then established separately for river water and groundwater. Model performance and stability were evaluated using diagnostic indicators， including Q values， residual distributions， observed-versus-predicted relationships， coefficients of determination， and bootstrap results. Finally， the PMF factor profiles， factor contributions， monitoring locations， surrounding land-use characteristics， and regional environmental settings were jointly considered for pollution source interpretation. The PMF results indicated that four major factors were resolved for river water， including industrial process discharge， seasonal hydrological conditions， urban anthropogenic activities and industrial metal pollution， and domestic sewage or nitrogenous organic pollution. For groundwater， four major factors were identified， representing water–rock interaction， salinization and ionic enrichment， agricultural fertilization or domestic sewage-related nitrogen pollution， and geogenic release under reducing conditions. The river water results showed that water quality was mainly influenced by anthropogenic activities， including industrial discharge， urban runoff， and domestic wastewater， with clear spatial differences among monitoring stations. In contrast， groundwater quality was controlled not only by anthropogenic inputs but also by hydrogeochemical processes， such as aquifer geology， water–rock interaction， groundwater residence time， and redox conditions. The spatial analysis further showed that the distributions of PMF factors corresponded reasonably with industrial areas， urbanized areas， agricultural land use， and local geological settings. Overall， river water in Taoyuan City is more sensitive to short-term surface pollution inputs and human activities， whereas groundwater more strongly reflects long-term hydrogeochemical processes and accumulated pollution effects. This study demonstrates that combining PMF with land-use and spatial analyses can effectively identify pollution sources and contribution characteristics in both river water and groundwater. The results may provide useful information for pollution hotspot identification， water quality monitoring design， and the development of pollution control strategies in Taoyuan City.
