本研究以環保署設置於新北市之一般測站為基礎，統計2001-2010 年之空氣品質與氣象監測數據，利用不同統計方式，分析其年、月、小時之變化變動情形，以了解空氣品質之變化狀況。並對同測站內與不同測站間之污染物測項，以變異數分析 (ANOVA) 及相關分析 (correlation analysis)，來探討其差異及相關程度。此外再利用迴歸方式，得知臭氧受其他參數之影響情形。由研究結果得知，本市之原生性污染物呈現逐年降低趨勢，但二次污染物除在萬里站外，有逐年略升之現象。CO、NOx、SO2 在各月間差異並不明顯、PM10 在各站均以3-4 月較高、6-9 月較低。O3 以春、秋二季較高，夏季6-8 月較低。每日逐時變化上CO 與NOx 起伏較大，以半夜最低，上下午交通尖峰各有一高值；O3 則依日出而升、日落而降；PM10 與SO2 在每日間之變化起伏並不明顯。由測站內之污染物與氣象因子間之相關性研究，可知CO 多與NOx、NMHC (Non-Methane Hydrocarbon) 呈現較高正相關，與PM (particulate matther)、SO2 相關性則因站而異，與O3 呈現低負相關，可知一次污染物主要來源可能為附近移動源，二次污染物則可能有較多移流及累積等因地而之現象。由測站間相同指標物相關性研究，可知PM10 與O3 間呈高值度相關，代表受大尺度之因素造成；CO 與NOx 在各站間均為中度之正相關，係受區域性污染源影響；SO2 在各測站間之相關性較低，僅以地理位置相近之板橋、新莊、土城間較相關，表示可能受區域性之固定源影響。經由變異數分析顯示無論何項指標物，測站間均有顯著差異。在臭氧與其他變因之關係上，若以無截距方式設定線性關係，將明顯較有截距者為佳，而若僅以每日13-15 時測值進行迴歸，又將較全日測值之線性更佳。

In this study, we used the recorded database of 11 automatic air quality monitoring stations, which set up by the environmental protection administration, by year from 2001 to 2010 in New Taipei City to analysis the relationship between the main air pollutant ozone and other pollutants and meteorological factors. The results showed that the air quality have great improvement of primary air pollutants such as nitrogen oxide, carbon monoxide and sulphur oxide in last ten years, it means the air pollution control of stationary and mobile pollution sources working well. According to the statistics of 10-year Taiwan's Pollutant Standards Index (PSI), the major index pollutant of poor air quality days (Pollution Standards Index > 100) was ozone especially at Xindian station. From pollutants correlation analysis within each monitoring station, CO has strong positive correlation with NOx, NMHC and middle positive correlation with PM10, SO2 this inferred the domination of combustion sources. According to the correlation of one pollutant between stations, PM10 and O3 were high and CO and NOx were middle correlation revealing that PM10 and O3 were large-scale and CO and NOx were regional pollutants. Humidity presents middle negative correlation with O3 that means humidity restrains the generation of O3. Ozone also has negative correlation with its precursors such as NMHC and NOx. The results of one-way ANOVA (analysis of variance) of these stations database for recent 5 years indicated that there were significant difference between monitoring stations. According to the result of linear regression analysis of O3 with other parameters, without intercept was much better than with intercept regression. Besides, using the database of highozone-hour such as 1pm to 3pm were better than whole-day-hour. Therefore, we suggested that one should take the high concentration hour data of other parameters and regression without intercept to predict O3’s trend.