Please use this identifier to cite or link to this item: http://hdl.handle.net/11455/4946
標題: 台灣地區原生污染物及二次氣膠之分析與模擬暨氣狀污染物垂直剖面之觀測
作者: 陳建隆
Chen, Chien-Lung
關鍵字: 空氣品質模式;air quality model;軌跡模式;傳遞係數;貢獻濃度;原生污染物;二次氣膠;濃度垂直剖面;主成份分析;trajectory model;transfer coefficient;contributed concentration;primary pollutants;secondary aerosol;vertical profile;principal component analysis
出版社: 環境工程學系
摘要: 
本研究利用主成份分析統計方法,探討污染物在地面水平分布特性,將台灣初步劃分為8個氣源區,並觀測草屯地區近地表1700公尺內氣象因子垂直剖面及氣狀污染物隨時間及高度分布情況。而所發展中興大學PM軌跡傳遞係數模式已初具雛形,已有系統的考慮乾沈降、溼沈降、水平與垂直的擴散、氣固相的轉換,對於粒狀物乾沈降模式加以驗證、污染物濃度一週變化的調整、污染物年排放量修正及模式參數敏感度測試,並將模式應用在台灣北中南地區,模擬空氣品質及二次氣膠濃度,計算各網格的傳遞係數及對受體點的貢獻濃度,探討污染源與受體點之間的關係,得知對受體點衝擊較大的貢獻區域,可做為空氣污染防制、減量措施、總量管制策略及國土使用規劃的參考。本研究主要成果如下:
(1) 綜合環保署1994年至2000年59個空氣品質測站污染物濃度日均值主成份分析結果,將台灣地區分為北部氣系、桃園沿海氣系、中部氣系、中部沿海氣系、南部氣系、宜蘭氣系、花蓮氣系以及台東氣系等8個氣源區。
(2) 草屯地區三次密集探空實驗結果顯示,白天的垂直對流混合較好,SO2的垂直剖面較沒有規則,CO濃度隨高度增加而降低,在1000公尺處的濃度急劇減少,約為地面濃度的0.5倍。NO2在地面50公尺內因移動源排放及NO的轉換,濃度隨高度增加而稍微增加。雖然地面有沈降的作用會造成濃度梯度,仍可觀察到不論在晴朗或陰天的夜晚,因夜間煙霧反應無法進行,接近地表100公尺內的NO2+O3滴定效應相對的顯著。
(3) 由臭氧濃度之垂直剖面特性,發現白天臭氧的高值出現在500公尺左右,在夜間邊界層上常存在相對的高臭氧濃度,故地面臭氧濃度難以代表近地表1000公尺內的臭氧濃度情況,而白天地面的臭氧濃度會受清晨逆溫層上臭氧的向下混合效應所影響。
(4) 排放量資料有必要針對經濟生產、交通活動的變化、區分工作日與休假日加以修正,尤其在都會地區,考量排放量短期間變化是有必要的,考量排放量一週的變化可改善模式模擬的相關係數及成功模擬事件日的比率。而以溼絕熱傾率替代觀測探空資料在台灣北部地區是可行的,空氣品質模擬相關係數及成功模擬事件日比率,並沒有明顯的差異。
(5) 模式參數敏感度測試結果顯示,對PM10濃度而言,以乾沈降速度及原生PM10排放速率影響最大,SI分別為-118 %及91 %,而氣固相轉換速率、洗滌係數、軌跡線高度及氣狀物轉換為二次氣膠在細粒中的比例則對PM10濃度影響較小,SI在5 %以下。相同地,乾沈降速度及排放速率對於CO、NOx、SO2及PM2.5濃度的影響也較顯著。
(6) 模式驗證1998年古亭、忠明及鳳山測站觀測值與模擬值結果的比較,CO、NOx、SO2、PM2.5及PM10之相關係數為古亭測站0.75、0.69、0.39、0.55與0.55,忠明測站0.72、0.51、0.50、0.48與0.53,鳳山測站0.15、0.52、0.08、0.54與0.52,和一些以統計模式之模擬結果相較並不遜色。
(7) 1998年9月-12月三個時段於台中市懸浮之微粒密集採樣分析進行二次氣膠模擬,結果顯示在發生PM10污染事件的7天中,台中及通霄電廠因SO2及NOx排放後轉換成二次氣膠及衍生的銨鹽,對於忠明測站的貢獻量硫酸鹽佔了58%,硝酸鹽佔了18%,銨鹽佔了44%,而PM2.5則佔了13%。但模式對於高煙囪的擴散及沈降機制仍有待改善,以點源排放為主的SO2及高雄都會區的CO,模擬之相關係數皆較差,夏季模式模擬高估,顯示有必要對模式機制再加檢驗,各參數對於季節性的變化也要考量。
(8) 中興大學PM軌跡傳遞係數模式利用模擬的貢獻濃度表達污染物與受體點遙距相關(teleconnection)的現象,對遠方受體點所造成的衝擊,可以推估空氣品質及探討污染源與受體點的關係。
(9) 1998年台灣地區PM空氣品質的惡化應有受到大陸沙塵暴影響。對於古亭測站,模式未能模擬出高值的21個事件日中,便包含了12個沙塵暴事件日,顯示本研究目前無法模擬沙塵暴數千公里的長程傳輸,概因模擬尺度的大小、邊界值的假設與觀測,以及缺乏結合大尺度氣象場之模擬結果所導致。而沙塵暴對於古亭、忠明及鳳山測站的影響則由北而南的遞減。

This study discusses the characteristics of the pollutant concentrations distribution near ground and in the atmospheric boundary layer. Sixty air quality stations are grouped using principal component analysis (PCA) in order to classify the air basin regions. Using a tethered balloon, vertical air pollutant concentrations (CO, SO2, NO, NO2 and O3) were measured in central Taiwan during field campaigns in the winters of 1999 and 2001. These data were collected in order to examine the temporal and vertical variations of pollutants. Then, A new algorithm has been derived for trajectory models to determine the transfer coefficient of each source along or adjacent to a trajectory and to calculate the concentrations of CO, SO2, NOx, sulfate, nitrate, fine particulate matter and coarse particulate matter at a receptor. The mechanisms of wet scavenging, dry deposition, vertical and horizontal diffusion, and gas-to-particle conversion processes have been included in the determination of transfer coefficients. The dry deposition velocity of particulate, the emission rates of pollutants and the sensitivity of calculated concentrations to a variety of model parameters and inputs are tested. The idea of determining transfer coefficient by an air quality model is important for the design of a better abatement strategy, the total capacity control strategy and the national landuse planning project. This study shows some results as follows:
(1) Using 1994 to 2000 data from sixty air quality monitoring stations (AQMS) for PCA, AQMS with the same maximum loading factor are grouped together. This study classifies Taiwan into eight air basins, named northern, northern coastal, central, central coastal, southern, Ilan, Hualien, and Taitung.
(2) The vertical profiles showed that there were no significant trends in altitude for SO2. The CO concentration decreases sharply above 1000 m, while showing a decrease with height throughout the day. Nitrogen dioxide presents a slight increase gradient with height in the lowest 50 m of the atmosphere. The production of NO2 is due to vehicle emissions and NO transformation. However, the loss of dry deposition dominates near the surface. Except for a slight decrease near the ground the titration of O3 by NO to produce NO2 was observed notably below 100 m height in nighttime.
(3) The vertical profiles of the mean ozone concentrations demonstrate that high ozone concentrations appear at about the 500 m height in the daytime. Higher ozone concentrations are present above the NBL at nighttime. A fair correlation appeared between the maximum ozone concentration at the surface during the daytime and average ozone concentration above the nocturnal boundary layer.
(4) The model that took into consideration the day-of-the week variability of emissions improved the simulated results of the correlation coefficients. However, if there is no airsonde data, the lapse rate above mixing height is fixed to be the wet adiabatic lapse. There is no significant improvement in using observed lapse rate than using wet lapse rate.
(5) The sensitivity analysis shows that the calculated PM10 concentration is most sensitive to its dry deposition velocity as well as its emission rate. Sensitivity index of the deposition rate is —118% and that of the emission rate is 91%. The PM10 concentrations are less sensitive to the gas-particle conversion rates, the trajectory height and the size fraction of converted second aerosols. Similarly, it is found that CO, NOx, SO2 and PM2.5 are most sensitive to their emission rates and dry deposition velocities.
(6) The correlation coefficients of the calibration months in 1998 for CO, NOx, SO2, PM2.5 and PM10 are 0.75, 0.69, 0.39, 0.55 and 0.55 at Gu-Ting station, 0.72, 0.51, 0.50, 0.48 and 0.53 at Chung-Ming station and 0.15, 0.52, 0.08, 0.54 and 0.52 at Fong-Shan station. The above correlation coefficients can be compared with statistical models.
(7) The major advantage of the Gaussian trajectory transfer-coefficient model is that the sources' apportionment of the receptor can be determined in a single model run. The contributed concentrations from the two power plants for Chung-Ming station during the campaigns from September to December 1998 are calculated. 58% of secondary sulfate aerosol, 18% of secondary nitrate aerosol, 44% of secondary ammonium aerosol and 13% of PM2.5 were contributed from the two power plants during the episode days. However, the mechanism of plume dispersion need to be improved for the overestimation.
(8) The trajectory transfer-coefficient model can fund the source-receptor relationship for air pollution problems associated with long-range transport. According to contour plots of contributed concentrations to Gu-ting station, “teleconnections” between source emissions and their contributions to the station can be identified.
(9) There are twelve uncaptured PM10 episodes (15 Apr.-20 Apr., 17 - 18 Oct., 5 Nov., 10 Nov., 8 Dec. and 15 Dec.) corresponded with periods when Asian dust storms affected Taiwan in 1998. According to the ratios of PM2.5 to PM10 at Gu-Ting, Chung-Ming and Fong-Shan stations, the influence of dust storm was significant in northern Taiwan and decreasing from northern to southern Taiwan.
URI: http://hdl.handle.net/11455/4946
Appears in Collections:環境工程學系所

Show full item record
 

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.