氣象場是進行空氣品質模擬的必要輸入，因此氣象模擬的好壞便會直接或間接地影響空氣品質模擬。風場決定污染物的傳送距離及方向，以及邊界層內的垂直混合程度，進而導致污染物的累積與擴散。尤其，污染的來源(排放量)主要由地面提供(交通幹道、工廠、森林等)，若能更精準地掌握近地面的風場，預期對於臭氧事件的模擬將能得到改善。 為加強地面風場的模擬，我們於模擬中應用四為資料同化分析納進法( FDDA analysis nudging )，將台灣環境保護署七十餘個地面監測站的氣象風場資料作為地面分析納進( surface analysis nudging )目標。然而，我們發現由於這些監測站密集地集中在台灣本島，相較於其他沒有監測站分布的區域(如海上)，模擬網域中分析場的風速場會產生無法預期的高風速情形。因此，我們額外加入”虛擬測站”做分析納進，能有效改善上述情形。 為了解分析納進法對於地面風場的改善於空氣品質模擬的影響，我們設計了三組實驗，分別測試納進係數、納進變數，納進網域對於模擬氣象場的影響，並探討以上情況下的大氣化學模擬結果與基本案例( 無任何納進的實驗 )的比較。結果顯示，納進以後的模擬的確能有效改善地面風場，以符合觀測值。納進變數，選擇同時調整風場與溫度場，導致垂直速度增加，污染物被傳送至較高空，而地面的臭氧無法累積；納進係數，雖然沒有開啟地面氣溫的納進，但是隨著係數使用的越大，地面氣溫的模擬有異常增溫的情況，同時，模擬的邊界層高度也有隨之增高的情形 ，這也對模擬的化學場有不良的影響；納進網域則討論三維分析納進( 3D analysis nudging )對於綜觀天氣形態的解析，由於模擬期間初期正好有納坦颱風侵台，納進網域實驗中發現，加強三維分析納進對於颱風路徑與風場有截然不同的掌握，但是並非主要臭氧事件發生日，因此對於化學場的模擬沒有顯著的差異。 由以上實驗結果顯示，臭氧的前驅物 NOx 以及 NMHC隨著風場的納進，其增長的趨勢與分布有顯著的不同，但是對於臭氧的生成與演變，分析納進實驗與基本案例沒有很大的不同。另外，硝酸與硫酸鹽類的濕沈降方面也有因為納進實驗產生些微的差異。 Meteorological fields are necessary input for air quality modeling. Wind field is responsible for pollutant transport and turbulent mixing in the troposphere. In this work we studied the impact of analysis nudging of meteorological variables on model performance both for meteorological variables and for air quality variables. Taiwan Environmental Administration has maintained a network of densely distributed monitoring stations for air quality with associated meteorological observations. We applied this network of 72 meteorological observations to nudge hourly wind speed, wind direction and temperature. Then we used the new meteorological data to conduct air quality study. We first noted that the standard analysis nudging technique as in MM5 and WRF models needs to be modified in order to be applicable to simulation domains such as Taiwan where nudging observations are all located near the center of the simulation domain. In such case surface nudging with the MQD method leads to undesirable changes in oceanic areas around Taiwan. With our modified MQD technique we are able to develop physically acceptable nudging meteorological fields. We first studied the impact of variations of the three surface nudging constraints, nudging coefficients, nudging variables and nudging domain on meteorological simulation. We found that surface nudging of wind field every 6 hours can reduce hourly surface wind speed to near observations for most of the simulation period. But temperature nudging can lead to unacceptable results if the nudging coefficient is too large. We found this is mostly due to unreasonably fast and deep planetary boundary layer development in the morning and at the evening. Such nudging will while improves surface temperature simulation does lead to overly strong vertical mixing and highly reduced surface pollutant mixing ratios very different from actual observation. Careful structuring of 3-D nudging domains also can improve meteorological simulations such as for the Typhoon Nockten period in 2004 which happens to be within the early part of our selected study period. For air quality study we found that surface nudging while improved simulation of surface wind field it did not change surface ozone simulation significantly, at least around the land based monitoring stations. But the change in 3-D wind structure did change local circulations that led to differences in off shore flow of pollutants and deposition of acidic materials.
