|摘要: ||大氣邊界層(Planetary Boundary Layer, PBL)位於地球大氣最底層，其結構發展除了與太陽輻射加熱、人類經濟活動及植被分布密切相關外，日夜的演變也影響許多小尺度天氣現象與整體空氣品質。但過去研究中，大氣邊界層的資料受限於觀測技術，缺乏高解度的時間與空間資料，且不論是氣象塔、探空氣球、繫留氣球或是飛機觀測都有其優缺點與使用限制，故考量到人力和觀測成本，許多觀測手段是無法做到整天或每天的常態觀測，目前台灣大多以探空資料或地面光達遙測來取得大氣邊界層熱力結構與空氣污染物的垂直分布。無人機是近年來新興的觀測平台之一，成本低廉，所需的人力也相對簡便，因此本研究旨在建構一套適用於大氣邊界層的無人機觀測系統，並以此來探討邊界層垂直發展結構與空氣污染物的變化，希望有助於邊界層機制參數化的建立與改善氣象及空污模式的預報。|
本系統搭配三種無人機酬載氣象及氣膠光學儀器，以獲取大氣邊界層溫度、相對濕度、壓力、風向、風速、氣膠粒子濃度與粒徑分布的垂直分布資料。為了減少所使用的儀器於量測上的誤差，先後進行了室內與室外的平行比對，室內溫濕度與壓力平行比對的均方根誤差(Root-Mean-Square Error, RMSE)為0.30℃、1.84%、0.11 hPa，室外日間與夜間3 km內溫濕壓平行比對的RMSE則分別為0.74℃、3.54%、0.72 hPa及0.21℃、3.34%、0.14 hPa，誤差皆於合理範圍且資料具高度相關性。實際觀測上，從2017年開始針對境外傳輸、本地污染及夏日北台灣邊界層發展結構等案例進行密集觀測，並整合探空氣象、剖風儀雷達風場與光達氣膠垂直分布等資料以驗證本系統實用與可靠性。結果分析顯示，本系統於0-3 km內的氣象探測結果與氣象局例行施放的探空觀測具良好的一致性，於邊界層高度的偵測與光達消偏振反演幾乎吻合。氣膠量測結果比較光達背向散射的連續觀測資料之中，垂直分布不連續處與本系統觀測的逆溫位置一致，氣膠垂直分布也與本系統所得類似。相較於與剖風儀雷達反演的風場，無人機於離地100 m後所量測的風向與風速RMSE分別為19.02°與1.91 m s-1，故仍有誤差待評估與修正。整體而言，本研究驗證本團隊所自行開發的無人機觀測系統及技術，可以有效應用於大氣邊界層內觀測，解釋大氣邊界層發展與空氣污染垂直分布間之關係，並可作為地面遙測觀測的驗證工具。
;The Planetary Boundary Layer is located at the bottom of the Earth′s atmosphere. Its structural development is closely related to solar radiation heating, human economic activities, and vegetation distribution. The evolution of day and night also affects small-scale weather and air quality. However, in the past studies, the data of the atmospheric boundary layer was limited by observation techniques, lacking high-resolution time and space data. Whether meteorological tower, radiosonde, tethered balloon or aircraft observation, both have advantages, disadvantages and use restrictions. Considering the cost of manpower and usage, many observation methods are unable to observe normal or daily normal observations. At present, the vertical distribution of the thermal structure of the atmospheric boundary layer and air pollutants is mostly detected by sounding data or ground-based remote sensing in Taiwan. Unmanned aerial vehicle (UAV) is one of the new observing platforms in recent years. The cost is much cheaper and the manpower needed is relatively simple. Therefore, the purpose of this study is to construct the UAV observation system for the atmospheric boundary layer and to explore the vertical structure of the boundary layer and the change of the air pollutants. It will help to establish the parameterized boundary layer mechanism and improve the prediction of the weather and air pollution model.
The system is equipped with meteorological and aerosol optical instruments on three kinds of drones to obtain the vertical distribution data of atmospheric boundary layer temperature, humidity, pressure, wind direction, wind speed, aerosol particles concentration and size distribution. In order to reduce the error of the instrument used in the measurement, the parallel comparison between indoor and outdoor has been carried out. The root-mean-square error (RMSE) of indoor temperature, humidity, and the pressure is 0.30 °C, 1.84%, 0.11 hPa, respectively. The RMSE of outdoor temperature and humidity within 3 km during outdoor day and night are 0.74 °C, 3.54% and 0.21 °C, 3.34%, respectively. The error is within a reasonable range and the data are highly correlated in actual observation, since 2017, this study focuses on the case for long-range transport, local pollution and the development structure of northern Taiwan boundary layer in summer, and integrates the data of sounding, wind profiler and the vertical distribution of the aerosol which is inversion by lidar to verify the practicality and reliability of the system. The results show that the meteorological observation results of the system within 0-3 km are in good agreement with the sounding observations by the central weather bureau. The detection of the boundary layer height is almost consistent with the polarization depolarization data. The main error source is the radiant heating and the response time of the sensor. The aerosol measurement results compare the continuous observation data of the backscattering signal. In addition, the aerosol results compared with the continuous observation data of the light backscattering. The vertical distribution of the aerosol distribution is similar to that of the system, except that the vertical distribution discontinuity is consistent with the temperature inversion observed by the system. Compared with the wind field inversion with the wind profiler, the wind direction and wind speed RMSE measured by the drone after 100 m from the ground are 19.02° and 1.91 m s-1 respectively, so there are still errors to be evaluated and corrected. As a whole, this study proves that the unmanned aerial vehicle (UAV) system and technology developed by our team can be applied to the atmospheric boundary layer to explain the relationship between the development of atmospheric boundary layer and the vertical distribution of air pollution, and can be used as a verification tool for remote sensing.