摘要
本研究的目的在於能夠運用全球定位系統(Global Positioning System, GPS)觀測資料反演出天頂向遲延量(Zenith Total Delay, ZTD)，再由天頂向遲延量配合地面觀測資料轉換為可降水量(Precipitable Water, PW)，並探討可降水量與真實降雨量之間的關係，最後將天頂向遲延量帶入WRF氣象模擬，探討天頂向遲延量對氣象模擬是否有幫助。本文所選案例是2012年的610水災，所選時間為6月9日至6月14日的梅雨期間，在這段期間內大量降雨對台灣造成重大災情。而GPS資料運算方面是使用瑞士伯恩大學(University of Bern, Switzerland)發展的Bernese GPS Software 5.0做為運算軟體，使用的氣象模擬軟體為WRF (Weather Research and Forecasting Model) V3.5版本，GPS網狀觀測資料、地面觀測資料是由中央氣象局提供。研究結果顯示，反演出的可降水量在曲線趨勢上與真實降雨量曲線趨勢及降雨分佈有一定程度的關係。可以觀察出高強度降雨前的水氣上升以及降雨後的水氣明顯下降。在海拔高於兩千公尺山區推算出的平均可降水量為41mm，明顯低於在海拔較低的平均65mm，較可能的原因是因測站在海拔較高區域GPS訊號所受到的路徑遲延量相比海拔較低時少。在WRF氣象模擬方面，經過ZTD資料同化過後的模擬結果，在24小時及48小時預報方面也較未使用ZTD資料的結果接近真實降雨量。在ETS及BIAS得分方面也是較佳的，結論當進行WRF氣象模擬時配合上品質較佳的GPS反演ZTD資料，在這案例中相較於未使用ZTD資料的WRF氣象模擬在分數及降雨分佈的上會有些許的改善。
並列摘要
The accuracy of the Global Positioning System (GPS) has been studied extensively. In this research we will use the satellite navigation ground-based observation to improve weather forecast. This study is about the 610 flood (2012) in Taiwan. The GPS observation data and surface observation data are provide by Taiwan Central Weather Bureau (CWB) and GPS data process software is Bernese5.0. In this research we have two part. Frist part we compute the GPS observation data to get Zenith Total Delay (ZTD), and using the surface temperature and the pressure to estimate the Precipitable Water Vapor (PWV), then we will compare the relationship of real rainfall and GPS station elevation. In second part we will investigate the impacts of ZTD data assimilation in Weather Research and Forecasting (WRF) model. The experiments results show that the rainfall occur when the PW is increasing, but we cannot infer that the high PW will occur rainfall, because there have a lot rainfall predisposing factor like Cloud physics、temperature、wind direction etc. In the WRF simulation experiments results, ETS and BIAS shows that the initial field with ZTD data assimilation are better than the initial field without ZTD data assimilation, although sometimes the ZTD performance not so well, but in this experiments the ZTD still shows little assist in improve weather forecast.