| 摘要: | 本論文主旨在於改進電離層電波掩星法反演電離層E層，在進行誤差校正前，必須先了解福爾摩沙三號衛星觀測資料的特性，文中利用2006年7月到2011年12月觀測到的電子濃度進行電離層E層的研究與探討。在進行資料分析前，需進行資料品質的控管，品質控管利用ionPrf電子濃度剖面的擾動量、F層電子濃度420到490公里的斜率、NOAA所提供的地磁擾動指數以及ionphs電離層散塊E層進行資料的篩選。從福爾摩沙三號衛星觀測電離層E層資料全球分佈以及日變化可以得知在日間電離層E層在地磁緯度區間南北緯30到50度以及磁赤道地區有明顯電子濃度異常的現象，在地磁緯度區間南北緯10度到30度則有電子濃度谷值，在磁緯度70度附近觀測到極光橢圓區。
使用國際參考電離層模型去模擬電離層80到2000公里的時空分佈，進而利用真實福爾摩沙三號衛星軌道去反演電子濃度剖面，並將模擬反演電子濃度剖面和國際參考電離層模型電子濃度比較，可以得知電波掩星法的反演誤差量。模擬結果顯示日間電波演星法反演電離層電子濃度從底層到高層皆在地磁緯度30到50度呈現高估，在地磁緯度10到30度則為低估，和實際福爾摩沙三號衛星觀測電子濃度資料一致。主要造成誤差的原因為電離層電子濃度不符合球對稱的假設，模擬結果顯示平均在日夜間電離層F層的誤差在35%以內，日間電離層E層誤差在200%以內，隨緯度而有不同的變化，夜間電離層E層則大於200%。除了檢視電子濃度的誤差量，文中也檢視了垂直電子含量的誤差，整體而言，垂直電子濃度含量在30%以內。
經過模擬和觀測資料相互比較後發現在不同地磁緯度以及當地時間變化上有強度上的差異，整體而言模擬反演E層電子濃度剖面符合COSMIC觀測到的E層電子濃度。根據全球電離層探測儀網、COSMIC觀測資料、IRI模型電子濃度和電波掩星法反演電子濃度，本文檢視並模擬2006年7月到2011年12月的資料。COSMIC量測以及IRI模型模擬電波掩星法都顯示百分比誤差(PE)和方均根誤差(RMSE)在地磁緯度和當地時間有很大的關係性，最小值發生在早上和高緯度地區，最大值發生在下午和中緯度地區。除此之外，PE和RMSE的季節變化似乎和緯度有關。在COSMIC觀測到的NmE移除IRI模型模擬電波掩星法造成的誤差後，統計結果也顯示校正後的資料，隨著不同月份COSMIC NmE仍然存在了-20%到38%的誤差量，大於IRI預測NmE的PE誤差量，範圍介於-6.5%到20%；校正後的RMSE範圍介於3.4*10^4到6*10^4 #/cm3，IRI模型的RMSE值介於1.2*10^4到2.3*10^4 #/cm3。
;The aim of this paper is the improvement of GPS Radio Occultation Retrieval Error of E Region Electron Density. Before the error correction was made, it was necessary to understand the feature of FORMOSAT3/COSMIC electron density profiles. On the basis of 66 month data (July 2006- December 2011), the investigation and analysis of the ionospheric E region electron density were made. In order to screen the data in harmony with the physical principle, the quality control algorithm was developed, including mean deviations, F region slope in a range of 420 and 490 km, Kp index provided by NOAA, and the fluctuations of signal-to-noise (SNR) ratio and excess phase. The global distribution and diurnal variations of COSMIC observation in E region show that there are three anomalous daytime enhancements in the magnetic latitude region ±30°–50° and magnetic equator, electron density depletion in the magnetic latitude region ±10°–30°, aural oval situated around 70° magnetic latitude.
The International Reference Ionosphere (IRI) model was selected to simulate temporal and spatial distributions of global electron densities from 80 to 2000 km in 2008. The COSMIC geometry were utilized to retrieve the electron density in each occultation events. The retrieval error were computed by comparing the retrieved electron density with IRI electron density in E and F region. The major cause of the retrieval error is ionospheric electron density which violate assumption of spherical symmetry. The result show that the simulation-retrieved electron densities seem to be significantly overestimated in the magnetic latitude regions ±30°–50° and magnetic equator, and underestimated in the magnetic latitude regions ±10°–30°, which is consistent with COSMIC observation. The retrieval error were ±35 % in the daytime and nighttime F region, ±200% in the daytime E region depended on the different latitudinal region, and greater than 200% in the nighttime E region. Besides the retrieval error of the electron density, the retrieval error of the vertical TEC was also examined. Overall, the retrieval error of the vertical TEC were ±30 %.
Despite regional discrepancies in the magnitudes of the E region electron density, the IRI model simulations can, on the whole, describe the COSMIC measurements in quality and quantity. On the basis of global ionosonde network and the IRI model, the retrieval errors of the global COSMIC-measured E region peak electron density (NmE) from July 2006 to July 2011 are examined and simulated. The COSMIC measurement and the IRI model simulation both reveal that the magnitudes of the percentage error (PE) and root mean square error (RMSE) of the relative RO retrieval errors of the NmE values are dependent on local time (LT) and geomagnetic latitude, with minimum in the early morning and at high latitudes and maximum in the afternoon and at middle latitudes. In addition, the seasonal variation of PE and RMSE values seem to be latitude dependent. After removing the IRI model-simulated GPS RO retrieval errors from the original COSMIC measurements, statistics show that the monthly residual errors remained in the corrected COSMIC-measured NmE vary in a range of -20% - 38%, which are comparable to or larger than the percentage errors of the IRI-predicted NmE fluctuating in a range of -6.5% - 20%. The RMSE after correcting are in a range of 3.4*10^4 - 6*10^4 #/cm3, which are greater than the RMSE of the IRI-predicted NmE fluctuating in a range of 1.2*10^4 - 2.3*10^4 #/cm3. |
