| 摘要: | 本論文旨在評估福爾摩沙衛星三號與福爾摩沙衛星七號之掩星觀測資料對於電離層數值天氣預報的影響。本研究使用系集卡曼濾波技術同化電離層-熱氣層耦合模式與福爾摩沙衛星三號及福爾摩沙衛星七號所測得之電子密度剖線與斜向全電子含量資料，以期能改善電離層資料同化系統的現報與預報能力。
首先，本研究將評估電離層與熱氣層之耦合作用對於電離層本身的現報與預報之影響。美國國家大氣研究中心所發展的資料同化開發模組將被用於同化福爾摩沙衛星三號電子密度剖現資料與熱氣層-電離層耦合環流電動模式(Thermosphere-Ionosphere-Electrodynamics General Circulation Model, TIE-GCM)。此資料同化開發模組與TIE-GCM之結合將有利於在數據同化與預報的過程中，電離層與熱氣層自洽性的耦合。觀測系統模擬實驗將被用來評估同化不同電離層與熱氣層變數對於電離層現報與預報之影響。實驗結果表明，電離層現報與預報之能力可藉由同時同化電離層與熱氣層的變數獲得改善，且中性大氣成分之影響尤為重要。
當利用福爾摩沙衛星三號與七號所測得之全球定位系統訊號反演電子濃度剖線資料時，需假設全球的電子濃度分布為球面對稱之分布。同化電子濃度剖線會同時將此假設所造成的誤差引入到數值模型中，而斜向全電子含量的資料則無此問題。故本研究的第二部分將著重在同化斜向全電子含量資料。然而TIE-GCM的上邊界高度過低，無法直接與斜向全電子含量進行同化。因此，在第二部分的實驗將使用美國大氣暨海洋總署所發展的新資料同化系統， Community Gridpoint Statistical Interpolation (GSI) Ionosphere，來取代先前的資料同化系統。此兩系統主要差異在於所使用的數值模型。在GSI Ionosphere中所使用之數值模型是由全球電離層電漿模型(Global Ionosphere Plasmasphere, GIP)與TIE-GCM所組成。此模型中，GIP沿磁力線進行電離層與電漿球層中電漿的模擬，熱氣層部分則由原本TIE-GCM中的中性大氣模組來模擬。相較於TIE-GCM，由於此模型之中低緯度封閉磁力線部分無須設定上邊界條件，且可考慮到熱氣層、電離層、與電漿球層的之間的耦合，故其應用更廣泛且彈性，可涵蓋的高度也較廣。
在第二部分的研究中，福爾摩沙衛星七號斜向全電子含量之觀測系統模擬實驗將被用來研究GSI Ionosphere在電離層現報與預報上的表現。與此同時，系集卡曼濾波器之參數對於同化福爾摩沙衛星七號斜向全電子含量之結果的影響也將被評估。本研究中主要評估的系集卡曼濾波器參數包括系集樣本大小與限地化參數。研究發現，利用GSI Ionosphere同化斜向全電子含量可大幅改善電離層中低緯度現報之結果。
在完成觀測系統模擬實驗後，真實的福衛三號斜向全電子含量資料將更進一步被同化以測試GSI Ionosphere在真實情況下對電離層現報之能力。在第三部分的研究中，我們主要關注在太陽平靜且無磁暴時期，GSI Ionosphere在電離層中底緯度的現報表現。實驗發現，GSI Ionosphere可成功同化真實的福衛三號斜向全電子含量資料並展現出優異的電離層中低緯度現報能力。我們可以預期，待未來福爾摩沙衛星七號升空並取得更多的觀測資料後，此系統的電離層現報與預報之能力將可大幅提升。
;This dissertation evaluates the ability of the Formosa Satellite-3/Constellation Observing System for Meteorology, Ionosphere and Climate (FORMOSAT-3/COSMIC) and the FORMOSAT-7/COSMIC-2 Global Navigation Satellite System (GNSS) Radio Occultation (RO) data to improve ionospheric specification and forecasting in the contest of ionospheric Numerical Weather Prediction (NWP) aided by data assimilation. For this purpose, both electron density profile and slant Total Electron Content (sTEC) data provided by the FORMOSAT-3/COSMIC and the FORMOSAT-7/COSMIC-2 missions are assimilated into a coupled first-principles model of the thermosphere and ionosphere by using ensemble square root filters. Along with evaluation of observational data, the dissertation investigates roles of incorporating ion-neutral coupling into data assimilation processes and the impact of different auxiliary ensemble data assimilation methods.
The role of ion-neutral coupling to ionospheric NWP systems is addressed first in this study. The FORMOSAT-3/COSMIC electron density profiles are assimilated into the Thermosphere-Ionosphere-Electrodynamics General Circulation Model (TIE-GCM), using Ensemble Adjustment Kalman Filter (EAKF) implemented in the Data Assimilation Research Testbed (DART). Combining the EAKF and the TIE-GCM allows a self-consistent treatment of ion-neutral coupling in both analysis and forecast steps of data assimilation. Both thermospheric and ionospheric variables are updating according to the background error covariance in the EAKF, and the updated state variables further affect other state variables through the physical ion-neutral coupling processes described by the TIE-GCM in Observing System Simulation Experiments (OSSEs). By updating different combinations of thermospheric and ionospheric variables in OSSEs, the impact of coupling processes on ionospheric NWP can be assessed. It is demonstrated that the incorporation of ion-neutral coupling can significantly improve the ionospheric electron density analysis and forecast, with the thermospheric composition being the most significant state variable.
Next, in the second part of this study, RO sTEC data, instead of electron density profiles, are attempted with the goal of further enhancing the ability of the FORMOSAT-3/COSMIC and FORMOSAT-7/COSMIC-2 to improve ionospheric specification and forecasting by eliminating the errors in electron density profiles that are introduced by the assumption of a spherical symmetric ionospheric electron density distribution in retrieval processes. Since the upper boundary of the TIE-GCM is 500-700 km, which is too low for assimilation of GNSS RO sTEC, another data assimilation system, the Community Gridpoint Statistical Interpolation (GSI) Ionosphere is used. The GSI Ionosphere is constructed using the GSI Ensemble Square Root Filter (EnSRF) and the Global Ionosphere Plasmasphere/TIE-GCM (GIP/TIE-GCM). Because the EAKF and the EnSRF are methodologically equivalent, the major difference between the DART/TIE-GCM and the GSI Ionosphere is the first-principles models employed in these two systems. The GIP/TIE-GCM is a coupled model of the thermosphere, ionosphere, and plasmasphere, and the plasma in the GIP/TIE-GCM are simulated along geomagnetic flux tubes, allowing the model to be extended to the plasmasphere, which adds flexibility for various future applications, such as incorporating the model with observation at higher altitude or studying the role of coupling of plasmasphere and ionosphere on ionospheric NWP.
In order to make quantify assessments of capability of the GSI Ionosphere to improve the low- and mid-latitude ionospheric specification and forecasting through assimilation of GNSS RO sTEC, a number of OSSEs are carried out by using synthetic FORMOSAT-7/COSMIC-2 data. An additional question addressed with the GSI ionosphere is the comparative evaluation of the FORMOSAT-3/COSMIC and FORMOSAT-7/COSMIC-2 observing systems. The effect of major EnSRF parameters, including the ensemble size and covariance localization schemes, on the assimilation analysis is investigated extensively, and are optimized to yield the highest quality assimilation analysis within the parameter range explored in this dissertation. The result shows that assimilation of sTEC data from the FORMOSAT-7/COSMIC-2 mission through the use of the GSI Ionosphere could potentially improve the low- and mid-latitude ionospheric specification considerably.
In the third part of this study, for the purpose of demonstrating the ability of the GSI Ionosphere data assimilation system to improve low- and mid-latitude ionospheric monitoring and forecasting with actual data rather than synthetic data, a case study is carried out under geomagnetically quiet and low solar activity conditions with the FORMOSAT-3/COSMIC GNSS RO sTEC data. Real FORMOSAT-3/COSMIC GNSS RO sTEC data during January 01, 2013 and January 02, 2013 is assimilated into the GSI Ionosphere, and the result is compared with the Center of Orbit Determination in Europe (CODE) Global Ionosphere Maps (GIMs). Result shows that, by assimilating sTEC data into the GSI Ionosphere, the feature of the Equatorial Ionization Anomaly (EIA) in the GSI Ionosphere becomes closer to that of the CODE GIMs. However, the correction of EIA magnitude by using EnSRF still need to be addressed. Comparing with the FORMSAT-3/COSMIC, the FORMOSAT-7/COSMIC-2 will be able to provide more dens data volume in low- and mid-latitude region. As a result, this issue can be solved by incorporating the FORMOSAT-7/COSMIC-2 data into the GSI Ionosphere in the future, since the sTEC data volume will be increased considerably |
