合成孔徑干涉雷達(InSAR)是一種用於監測地球表面的有用技術，其常見的應用之一是繪製地形圖。對於重複式軌道干涉的雷達系統而言，傳統的InSAR方法在取得高解析度地形圖常會受到限制，因為不同類型的去相關、大氣延遲、散射特性、表面變形和幾何變形都是影響最終成果精度的因素，這使得探索誤差來源及其影響程度成為一項複雜的挑戰。因此，此研究的目的是利用理論公式討論地表變形和對流層延遲對最終地形產品的影響。為了減輕全相位回復誤差的影響，此研究在InSAR處理過程中額外套用相干值門檻。因此，使用相干值門檻的有效性是在誤差分析的討論之外的另一個主題。 研究結果顯示，相干值門檻的應用可以幫助約89%的像對提高計算地形的準確性。 此外，隨著門檻值的提升，全相位回復錯誤的發生顯著減少。然而，相干值門檻如何影響全相位回復步驟的確切原因仍然是一個問題。在河流和丘陵等低相干區域很難觀察到相門檻值的一致影響，然而在高相干值區域的影響明顯較小。在回歸分析中，雖然某些單個圖像對中可以觀察到顯著的相關性，但是當所有像對合併進行分析時，相關性即變得非常小。除此之外，相關性的一些符號與理論假設相矛盾。 ;Interferometric Synthetic Aperture Radar (InSAR) is known as a useful technique for monitoring earth surface. One of the common applications is topography mapping. For a repeat-pass spaceborne radar system, conventional InSAR approach is limited in achieving high vertical accuracy owning to its inherent properties, such as different sorts of decorrelation, atmospheric delays, scattering properties, surface deformation, and geometric distortion. These are factors that contaminate final results during the whole imaging system, which makes it a sophisticated challenge to explore the drivers and their exact influences. The aim of this work is to discuss the impact from surface deformation and tropospheric delays on end topography products based on mathematic description. In order to alleviate the influence of phase unwrapping error, coherence masks are applied during InSAR processing. Therefore, the effectiveness of coherence masks is also one topic that is concerned in this research in addition to the discussion of the error analysis. The result has shown that the application of coherence mask can improve the accuracy of measured heights for about 89% image pairs in our experiments. Also, the occurrence of phase unwrapping errors can be significantly reduced as the threshold increases. However, the precise reason for how coherence mask influence phase unwrapping remains a question. It is difficult to observe a consistent influence of coherence thresholds in low coherence areas, such as rivers and hills, while the influence in high coherence area is subtle. In error regression analysis, some significant correlations between the errors and the two considered error sources, refractivity and deformation. The relationship between the errors and deformation is stronger than that with refractivity. Although the relationship in the experiment does not meet the expectation built from the derived mathematical description, the work has constructed a reasonable theoretical relationship and conducted a preliminary attempt to fit the relationship in reality.
