近年來，太空搭載之合成孔徑雷達感測器之相關技術迅速發展，提升地質災害監測的精度。永久散射體干涉技術（PSInSAR）是一種時序分析方法，可利用穩定相位的雷達目標，達到毫米級的地表變形監測精度。本研究應用PSInSAR技術，分析2019至2023年間宜蘭、雲林與屏東三縣的地層下陷情形。期間最大沉陷速率出現在雲林，每年達6公分。研究特別關注2021與2023年極端乾旱事件對地表沉陷的影響。整合下陷與降雨資料後發現，降雨變異與地層下陷具明顯關聯，三縣市的相關係數均高於0.7，其中雲林六個氣象站平均達0.89。為驗證結果準確性，本研究引入GNSS資料進行多參考點克利金內插校正，並採用留一交叉驗證法進行評估。結果顯示，PSInSAR與GNSS的變形速率高度一致，年變形率相關係數達0.95。與水利署2019至2023年公告之地層下陷範圍相比，本研究估算之下陷面積差異均在±10%以內，顯示PSInSAR具有良好的空間精度與穩定性。整體而言，本研究結合PSInSAR與GNSS技術，驗證其地表變形觀測之可靠性，對水資源管理與地層下陷防治具重要參考價值。;In recent years, spaceborne Synthetic Aperture Radar (InSAR) technology has rapidly advanced, enhancing the accuracy. Persistent Scatterer InSAR (PSInSAR), a time-series analysis method, uses radar targets with stable phase characteristics to achieve millimeter-level ground deformation monitoring. This study applied PSInSAR to analyze land subsidence in Yilan, Yunlin, and Pingtung counties from 2019 to 2023. The highest subsidence rate occurred in Yunlin, reaching up to 6 cm/year. The study particularly focused on the impact of extreme droughts in 2021 and 2023. By integrating deformation and rainfall data, a strong correlation was observed between rainfall variability and land subsidence, with correlation coefficients above 0.7 in all three counties and an average of 0.89 across six weather stations in Yunlin. To verify the accuracy of the results, GNSS data were incorporated for multi-reference kriging correction, and leave-one-out cross-validation was performed. The results showed high consistency between PSInSAR and GNSS deformation rates, with an annual rate correlation of 0.95. Compared to the official subsidence areas published by the Water Resources Agency from 2019 to 2023, the estimated area differences were within ±10%, demonstrating the spatial accuracy and stability of the PSInSAR method. Overall, this study confirms the reliability of combining PSInSAR and GNSS for ground deformation monitoring and provides valuable insights for water resource management and subsidence mitigation.
