利用合成孔徑雷達干涉技術及數值模擬方法分析雲林地區乾濕季地層下陷引起之水平變形

Abstract

合成孔徑雷達干涉技術 (Interferometric Synthetic Aperture Radar, InSAR) 為監測地層下陷之重要技術，具有面狀監測之優勢，可有效掌握地層下陷之時空變化。儘管地層下陷以垂直變形為主，地層下陷引起之水平變形可能使橋梁墩柱傾斜，甚至造成地表破裂，與垂直變形同樣具有嚴重之災害性。本研究以雲林地區為例，分析地層下陷引起之水平變形。雲林位處濁水溪沖積扇南側，長期因超抽地下水而發生地層下陷，1992年至今累積下陷量已超過190 cm，下陷中心位於土庫、元長一帶。此外，台灣高鐵恰好通過雲林地區下陷中心，因此需關注地層下陷可能對高鐵軌道造成之損害。雖然雲林地區已有許多InSAR監測成果，並發現地層下陷存在乾濕季變化，然而，由於InSAR沿視衛星 (Line of Sight, LOS) 方向進行測量，過去研究多假設水平變形可忽略，或將GNSS水平速度在LOS方向之貢獻扣除，以將LOS方向變形投影到垂直向，故缺乏InSAR對雲林地區下陷相關水平變形之研究。因此，本研究擬提供InSAR對雲林地區水平變形之觀測結果，並分析其季節性變化，討論雲林地區水平變形與地層下陷之關聯。為評估地層下陷引起水平變形之機制，本研究進一步以位於雲林地區下陷中心之秀潭國小地層資料建立數值模型，模擬地下水位變化於秀潭國小周圍造成之垂直及水平地表變形。本研究使用2018年5月至2021年5月Sentinel-1衛星之升、降軌影像，以短基線子集法 (Small Baseline Subset, SBAS) 及2D反演解算InSAR垂直向及東西向長期速度場，並將變形場依據乾濕季分為6期討論。分析通過下陷中心之剖面顯示，長期InSAR速度場於下陷中心存在數個具有較高下陷速度之下陷錐，最大下陷速度達78.5 mm/yr。同時，下陷錐位置存在約10 mm/yr之水平變形梯度。乾濕季變形場顯示，僅有濕季於剖面沿線出現較顯著之下陷錐，並發生高達10-15 mm之水平變形梯度，乾季則以區域性下陷為主，未有明顯下陷錐，且不存在明顯水平變形梯度。數值模型方面，本研究模擬三種地下水位情境造成之地表變形，在不同情境下水平變形皆較垂直變形快達到最大變形。另外，本研究分析不同數值模型參數對於地表變形之參數敏感度。其中，楊氏模數及蒲松比參數敏感度較高，大幅影響最終變形量，且水平變形對此兩項參數之敏感度較高，水力傳導係數之參數敏感度則相對較低，但影響變形速度。分析雲林地區於秀潭國小一帶出現下陷錐及高水平變形梯度之成因，數值模型成果顯示抽水可造成地表出現下陷錐及高水平變形梯度，且三種抽水情境之最大水平變形及垂直變形比約為0.20-0.27，InSAR於濕季量測之最大水平變形及垂直變形比則約為0.13-0.15，與數值模型成果相近，推測InSAR偵測到之下陷錐及高水平變形梯度由濕季過量抽水造成。高鐵沿線InSAR地表變形成果顯示，高鐵通過秀潭國小東側下陷錐，並於下陷錐處出現10 mm/yr之水平速度梯度。乾濕季地表變形成果亦顯示，高鐵沿線於下陷錐處出現高水平變形梯度，濕季水平變形梯度達約15 mm，乾季則約為10 mm，顯示高鐵沿線地表因過量抽水，出現較大水平變形梯度，並在濕季較為嚴重。本研究結合InSAR水平變形觀測結果，並由數值模型驗證地層下陷引起水平變形之機制，未來可做為地層下陷管理與防災之參考。

Interferometric Synthetic Aperture Radar (InSAR) is a crucial technology for monitoring land subsidence, offering the advantage of areal monitoring and effectively capturing the spatiotemporal variations of subsidence. Although land subsidence is primarily characterized by vertical deformation, the horizontal deformation induced by subsidence can lead to tilting of bridge piers or even ground fissures, posing hazards as severe as those caused by vertical deformation. This study focuses on horizontal deformation resulting from land subsidence in the Yun-Lin area as a case study. Located on the southern side of Chuoshui River alluvial fan, Yun-Lin area has experienced long-term land subsidence due to excessive groundwater extraction. Since 1992, the accumulated subsidence has exceeded 190 cm, with the subsidence center located around Tu-Ku and Yuan-Chang areas. Moreover, Taiwan High-Speed Rail (THSR) passes directly through the subsidence center in Yun-Lin area, raising concerns about potential damage to the railway tracks caused by subsidence. Numerous InSAR monitoring results on land subsidence are available for Yun-Lin area, which have also revealed seasonal variations in subsidence between wet and dry periods. However, due to InSAR measuring along the satellite’s line of sight (LOS) direction, past studies often assumed that horizontal deformation could be ignored, or subtracted the contribution of horizontal velocities in the LOS direction by GNSS data, to project LOS deformation to the vertical direction. Consequently, research on horizontal deformation associated with subsidence in Yun-Lin area by InSAR remains lacking. Therefore, this study aims to provide InSAR observations of horizontal deformation in Yun-Lin and analyze its seasonal variation to explore the relationship between horizontal deformation and land subsidence. To evaluate the mechanism behind horizontal deformation caused by subsidence, the study further develops a numerical model using stratigraphic data from Siu-Tan Elementary School (STES), located at the subsidence center in Yun-Lin area, to simulate vertical and horizontal surface deformation due to changes in groundwater levels around the area. This study utilizes ascending and descending orbit images from Sentinel-1 satellites between May 2018 and May 2021. By applying the Small Baseline Subset (SBAS) technique and 2D inversion, the long-term vertical and east-west velocity fields are derived. Cross-sectional analysis through the subsidence center reveals several subsidence cones with high subsidence rates, with the maximum rate reaching 78.5 mm/year. Around these subsidence cones, it shows horizontal deformation gradient of approximately 10 mm/year. The deformation field is further divided into six periods based on wet and dry seasons. It shows that significant subsidence cones and horizontal deformation gradients of up to 10–15 mm only appear along the cross-section during wet seasons. In contrast, dry seasons are characterized by regional subsidence without subsidence cones or horizontal deformation gradients. In the numerical modeling section, three groundwater level scenarios are simulated to evaluate surface deformation. In all scenarios, horizontal deformation reaches its maximum earlier than vertical deformation. Analysis on parameter sensitivity indicates that Young’s modulus and Poisson’s ratio significantly influence the final deformation magnitude, with horizontal deformation being more sensitive to these parameters. In contrast, hydraulic conductivity shows relatively low sensitivity but affects the deformation rate. The formation of subsidence cones and high horizontal deformation gradients near STES in Yun-Lin area is analyzed, and numerical modeling results show that groundwater extraction can cause both phenomena. The maximum horizontal-to-vertical deformation ratio across the three extraction scenarios ranges from 0.20 to 0.27, which is comparable to the InSAR-measured ratio of approximately 0.13 to 0.15 during the wet season, suggesting that the subsidence cones and high horizontal deformation gradients detected by InSAR are caused by excessive groundwater pumping during the wet season. InSAR surface deformation results along THSR reveal that the railway crosses a subsidence cone east of STES, where a horizontal velocity gradient of about 10 mm/year is observed. Seasonal surface deformation results also show a significant horizontal deformation gradient at the subsidence cone along the railway, approximately 15 mm in the wet seasons and 10 mm in the dry seasons, indicating that excessive groundwater extraction causes a larger horizontal deformation gradient along THSR, which is more pronounced during the wet season. This study combines InSAR-derived horizontal deformation observations with numerical modeling to validate the mechanism of horizontal deformation caused by land subsidence. The results can serve as a valuable reference for land subsidence management and disaster prevention in the future.