透過密集陣列微震分析建立宜蘭平原地熱探勘之高解析度剪力波速度模型

Abstract

宜蘭平原位於臺灣東北部，位於沖繩海槽西南端，受弧後擴張與岩漿入侵影響，是一潛在地熱活動顯著的區域。為探討其地熱潛力與地下構造，本研究於 2022 年 7 月至 2023年 1 月期間於整個宜蘭平原佈設了 81 個測站的區域測站網與紅柴林地區 186 個測站的高密度陣列，進行為期六個月的微地動連續觀測。首先本研究利用水平-垂直頻譜比（H/V spectral ratio）方法來描繪宜蘭平原的岩盤形貌。透過傅立葉轉換分析各測站的三分量(Z, N, E)資料，計算並平滑化 H/V 振幅頻譜比，以識別共振頻率的頻譜峰值，並轉換成岩盤上覆沉積層的厚度，結果與先前的地震剖面及S-P 轉換相位研究結果高度一致。接著針對紅柴林地區，應用極化度橢圓率分析法（DOP-E）於時頻域中從環境噪訊中擷取 0.2 - 5 Hz 穩定的雷利波橢圓率訊號，並透過鄰域演算法（Neighbourhood Algorithm）進行反演，建立每個站的淺層一維剪力波速度（Vs）模型，進而透過密集的測站分佈合成高解析的三維模型。模型結果顯示在 3 至 4 公里深處有多處低速異常區，包括員山一號井（臺灣首口深層地熱型研究井）下方、紅柴林東南方及利澤地區，顯示這些區域可能具有潛在地熱儲集層的特徵。與大地電磁（MT）資料的比對也顯示兩者在 1 公里內的淺層構造十分相符，高、低電阻區大致對應到高、低剪力波速區。驗證本研究模型的可信度。而相對MT 技術的空間解析力會隨深度降低，本研究之速度成像的空間解析力主要基於測站間距，故可提供較清晰的深部構造資訊。綜上所述，本研究展示利用非侵入式的單站微地動分析能作為高效低成本的地熱探勘方法，快速評估張裂盆地的岩盤結構與地熱潛能，為未來宜蘭平原地熱資源的開發與鑽井規劃提供重要參考。

The Ilan Plain, located in northeastern Taiwan at the southwestern tip of the Okinawa Trough, is influenced by back-arc extension and magmatic intrusion, making it a region with significant geothermal potential. To investigate its geothermal resources and subsurface structures, this study deployed 267 seismic stations, including 81 stations across the Ilan Plain and 186 high-density stations in the Hongchailin area, for six months of continuous ambient noise monitoring from July 2022 to January 2023. The study first applied the Horizontal to Vertical Spectral Ratio (HVSR) method to delineate the basement geometry of the Ilan Plain. By performing Fourier transforms on three-component seismic data (Z, N, E) at each station, we computed and smoothed the amplitude spectra to identify peak frequencies related to resonance. These peaks correspond to the thickness of sediments overlying the basement and are highly consistent with previous seismic profiles and S-to-P converted phase studies. Based on the estimated sediment thicknesses, we constrained the layering depths required for inversion. The Degree of Polarization Ellipticity (DOP-E) method was then applied in the time-frequency domain to extract stable Rayleigh wave ellipticity signals from ambient noise. These measurements were inverted using the Neighborhood Algorithm to obtain high-resolution one-dimensional shear-wave velocity (Vs) models. The results revealed significant low-velocity zones beneath Yuanshan Township, which hosts Taiwan’s first deep geothermal well, as well as southeast of the Hongchailin area, indicating high geothermal potential. Synthetic tests confirmed the accuracy and stability of the inversion workflow. Sensitivity kernel analysis demonstrated that the model resolution extends to a depth of 5 kilometers. Several low-velocity anomalies were identified at depths between 3 and 4 kilometers, including beneath the Yuanshan geothermal well, southeast Hongchailin, and the Lize area, suggesting the presence of potential geothermal reservoirs. A three-dimensional tomography model was constructed using the one-dimensional Vs results from all stations, generating velocity cross-sections from 1 to 5 kilometers in depth. These images reveal lateral variations in the subsurface shear-wave velocity. A comparison with magnetotelluric (MT) resistivity data showed strong agreement within the upper 1 kilometer, supporting the validity of our Vs model. While MT suffers from limited resolution in shallow regions, the tomography developed in this study provides clearer imaging of deeper structures. In conclusion, this study demonstrates that microtremor analysis is an effective, non-invasive, and efficient single-station method for rapidly assessing basement structures and geothermal potential in rifted basins. The results offer valuable insights for future geothermal exploration and well-targeting strategies.