利用地震噪訊H/V頻譜比分析日本隱沒帶增積岩體淺層構造

Abstract

本研究分析2011年9–12月於紀伊半島外海布放之 147 部海底地震儀三分量環境噪訊，應用地震噪訊H/V頻譜比分析 (Horizontal-to-Vertical Spectral Ratio analysis, HVSR) 解析海床淺層構造。當地震波在地層傳播時，在不同波速阻抗的界面上發生反射和折射。當剪力波經過有明顯阻抗差異的界面時，水平方向的振動會產生共振效應。因此將水平頻譜除以垂直頻譜，其能量波峰對應此地層的共振頻率，稱為波峰頻率。因此可根據波峰頻率估算主要速度變化層 (通常為沉積層) 的波速與厚度。 研究區域位於菲律賓海板塊隱沒至歐亞大陸板塊的聚合邊界南海海槽，測站分布在4條縱向跨越弧前盆地 (熊野弧前盆地及室戶弧前盆地的東緣)、大陸斜坡和南海海槽，以及1條東西橫跨兩弧前盆地的測線。在所有測站獲得的HVSR曲線中，有135個測站辨識出波峰頻率，頻率愈低代表界面愈深，而HVSR振幅愈高則表示速度阻抗差愈大。因此基於各站的波峰頻率，以及對應的HVSR振幅，使用k-means群聚分析分成7群。分析結果指出：海溝區呈0.6–2 Hz低頻高振幅特徵，對應70–220 m厚之低速沉積層並可追蹤至相變阻抗面；熊野弧前盆地的多峰HVSR頻率隨測線向盆地中心遞減；大陸斜坡區域則以 > 6Hz的高頻低振幅HVSR波峰為主，顯示薄層沉積層或基盤裸露。之後也選擇波峰較顯著且連續的海槽段和弧前盆地段進行HVSR曲線逆推，取得淺層地震波速度模型，和前人研究結果比對後，發現Vp/Vs比模型可描繪出沉積構造單元和波速剖面中低速帶形貌。 HVSR波峰頻率在三個月時序中，除布放初期的壓密效應與偶發外來噪訊外，整體變化甚微，而其空間變化則與海溝、大陸斜坡與弧前盆地等構造單元呈系統性對應。本研究藉由長時間、大空間尺度觀測的環境噪訊資料，透過波峰頻率在時空中的分布特徵，顯示深海區域HVSR與地質構造具有高度相關性，並以此為基礎討論增積岩體淺層剪力波速度阻抗層面的深度，量化隱沒帶淺層構造物理特性。

This study analyzes three-component ambient seismic noise recorded by 147 ocean bottom seismometers (OBS) deployed off the Kii Peninsula between September and December 2011. We apply ambient-noise Horizontal-to-Vertical Spectral Ratio (HVSR) analysis to delineate the shallow structure. When seismic waves propagate through layered media, reflections and refractions occur at interfaces with contrasting seismic impedances. A shear wave crossing a pronounced impedance contrast excites resonance in the horizontal components; dividing the horizontal by the vertical spectrum yields an HVSR peak whose frequency represents the fundamental resonance frequency of that interface. The peak frequency therefore constrains the depth and shear-wave velocity of the principal velocity contrast, commonly marking the base of the sedimentary layer. The study area spans the Nankai Trough—the convergent margin where the Philippine Sea Plate subducts beneath the Eurasian Plate. OBS stations are arranged along four trench-perpendicular profiles that cross the forearc basins (eastern Kumano and Muroto), the continental slope, and the trench, plus one trench-parallel profile intersecting both forearc basins. Distinct HVSR peaks are identified at 135 stations; lower peak frequencies indicate deeper interfaces, whereas larger HVSR amplitudes imply stronger impedance contrasts. Using the peak frequency and its corresponding amplitude, we perform k-means clustering and classify the observations into seven groups. The trench sector exhibits low-frequency (0.6–2 Hz) high-amplitude HVSR peaks that correspond to a 70–220 m-thick low-velocity sediment layer and can be traced to an impedance associated with a silica phase transition. In the Kumano forearc basin, multi-peak HVSR spectra show frequencies that decrease landward toward the basin center. Along the continental slope, HVSR peaks are characterized by high frequency (> 6 Hz) and low amplitude, indicating a thin sediment cover or exposed basement. Representative trench and forearc basin sites with clear, continuous peaks are inverted by Monte-Carlo modeling of the full HVSR curve, yielding shallow shear-wave velocity models; comparison with previous seismic and borehole studies confirms that the derived Vp/Vs ratio model delineates sedimentary units and low-velocity zones. Over the three-month observation, peak frequencies remain nearly constant—apart from minor increases attributable to initial instrument compaction and several anthropogenic noise—and their spatial pattern systematically correlates with structural domains such as the trench, continental slope, and forearc basins. Leveraging long-duration, wide-aperture ambient-noise data, we show that the spatiotemporal distribution of peak frequencies exhibits a strong correlation between deep-marine HVSR responses and underlying geological structures. Building on this relationship, we construct a shallow shear-wave impedance map that quantitatively constrains the physical properties of the upper Nankai subduction zone.