海洋地球物理調查常依賴聲學觀測來推估海床形貌、淺部沉積構造與動態地質過程。然而，在深海環境中，直接量測常受限於儀器佈放、資料覆蓋範圍與重複測量成本。因此，如何有效利用既有主動式與被動式聲學資料，發展可用於海床環境變化與海底儀器狀態監測的方法，是重要的研究課題。本研究整合單音束測深資料、控制震源空氣槍實驗資料與船舶輻射噪音，探討不同聲學觀測在海床氣體滲漏、海底地形變化、聲波多重路徑傳播，以及海底地震儀幾何狀態分析上的應用。利用歷史單音束測深資料分析臺灣東北部南沖繩海槽海床氣體噴發特徵。雖然單音束資料受限於二維航跡取樣，但其長期累積的觀測紀錄可提供重要的歷史資訊。透過重新處理聲學回波資料，辨識出水柱中與氣體噴發相關的異常聲學反應，並探討其空間分布與海床流體活動之間的關係，顯示既有單音束資料經適當處理後，可作為重建過去海床流體活動的重要資料來源。其次，使用臺灣西南外海 Pointer Ridge 區域的環形空氣槍震測資料，建立受控幾何條件下的多重路徑聲波分析架構。環形震源幾何提供近似均勻的方位角覆蓋，使海底地震儀可在相近震源距離下接收不同方向的聲波訊號。透過 cepstrum 分析，萃取直達波與海床反射相關路徑之間的走時差，並結合 BELLHOP 聲線追跡與幾何聲線計算，比較觀測值與理論模型之間的差異。結果顯示，多重路徑走時差不僅受反射點水深控制，也受到海床坡度與局部地形起伏影響。將此方法應用於臺灣東部外海 MACHO 海底纜線觀測系統之 EOS4 測站，探討 2018 年花蓮地震序列前後海底地震儀狀態與周圍海床環境可能的變化。研究中結合 AIS 船舶軌跡、船舶輻射噪音、cepstrum 分析、奇異值分解濾波與 H/Z 頻譜比值分析，從被動聲源中萃取多重路徑走時差。結果指出走時差殘差在事件前後皆呈現西北側為正、東南側為負的大尺度空間分布，該特徵主要源自原始測站座標偏差，而非地震造成的海床變化。將測站位置向西約 300 公尺、向北約 600 公尺的試驗位置重新計算後，大尺度殘差梯度明顯降低；剩餘的局部異常主要分布於海底水道與斜坡區域，可能反映局部水深、坡度與反射幾何差異。事件後壓力資料顯示等效水深增加約4公尺，部分頻帶的水平與垂直分量頻譜比下降，並伴隨感測器方位旋轉，顯示儀器可能發生輕微姿態調整、淺層掩埋或與沉積物耦合狀態改變，但未發現明確的大尺度同震水平位移。本研究展示多重路徑聲學訊號可作為連結主動式與被動式海洋聲學觀測的有效工具。受控空氣槍實驗可建立多重路徑走時差對海床地形與坡度的敏感性，而船舶輻射噪音則提供長時間、廣空間覆蓋的被動聲源，有助於監測海底地震儀狀態與海床環境變化。;Marine geophysical investigations commonly rely on acoustic and seismic observations to characterize seafloor environments， shallow subsurface structures， and dynamic geological processes. In deep-ocean settings， however， direct observations are often limited by survey coverage， deployment geometry， and the cost of repeated measurements. Existing acoustic and seismic datasets may therefore contain environmental information beyond their original survey objectives. This study develops a multi-scale signal-analysis framework that integrates historical single-beam echo sounder records， controlled air-gun experiments， passive ship-radiated noise， and ocean-bottom seismic and pressure observations to characterize seafloor gas seepage， bathymetric effects on acoustic propagation， and changes in Ocean Bottom Seismometer (OBS) geometry and coupling offshore Taiwan. First， historical single-beam echo sounder data from the southernmost Okinawa Trough were reprocessed to investigate water-column acoustic anomalies associated with seafloor gas seepage. Although single-beam echo sounder data are limited by their two-dimensional acquisition geometry， their long-term availability provides valuable historical observations. By reprocessing acoustic backscatter records， this study identifies anomalous water-column acoustic responses associated with gas seepage and examines their spatial relationship with seabed fluid activity. These results demonstrate that historical single-beam echo sounder data， when properly processed， can provide useful constraints on past seabed fluid activity. Second， controlled circular air-gun data from the Pointer Ridge area offshore southwestern Taiwan were used to establish a multipath acoustic analysis framework under well-constrained source geometry. The circular source configuration provides quasi-azimuthally uniform coverage around the OBS， allowing acoustic signals from different azimuths to be recorded at similar source–receiver distances. Cepstral analysis was applied to extract travel-time differences between the direct arrival and seafloor-related multipath arrivals. The observed travel-time differences were then compared with theoretical predictions from BELLHOP ray tracing and a custom geometric ray-based calculation. The results show that multipath travel-time differences are controlled not only by the water depth at the reflection point， but also by seafloor slope and local bathymetric variability. When the actual reflection geometry differs from a simplified flat-bottom assumption， systematic travel-time deviations occur， indicating that seafloor morphology plays a critical role in the interpretation of multipath acoustic observations. Finally， the framework was applied to passive ship-radiated noise recorded at the EOS4 station of the MACHO submarine cable observatory before and after the 2018 Hualien earthquake sequence. AIS vessel positions， cepstral analysis， Singular Value Decomposition filtering， pressure observations， and H/Z spectral ratios were jointly analyzed. The broad spatial pattern of Δτ was present in both the pre-event and post-event datasets， indicating that its distribution was primarily influenced by a bias in the original OBS coordinates rather than coseismic seafloor change. Recalculation using a trial position shifted approximately 300 m westward and 600 m northward substantially reduced this large-scale residual pattern. Localized residuals remained near channel and slope regions， where variations in bathymetry and reflection geometry likely affected the multipath travel times. Pressure， sensor-rotation， and H/Z observations further suggest minor instrument adjustment， shallow burial， or changes in sediment coupling， although no clear evidence of large coseismic horizontal displacement was identified. Overall， this study demonstrates that multi-scale acoustic and seismic observations can provide complementary constraints on gas seepage， seafloor morphology， multipath propagation， and instrument–seafloor interaction. Controlled air-gun experiments clarify the effects of bathymetry and slope on multipath travel-time differences， whereas passive ship noise enables retrospective monitoring without dedicated active-source surveys. The results also highlight the importance of accurate instrument coordinates， source positions， and bathymetric models when interpreting multipath acoustic signals.
