應用水下被動式聲學監測定位鯨豚活動

Abstract

本研究於雲林麥寮海域及彰濱海域分別佈設三組水下聲學記錄器，蒐集鄰近白海豚重要棲地之水下聲訊並進行分析，並透過海上調查來了解該海域之鯨豚種類。本研究同時開發基於到達時間差(Time Difference of Arrival, TDOA)結合粒子群演算法(Particle Swarm Optimization, PSO)之聲源定位方法，以提升定位精度及計算效率。 在定位演算法驗證方面，隨機選取36個聲源點位，利用Bellhop聲線計算模組模擬傳播至測站之接收訊號，再以開發之演算法進行定位推估。結果顯示整體之定位誤差約為16公尺，且隨著距離量測中心越遠，其誤差值也隨之上升。在鯨豚聲學資料分析方面，雲林麥寮海域哨叫聲主要集中於2023年11月19日至25日之間，其中僅S_A測站與其它測站間哨叫聲偵測數據具有顯著差異，且S_A測站哨叫聲聲紋偵測數量顯著低於其餘兩測站。該海域之哨叫聲簡單型聲紋為主，其中以平坦型及下降型居多，惟S_A測站偵測數量較少，其聲紋中以簡單型與複雜型比例各半。在彰濱海域海上調查確認於調查期間以瓶鼻海豚出沒為主，在哨叫聲分析顯示社交活躍度近岸高於離岸，且活動呈現間歇性特徵。該區域聲紋亦為簡單型為主，並以下降型聲紋居多，其聲紋頻率高於雲林海域之統計結果。 目前在鯨豚定位結果上，惟受限於多數聲紋僅由單一或兩個測站偵測，致可有效定位之聲紋有限，尚難重建鯨豚活動路徑。建議後續可增加被動式聲學監測(Passive Acoustic Monitoring, PAM)測站數量，或透過喀答聲進行鯨豚定位與活動軌跡分析。

This study deployed three sets of underwater acoustic recorders in the Mailiao sea area of Yunlin County and the Changbin sea area of Changhua County to collect and analyze underwater acoustic signals from key habitats of the Chinese white dolphin (Sousa chinensis taiwanensis). Concurrently, boat-based visual surveys were conducted to identify cetacean species present in these areas. A novel sound source localization method was developed based on Time Difference of Arrival (TDOA) combined with Particle Swarm Optimization (PSO) to enhance localization accuracy and computational efficiency. For algorithm validation, 36 sound source points were randomly selected, and the received signals at monitoring stations were simulated using the Bellhop ray-tracing model. The developed algorithm was then applied for localization estimation. Results indicated an overall localization error of approximately 16 meters, with errors increasing as the source location moved farther from the measurement center. In the analysis of cetacean vocalization data, whistle vocalizations in the Mailiao sea area were predominantly recorded between November 19 and 25, 2023. Significant differences in whistle detection counts were observed only between station S_A and other stations, with S_A exhibiting significantly fewer whistle detections. The whistle repertoire in this area was primarily simple types, mainly flat and descending contours, while S_A showed a more balanced proportion of simple and complex whistles. Boat-based visual surveys in the Changbin sea area confirmed the predominance of bottlenose dolphins (Tursiops truncatus) during the survey period. Analysis of vocalization data indicated higher social activity nearshore compared to offshore, with vocal behavior exhibiting intermittent patterns. The whistle types were mainly simple, with descending contours prevailing, and the whistle frequencies were statistically higher than those recorded in the Mailiao sea area. Currently, due to most whistles being detected by only one or two stations, the number of effectively localizable whistles remains limited, making it difficult to reconstruct cetacean movement paths. It is recommended that future studies increase the number of Passive Acoustic Monitoring (PAM) stations or incorporate click trains for cetacean localization and movement trajectory analysis.