鯨豚聲訊密度估算法探討－以臺灣西部海域白海豚為例

Abstract

在全球能源轉型與淨零碳排的趨勢下，臺灣積極推動離岸風力發電，然而部分臺灣西部海域作為離岸風電的開發區域鄰近中華白海豚（Sousa Chinensis）的重要棲息環境，如何在推動綠色能源發展的同時兼顧海洋生態保育，成為一項關鍵議題。本研究旨在應用被動式聲學監測（Passive Acoustic Monitoring, PAM）技術，利用佈放於彰化外海的聲學測站資料，計算該區域的鯨豚聲訊密度（Acoustic Density），並深入分析影響其變化的關鍵參數，為未來制定精確的鯨豚保育策略提供科學依據。 本研究利用了兩筆水下聲學數據，第一筆來自於2019年至2021年雲林麥寮港附近裝設的二個被動式水下聲學測站，並結合目視團隊的目視調查紀錄，計算後續分析所使用的鯨豚哨叫聲平均發聲率（r ̂）；第二筆資料為聲訊密度的主要計算資料，分析彰化外海2023年5月17日至5月30日期間，五個被動式聲學測站（UN1～UN5）共336小時的連續監測資料。透過密度估計公式，計算了各彰化外海測站每小時的聲訊密度值。結果顯示，聲訊密度在時空分佈上呈現顯著差異。例如，UN1測站於2023年5月17日5時的密度值達到1023.23隻/平方公里，而相近時段UN4測站偵測到多達801聲哨叫聲，其密度值卻僅為147.97隻/平方公里。此一結果顯示，哨叫聲數量並非影響聲訊密度值的唯一因素。 對聲訊密度的計算參數進行分析，在單一測站的聲訊密度變化中，哨叫聲數量是主要的影響參數，其對密度值具顯著正相關性，在跨測站比較時，偵測範圍的差異也成為影響聲訊密度值的重要參數。這種差異主要源於各測站所在環境因子的不同，包括噪音強度和音傳損耗。在五個測站間，噪音強度的差異不多，音傳損耗的計算頻率差異也不大，但UN1測站因其周圍海域噪音強度較高且底質為泥，可能因此導致其音傳損耗增加，偵測範圍小於其他測站，從而使其聲訊密度值被放大。此外，研究亦發現，哨叫聲發聲率的微小變動，會對聲訊密度計算結果產生巨大影響，凸顯了獲取準確本土化發聲率的重要性。 本研究提供了利用PAM資料計算聲訊密度的方法，其結果在時間與空間上均具備獨特的參考價值，此方法對海洋哺乳類動物監測與保育工作提供了初步的科學依據，未來若能擴大監測網絡、針對不同的時間與地點建立聲學資料庫，並結合目視調查資料進行驗證，將有助於提高聲訊密度估算的準確性，為實現經濟發展與生態永續的雙贏局面提供堅實的科學基礎。

In response to global energy transition and net-zero carbon emissions, Taiwan is actively promoting offshore wind power. However, development in the western Taiwan Strait encroaches upon the critical habitats of the Indo-Pacific humpback dolphin (Sousa Chinensis). This study addresses the challenge of balancing green energy with marine conservation by applying Passive Acoustic Monitoring (PAM) technology. We utilized data from acoustic stations off the coast of Changhua to calculate cetacean Acoustic Density and analyze key influencing parameters, providing a scientific basis for conservation strategies. The research uses two underwater acoustic datasets. The first, collected from 2019 to 2021 near Yunlin's Mailiao Port, combined with visual survey records, was used to calculate the average dolphin whistle production rate. The second, and primary, dataset consists of 336 hours of continuous monitoring from five PAM stations (UN1-UN5) off the coast of Changhua from May 17 to May 30, 2023. Using a density estimation formula, we calculated the hourly acoustic density for each station. Results revealed significant spatiotemporal variations. For example, UN1's density reached 1023.23 individuals/km² at 5:00 on May 17, 2023, while UN4, with 801 detected whistles in a similar period, yielded only 147.97 individuals/km². This indicates that the number of detected whistles is not the sole factor influencing acoustic density. Our in-depth analysis of the density calculation parameters reveals that while whistle count is the primary influencing factor within a single station, the detection area is the dominant factor in cross-station comparisons. This is due to local environmental variables, such as ambient noise and seabed composition, which affect transmission loss. Specifically, station UN1's detection area was smaller than others, likely due to higher ambient noise and a muddy seabed, which increased acoustic transmission loss and inflated the calculated density. The study also highlights that even minor variations in the whistle production rate can significantly impact the density results, underscoring the need for accurate, localized data. In conclusion, this research provides a robust method for calculating acoustic density using PAM data, offering valuable spatiotemporal insights for marine mammal conservation. Future efforts should focus on expanding the monitoring network and establishing a localized acoustic database to improve estimation accuracy and achieve a sustainable balance between development and preservation.