基于高精度声学图像的海洋牧场海床类型识别

人工鱼礁作为海洋牧场建设的基础生态工程，在海洋生态环境修复和渔业资源增殖方面起着重要作用。随着人工鱼礁的大量投放，海洋牧场海床逐步转变为高度异质性的底栖生境，底栖生态变化会直接影响到底栖及底层生物的栖息与繁衍。针对传统调查方法在海洋牧场中的局限性，聚焦天津海洋牧场，基于侧扫声纳后散射数据和多波束测深数据，从侧扫图像亮度、图像纹理及地形起伏度等3个维度，探索了天津海洋牧场海床类型识别、分类和面积量化的新方法。研究表明，天津海洋牧场共发现人工鱼礁、牡蛎类、淤泥类、蛤仔泥混类、牡蛎泥混类及碎贝泥混类等6种海床类型，基于声学图像特征的海床类型识别方法准确率达90%，可满足海洋牧场底栖生态研究的需求；通过对比分析，天津海洋牧场本底淤泥类海床消退为58.19%，贝类海床增至7.96%，贝类资源增殖效果明显；人工鱼礁周边新生牡蛎礁0.246 km2，并伴有0.303 km2的牡蛎泥混海床待发育为牡蛎礁，海洋牧场建设对牡蛎礁具有重要的修复作用。本研究探索了适宜于海洋牧场的海床声学分类方法，探讨了声学技术对传统底栖调查方法的补充与完善作用，明确了海床类型与声学图像特征的紧密联系，验证了海洋牧场对牡蛎礁的修复作用，成果可有效应用于近岸海域生境修复、贝类资源量精准评估及牡蛎礁演化机理等领域的研究。

Research on seabed classification of marine ranching based on high-precision acoustic images

As the foundation for ecological engineering of marine ranching, artificial reefs play an important role in the restoration of marine ecological environment and the proliferation of fishery resources. With a large number of artificial reefs, the seabed of marine ranching has gradually transformed into highly heterogeneous benthic habitats, which directly affects the development and reproduction of bottom organisms. How to accurately identify seabed habitats has become an important issue in the benthic ecological research of marine ranching, and remote sensing of the seabed using acoustic technique combined or individually are providing solutions to this problem. Currently, the acoustic seabed classification technique has been applied to ecological research in many sea areas successfully, but it is rarely used in marine ranching. Also, the science of acoustic seabed classification is at its nascence, and there is still no widely accepted agreement about the standardization of instruments and methods. Focus on the application of acoustic technique in seabed classification of marine ranching, we developed a new method of seabed telemetry and identification based on side scan sonar (SSS) and multi-beam echo sounder (MBES), and applied it in Tianjin Marine Ranching. The SSS and MBES acoustic images objectively reflect the hardness and topography of seabed respectively, which closely relates to seabed type. But most seabed types are not directly measurable using remote survey techniques. It is always necessary to measure the seabed types using in situ ground truthing methods, which can be linked in some way to the characteristics of acoustic images, and then the seabed types can be identified and classified by using acoustic images. In our research, 70 seabed in-situ samples were analyzed and connected with the corresponding acoustic images, and the results show that key image features to identify the seabed type were brightness and texture of SSS image and topography derived from MBES image. Based on the believable relationship between seabed type and acoustic image characteristic, the seabed type of Tianjin Marine Ranching was identified, classified and quantified. The results show that 1) six seabed types were found in Tianjin Marine Ranching, including artificial reefs, oysters, mud, clam-mud, oyster-mud and shell-mud, and the accuracy of seabed classification method based on acoustic characteristics is 90%, which can meet the needs of benthic ecological research in marine ranching; 2) the mud seabed has subsided to 58.13%, and the shellfish seabed has increased to 7.96%, which demonstrates that the shellfish proliferated in Tianjin Marine Ranching; 3) 0.266km2 of new oyster reef are closely surrounding artificial reefs, and 0.313km2 of oyster-mud seabed will be developed into oyster reef, which verifies the restoration effect of Tianjin Marine Ranching on living oyster reef. Compared with conventional benthic survey methods, our method of acoustic seabed classification based on high-precision acoustic images can avoid sampling errors, reduce human disturbance, discover accurate position of shellfish and improve the accuracy of seabed classification. This study explored the acoustic classification method suitable for the seabed of marine ranching, discussed the complement and perfection of acoustic technique to traditional benthic survey methods, clarified the close relationship between marine ranching seabed types and acoustic image characteristics, and verified the restoration effect marine ranching on oyster reefs. Our research can be effectively applied to coastal habitat restoration, accurate assessment of shellfish resources, and the evolution mechanism of oyster reefs.