Research On Acoustic Seafloor Classification Based On Deep-sea Multibeam Data

Seafloor polymetallic sulfides in hydrothermal fields of mid-ocean ridges are enriched with high-grade metals and rare metals,which will be important potential resources.As geological caps play an important role in the formation of SMS in the mid-ocean ridge,the study on the condition of seafloor substrate in the mid-ocean ridge is very meaningful.Acoustic seafloor classification based on multibeam data is a leading-edge field of hydrographic surveying and charting,which is a hot but difficult problem both in domestic and abroad.It has been successfully applied in shallow water,but may not work well in deep sea,particularly in the mid-ocean ridges that with complex seafloor topography and substrates.In this paper,the acoustic seafloor classification method based on deep-sea multibeam data has been systematically studied.An inversion method based on backscatter angular response curves to obtain the characteristic parameters was proposed and the intensity information in the backscatter image was also statistically analyzed according to different sediment coverage.They have been applied to a complex terrain area around Duanqiao hydrothermal field on the SWIR at 50.47°E.The main results are as follows:1.Based on the study of seafloor backscattering characteristics,the numerical simulation of backscatter angular response curves by using Jackson backscatter model has been carried out.Thus,the relationship between the seafloor backscatter strength(BS)and sonar frequency,grazing angle,seafloor characteristics has been analyzed and summarized.2.In-depth analysis of various factors affecting the quality of multibeam data was conducted,and research emphasizing on the method of removing jump points from bathymetry data and the correction of seafloor incidence angle for BS data were implemented.The program for the complete workflow of multibeam data was developed,and the multibeam data of the study area were processed to obtain highresolution topographic maps of the seafloor and backscattered grayscale maps,as well as high-quality intrinsic BS values of the study area.3.The feasibility of applying deep learning techniques to the automated substrate classification of seafloor camera data was explored.With the manual classification information of seafloor camera data from seven survey lines,a convolutional neural network model was built and trained,and finally achieved a fairly accurate autonomous determination of seafloor substrate types.This technique can greatly improve the efficiency of substrate classification for other seafloor camera data in later voyages,and provide sufficient priori information and verification data for multi-beam acoustic substrate classification.It also provides a reference for other automated data processing of deep-sea dataset.4.Proposed an inversion method of characteristic parameters based on deep-sea multibeam backscatter angular response curves.Firstly,multibeam BS data were grouped according to the vertical direction of the ship track,and the backscatter angular response curves were extracted from each grouped unit.Then the angular response curves were inverted based on Lambert functional model to obtain the characteristic parameters related to seafloor substrate types and the standard deviations of inversion results.Finally,combined with the ground truth information,the characteristic parameters and the standard deviations were statistically analyzed,and six substrate types are defined according to their value ranges.Thus,the substrate classification map of Duanqiao hydrothermal zone is drawn.The validity of this method was demonstrated.5.Based on the backscatter image of the study area,the statistical analysis of seafloor sediment coverage was carried out.Firstly,typical areas with different sediment coverage were identified,and the typical areas were grouped according to the type and the number of the known substrate types in the area.Then the BS values corresponding to the seafloor photo-based classification points in the typical areas were acquired,and the relationship between the two was also statistically analyzed to obtain the association between BS values and sediment coverage.Finally,the distribution of seafloor sediment coverage in the study area was mapped by using this association.6.Combining the topographic and geomorphic features of the seafloor in the study area,the distribution and genesis of three types of seafloor substrates,sediments,basalts and breccia were analyzed based on the photo-classification pointset,the seafloor classification map of Duanqiao hydrothermal zone,and the sediment coverage distribution map of the study area that obtained in this paper.The research results in this paper will serve for the implementation of China’s exploration contract for polymetallic sulfide resources in Southwest Indian ridge,and contribute to the seafloor substrate interpretation in other deep-sea regions.