Seafloor mapping using multibeam sonar
Seafloor habitat and its marine community have greatly affected by anthropogenic pressures from various human activities. Efforts to conserve and manage the marine habitat are challenging due to the difficulty to get the details of the seafloor data. Attention has been focused towards the multibeam...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2020
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/98285/1/ShahrinAmizulSamsudinMRAZAK2020.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Seafloor habitat and its marine community have greatly affected by anthropogenic pressures from various human activities. Efforts to conserve and manage the marine habitat are challenging due to the difficulty to get the details of the seafloor data. Attention has been focused towards the multibeam echo sounder system (MBES), a tool in mapping the seafloor habitats, due to its ability to produce a detailed seafloor map. The aim of this study is to utilize MBES output, namely the bathymetry, backscatter, and its derivatives in order to produce a seafloor habitat map using automated classification technique in Malaysian water. The objectives are: (i) to investigate the correlation between MBES backscatter image and signal-based method for seafloor sediment classification; (ii) to evaluate the importance of bathymetry and its derivatives in producing coral reef classification map; (iii) to perform automated technique in producing the coral reef classification map, and finally (iv) to assess the accuracy of the coral reef classification maps constructed from the techniques above. The study was conducted in two different locations: Sembilan Island, Perak and Tawau, Sabah. The results of the data reduction analysis using the Principal Component Analysis (PCA), Linear Pearson Correlation, and variable importance analysis showed four most significant derivative layers for the production of coral reef classification map were identified: (i) bathymetry, (ii) benthic position index (BPI), (iii) slope, and (iv) grey level co-occurrence matrices (GLCM) mean. The classification map constructed with the selected MBES derivatives using four different techniques (Support Vector Machine, Neural Network, QUEST decision trees, and CRUISE decision trees) had shown an encouraging results with two classifiers achieved the accuracy of more than 70% (Support Vector Machine with 73.61% and Neural Network with 70.14%). In sum, this classification seafloor habitat map has enhanced coral reef spatial distribution information, and this finding has an important contribution to the seafloor habitat mapping in Malaysia. |
|---|