Open Access

ARTICLE

Seabed Classification from Multi-Frequency Multibeam Data: A Study from Selorejo, Malang, Indonesia

Qaisara Yusriena Yusaini¹, Muhammad Abdul Hakim Muhamad^1,*, Raiz Razali^1,*, Rozaimi Che Hasan², Mohd Shahmy Mohd Said¹, Mohd Zainee Mohd Zainal^1,2, Ikhsan Nuradi³

1 School of Geomatic Science and Natural Resources, College of Built Environment, Universiti Teknologi MARA, Shah Alam, 40450, Malaysia
2 Faculty of Artificial Intelligence, UTM Kuala Lumpur, Level 7, Razak Tower, Jalan Sultan Yahya Petra, Kuala Lumpur, 54100, Malaysia
3 PT Sonar Nusantara Utama, Jl. Ciledug Raya No. 74B, Cipulir, Kec. Kby. Lama, Daerah Khusus Ibukota Jakarta, Jakarta, 12230, Indonesia

* Corresponding Authors: Muhammad Abdul Hakim Muhamad. Email: ; Raiz Razali. Email:

(This article belongs to the Special Issue: Advancements in Geospatial Methods and Technologies for Sustainable Built Environment and Engineering)

Revue Internationale de Géomatique 2025, 34, 535-552. https://doi.org/10.32604/rig.2025.065284

Received 08 March 2025; Accepted 30 June 2025; Issue published 06 August 2025

Abstract

Sediment mapping is a crucial component of environmental science, particularly in the marine environment, where the analysis of seabed sediments is essential for various purposes, including marine resource management, habitat preservation, and infrastructure development. Sediment refers to the solid particles that are transported and deposited in different areas. Multibeam echosounders have revolutionized the field of seabed sediment mapping by providing unparalleled resolution and accuracy in seafloor surveys. This study aimed to produce sediment maps by implementing multi-frequency, e.g., 200, 400, 550, and 700 kHz multibeam data using a machine learning algorithm, e.g., Support Vector Machine (SVM) and Random Forest (RF) approach. The study area of this research was located at Waduk Selorejo, Kecamatan Ngantang, Kabupaten Malang, Indonesia. To achieve the aim, bathymetry and backscatter mosaic with multi-frequency results are produced. Then, the second derivatives from bathymetry and backscatter mosaic are produced (slope, ruggedness, aspect, curvature, mean, homogeneity, entropy, correlation, phi, and characterization) were used as the predictors. All the predictors with multi-frequency each have 0.5 m of spatial resolution. To classify, Pearson’s correlation coefficient method is used to identify strong and weak correlations. The weak correlation was used to run the classification by using SVM and RF. The Angular Range Analysis (ARA) characterization is used as the ground truth data. The accuracy assessment of the multiple-frequency sediment map will be compared with the two algorithms of machine learning (e.g., SVM and RF). This study highlights the importance of sediment mapping and the potential of acoustic methods and machine learning algorithms for seafloor classification.

---

Keywords

---