Analysis Of Backscatter To Extraction Of Shoreline Using Machine Learning Methods In The Bangkalan Regency - Support Vector Machine

Aug 3, 2026

Upload by Miles

Coastal zones face instability and shoreline shifts driven by natural processes and human activities, requiring reliable monitoring to support sustainable management. This research uses Synthetic Aperture Radar (SAR) data to extract coastlines by segmenting the scene and applying a Support Vector Machine (SVM) classifier to separate water and land classes. Model performance is evaluated through accuracy and Kappa coefficient, yielding overall accuracies of 99.5% (2016) and 99% (2023).

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Coastal change and monitoring needs

SAR data and Sentinel-1 overview

Machine learning approach with SVM

Materials and Method

Research location

Data collection

Pre-processing SAR image

Classification with SVM

Analysis Of Backscatter To Extraction Of Shoreline Using Machine Learning Methods In The Bangkalan Regency Fahmi Arifin1, Ashari Wicaksono1,2* 1 Department of Marine Science, University of Trunojoyo Madura, Jl. Raya Telang PO BOX 2 KamalBangkalan, East Java, Indonesia 2 Laboratory of Oceanography, University of Trunojoyo Madura, Jl. Raya Telang PO BOX 2 KamalBangkalan, East Java, Indonesia Abstrak. Coastal areas are often threatened by natural and anthropogenic factors, causing instability and shoreline changes in the affected areas. Shoreline changes can be monitored with remote sensing techniques such as Synthetic Aperture Radar (SAR) data. The purpose of this research is to extract the coastline by segmenting the machine learning method and find out how far the machine learning model works to distinguish the water class and the land class. The method used in this research is the Support Vector Machine model to divide the water and land classes that will be utilized to obtain shoreline extracts from the model results, and evaluate the model by calculating the model accuracy. The overall accuracy results recorded in 2016 and 2023 are 99.5% and 99%, respectively, with Kappa Coefficients of 0.99018 and 0.98138. This study highlights the potential of SAR data and SVM methods in monitoring coastal dynamics and can serve as a reference for sustainable coastal management. 1. Introduction Coastal areas play an important role in supporting biodiversity and human activities, which also affect the economic development of the surrounding area [1] . Coastal areas are often threatened by natural and anthropogenic factors, causing instability and shoreline changes in the affected areas [2] . Shorelines can be affected by natural activities, such as tides, erosion, sedimentation and sea level rise, which require accurate monitoring to reduce negative impacts [3] . The coastline in Bangkalan Regency has experienced significant changes in recent years. This can occur due to natural processes and human activities, that are quite often carried out in coastal areas [20] . Shoreline changes can be monitored with remote sensing techniques, such as Synthetic Aperture Radar (SAR) data [4] . SAR is a radar system that records the earth's surface through electromagnetic wave signals [10] . SAR is an active sensor that emits energy radiation to get reflections from the object [5] . SAR has advantages over other optical images because it isnot affected by clouds, weather, or time (day and night) . These advantages can be utilized to *Corresponding author: [ashari.wicaksono@trunojoyo.ac.id](ashari.wicaksono@trunojoyo.ac.id) © The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 ([https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/)). extract the coastline without worrying about problems in other optical images [6] . SAR, especially Sentinel-1, has a high spatial resolution of up to 10 meters, which is good enough to be used for coastline extraction. Sentinel-1 has a wide global coverage with a period of 6- 12 days [7]. The frequency emitted by Sentinel-1 is around 5.405GHz, with a wavelength of 5.5 cm [8] . The use of SAR data can be utilized as coastline monitoring by extracting the coastline using several methods. The method that can be used is machine learning, one of which is the Support Vector Machine (SVM) method [9] . The utilization of this SVM method has been carried out in many studies, one of which is to identify Land Cover in an image. This research will utilize SVM to extract coastlines by separating SAR data from two different classes, for example water areas and land areas [9]. The purpose of this research is to extract the coastline by segmenting it using a machine learning method and to determine how well the machine learnig model works to distiguwish the water class from the land

FAQs

What is the main goal of this study?

To extract the coastline in Bangkalan Regency using SAR data and a machine learning approach, specifically by separating water and land classes with an SVM model.

Why is Sentinel-1 SAR suitable for shoreline extraction?

SAR enables coastline monitoring with advantages over optical imagery, since it is not affected by clouds, weather, or day-night conditions, and Sentinel-1 provides sufficient spatial resolution.

How is the SVM model evaluated in the research?

The model accuracy is calculated and reported along with Kappa coefficients, with overall accuracies reaching 99.5% in 2016 and 99% in 2023.

Aug 3, 2026

Upload by Miles