Scholarlink

<p>Bathymetry is one of the important parameters of the coastal environment. Conventional bathymetry measurements require large resources (humans, time and costs). Technological developments and remote sensing satellite imagery can be used to estimate coastal bathymetry, known as Remote Sensing-Derived Coastal Bathymetry and overcome several weaknesses of conventional measurements. This research aims to estimate coastal bathymetry from Sentinel 2 imagery using random forest (RF) and gradient tree boost (GTB) algorithms based on machine learning techniques. The novelty of this research lies in the combination of a specific regional application, optimized methodological configurations for Sentinel-2 data, and the incorporation of additional indices beyond the original Sentinel-2 bands to enhance bathymetry estimation in turbid estuarine waters affected by suspended sediment, leveraging the data and computational capabilities of Google Earth Engine for improved operational efficiency. Based on the results of this study, the Normalized Root Mean Square Error value for estimated bathymetry is 9.79% for GTB and 9.83% for RF when compared with the measured bathymetry results. This method makes it possible to monitor and map shallow water areas more effectively and efficiently, especially in estuary areas, however, this method is very dependent on the availability of good quality satellite imagery and at the same time as bathymetric measurements in the field. In the future, this study can be continued by utilizing other deep learning architectures.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of Sentinel-2 Imagery and Machine Learning for Predicting Coastal Bathymetry in the Cisadane Estuary, Indonesia

  • 期刊论文

  • 发表日期：

    2025年7月8日

查看全文
  • Hilmi Aziz,
  • Fajar Yulianto,
  • Mardi Wibowo,
  • Dhedy Husada Fadjar Perdana,
  • Amalia Nurwijayanti,
  • Imam Fachrudin

摘要

Bathymetry is one of the important parameters of the coastal environment. Conventional bathymetry measurements require large resources (humans, time and costs). Technological developments and remote sensing satellite imagery can be used to estimate coastal bathymetry, known as Remote Sensing-Derived Coastal Bathymetry and overcome several weaknesses of conventional measurements. This research aims to estimate coastal bathymetry from Sentinel 2 imagery using random forest (RF) and gradient tree boost (GTB) algorithms based on machine learning techniques. The novelty of this research lies in the combination of a specific regional application, optimized methodological configurations for Sentinel-2 data, and the incorporation of additional indices beyond the original Sentinel-2 bands to enhance bathymetry estimation in turbid estuarine waters affected by suspended sediment, leveraging the data and computational capabilities of Google Earth Engine for improved operational efficiency. Based on the results of this study, the Normalized Root Mean Square Error value for estimated bathymetry is 9.79% for GTB and 9.83% for RF when compared with the measured bathymetry results. This method makes it possible to monitor and map shallow water areas more effectively and efficiently, especially in estuary areas, however, this method is very dependent on the availability of good quality satellite imagery and at the same time as bathymetric measurements in the field. In the future, this study can be continued by utilizing other deep learning architectures.