Scholarlink

<p>Monitoring coastal water quality is crucial for protecting marine ecosystems and ensuring sustainable use of aquatic resources. This study presents an approach based on integrated remote sensing and machine learning to analyze the spatial distribution of relevant water quality indicators, Chlorophyll-a, turbidity (in Nephelometric Turbidity Units, NTU), and dissolved oxygen (DO), for the Bay of La Paz, Mexico. Quantitative data were obtained from Sentinel-2 satellite imagery from 2019 to 2024 to establish long-term trends and spatial variability and thereby to visualize maps properly and study water quality indicators more reliably and at scale and resolution. Further, in-situ sampling was carried out on March 10, 2021, providing direct field observations for model calibration and validation. This study applied XGBoost, SVR, MLP, and MDN machine learning techniques to each water quality indicator. The results had significant variations throughout the bay, with turbidity having an unusually high value of 97 NTU in some near-shore and estuarine boundaries; turbidity values were characteristically high due to resuspension of sediment and potential coastal run-off and tidal action. DO levels and chlorophyll-a concentrations were steady and within the mid-ecological range, suggesting that phytoplankton productivity and oxygen availability in the bay were in balance during this event. All things considered, machine learning-based models have demonstrated a great capacity to forecast and categorize water quality attributes, providing a scalable and affordable substitute for traditional field-based techniques. An integrated approach like this improves the ability to monitor and respond to ecological changes in near real time, which can facilitate better decisions for coastal management and policy formation. Sophisticated monitoring techniques with high ecological and economic value were successfully shown in the Bay of La Paz. By offering long-term insights on patterns in water quality, these techniques helped to preserve biodiversity. Our framework can be adapted and applied to other coastal environments under similar anthropogenic and climatic stressors in the future.</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

Monitoring Coastal Water Quality Using Remote Sensing and Machine Learning

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • K. Ambika,
  • P. Shanmugapriya,
  • Abdulaziz G. Alghamdi,
  • Hadeel Alsolai

摘要

Monitoring coastal water quality is crucial for protecting marine ecosystems and ensuring sustainable use of aquatic resources. This study presents an approach based on integrated remote sensing and machine learning to analyze the spatial distribution of relevant water quality indicators, Chlorophyll-a, turbidity (in Nephelometric Turbidity Units, NTU), and dissolved oxygen (DO), for the Bay of La Paz, Mexico. Quantitative data were obtained from Sentinel-2 satellite imagery from 2019 to 2024 to establish long-term trends and spatial variability and thereby to visualize maps properly and study water quality indicators more reliably and at scale and resolution. Further, in-situ sampling was carried out on March 10, 2021, providing direct field observations for model calibration and validation. This study applied XGBoost, SVR, MLP, and MDN machine learning techniques to each water quality indicator. The results had significant variations throughout the bay, with turbidity having an unusually high value of 97 NTU in some near-shore and estuarine boundaries; turbidity values were characteristically high due to resuspension of sediment and potential coastal run-off and tidal action. DO levels and chlorophyll-a concentrations were steady and within the mid-ecological range, suggesting that phytoplankton productivity and oxygen availability in the bay were in balance during this event. All things considered, machine learning-based models have demonstrated a great capacity to forecast and categorize water quality attributes, providing a scalable and affordable substitute for traditional field-based techniques. An integrated approach like this improves the ability to monitor and respond to ecological changes in near real time, which can facilitate better decisions for coastal management and policy formation. Sophisticated monitoring techniques with high ecological and economic value were successfully shown in the Bay of La Paz. By offering long-term insights on patterns in water quality, these techniques helped to preserve biodiversity. Our framework can be adapted and applied to other coastal environments under similar anthropogenic and climatic stressors in the future.