Scholarlink

<p>This work presents a transferable and distributed-ready computational framework for coastal water quality evaluation, aligned with high-performance and cluster-based data processing environments, namely Integrated Health Assessment Model (IWHA). IWHA is a data-driven framework leveraging machine learning to support coastal water quality evaluation. The proposed approach integrates open marine datasets, including variables like phosphate, dissolved oxygen, silicate and dissolved inorganic nitrogen. The proposed computational framework is divided into implementation layers, enabling the model to be transferable to different locations and time periods. Statistical analysis revealed strong relationships among nutrient variables, with high positive correlations between silicate and phosphate (<i>r</i> = 0.95) and between dissolved inorganic nitrogen and both phosphate and silicate (<i>r</i> = 0.91). Dissolved oxygen showed consistent negative correlations with dissolved inorganic nitrogen (<i>r</i> = − 0.72) and silicate (<i>r</i> = − 0.67), reflecting reduced oxygen availability under increased nutrient concentrations. Spatial analysis of Water Quality Index (WQI) values identified areas with persistently low water quality, particularly in coastal and island zones. Supervised machine learning techniques were applied to classify WQI categories using Principal Component Analysis (PCA) for dimensionality reduction. Among the evaluated models, Bagged Trees and Weighted K-Nearest Neighbors, achieved the most consistent performance, with macro-average F1 scores above 99%, retaining 95% of the variance, when six features were employed. The results demonstrate that IWHA provides a promising structure for integrating environmental data with supervised learning methods.</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

A transferable computational framework for integrated water health assessment in coastal ecosystems

  • 期刊论文

  • 发表日期：

    2026年7月16日

查看全文
  • Aika Miura,
  • Ali Ahmad,
  • Vitor Gabriel da Silva Ruffo,
  • Jaime Lloret

摘要

This work presents a transferable and distributed-ready computational framework for coastal water quality evaluation, aligned with high-performance and cluster-based data processing environments, namely Integrated Health Assessment Model (IWHA). IWHA is a data-driven framework leveraging machine learning to support coastal water quality evaluation. The proposed approach integrates open marine datasets, including variables like phosphate, dissolved oxygen, silicate and dissolved inorganic nitrogen. The proposed computational framework is divided into implementation layers, enabling the model to be transferable to different locations and time periods. Statistical analysis revealed strong relationships among nutrient variables, with high positive correlations between silicate and phosphate (r = 0.95) and between dissolved inorganic nitrogen and both phosphate and silicate (r = 0.91). Dissolved oxygen showed consistent negative correlations with dissolved inorganic nitrogen (r = − 0.72) and silicate (r = − 0.67), reflecting reduced oxygen availability under increased nutrient concentrations. Spatial analysis of Water Quality Index (WQI) values identified areas with persistently low water quality, particularly in coastal and island zones. Supervised machine learning techniques were applied to classify WQI categories using Principal Component Analysis (PCA) for dimensionality reduction. Among the evaluated models, Bagged Trees and Weighted K-Nearest Neighbors, achieved the most consistent performance, with macro-average F1 scores above 99%, retaining 95% of the variance, when six features were employed. The results demonstrate that IWHA provides a promising structure for integrating environmental data with supervised learning methods.