Scholarlink

<p>A water body’s quality is assessed based on restricted water samples. From a statistical perspective, there is inadequate confidence in determining the overall water quality and eutrophic status through point-based water sampling, typically constrained by physical and financial limitations. Therefore, this study focuses mainly on modeling the water quality through remotely sensed imagery, fuzzy logic (FL), and ground truth data. To monitor contaminants and evaluate surface water quality parameters (SWQPs) in the Canadian Saint John River (SJR), a novel Landsat 8-based-FL framework has been produced. Consequently, the most effective Landsat 8-based-FL models were developed to map concentrations of turbidity and total suspended solids (TSS), achieving coefficients of determination (R<sup>2</sup>) of 0.963 for turbidity and 0.942 for TSS. The models yielded root mean square errors (RMSE) of 0.857 nephelometric turbidity units (NTU) for turbidity and 0.415 milligrams per liter (mg/l) for TSS, with residual prediction deviations (RPD) of 2.979 and 3.711, respectively. Compared to regression methods, the FL modeling approach outperformed multiple regression results. Furthermore, using an independent validation dataset, the models achieved R² = 0.846 and RMSE = 1.254 NTU for turbidity, and R² = 0.809 and RMSE = 0.996&#xa0;mg/l for TSS, confirming the robustness of the developed Landsat 8-based-FL models. The results of this research demonstrated that it is possible to develop generalized Landsat 8-based-FL models that can accurately estimate concentrations of numerous SWQPs at various waterbodies. Ultimately, this research work is essential for decision-makers and local administrators, guiding them in taking appropriate actions at the opportune time.</p> Graphical Abstract <p></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 Fuzzy Logic to Satellite Reflectance for Precise Water Quality Management in the Saint John River, Canada: A Novel Approach

  • 期刊论文

  • 发表日期：

    2025年12月6日

查看全文
  • Essam Sharaf El Din

摘要

A water body’s quality is assessed based on restricted water samples. From a statistical perspective, there is inadequate confidence in determining the overall water quality and eutrophic status through point-based water sampling, typically constrained by physical and financial limitations. Therefore, this study focuses mainly on modeling the water quality through remotely sensed imagery, fuzzy logic (FL), and ground truth data. To monitor contaminants and evaluate surface water quality parameters (SWQPs) in the Canadian Saint John River (SJR), a novel Landsat 8-based-FL framework has been produced. Consequently, the most effective Landsat 8-based-FL models were developed to map concentrations of turbidity and total suspended solids (TSS), achieving coefficients of determination (R2) of 0.963 for turbidity and 0.942 for TSS. The models yielded root mean square errors (RMSE) of 0.857 nephelometric turbidity units (NTU) for turbidity and 0.415 milligrams per liter (mg/l) for TSS, with residual prediction deviations (RPD) of 2.979 and 3.711, respectively. Compared to regression methods, the FL modeling approach outperformed multiple regression results. Furthermore, using an independent validation dataset, the models achieved R² = 0.846 and RMSE = 1.254 NTU for turbidity, and R² = 0.809 and RMSE = 0.996 mg/l for TSS, confirming the robustness of the developed Landsat 8-based-FL models. The results of this research demonstrated that it is possible to develop generalized Landsat 8-based-FL models that can accurately estimate concentrations of numerous SWQPs at various waterbodies. Ultimately, this research work is essential for decision-makers and local administrators, guiding them in taking appropriate actions at the opportune time.

Graphical Abstract