Scholarlink

<p>Remote sensing technology has become an important way of water quality monitoring, and its characteristics such as fast, wide range and low cost are of great significance for large-scale water quality inversion. Based on Landsat 8 OLI/Landsat 9 OLI remote sensing images and the measured data provided by the national surface water quality automatic monitoring station. The middle reaches of the Yangtze River Basin were chosen as the study area. The Back Propagation (BP) neural network model was constructed to analyze turbidity, electrical conductivity, total phosphorus concentration (TP), total nitrogen concentration (TN) and gray value. The accuracy test was conducted, and the spatial distribution characteristics were analyzed. Combining with the land use and land cover graph, the causes of overall and local pollution in the study area were analyzed and deeply discussed. And a 1&#xa0;km buffer zone was established for the area with a&#xa0;high concentration of water quality parameters in the study area to further study the correlation of land use. The results show that the BP neural network model has a&#xa0;good correlation with turbidity, the correlation coefficient of the turbidity’s total value is 0.798, with a mean absolute error (MAE) of 10.358, a root mean square error (RMSE) of 17.254, a mean absolute percentage error (MAPE) of 29.0%. And the BP neural network model has low MAPE with TN, the correlation coefficient of the TN’s total value is 0.536, with a MAE of 0.597, with a RMSE of 1.202, with a MAPE of 26.3%. This means that the model can predict the trend of turbidity in surface water well and the inversion of TN has a good precision. It indicates that the inversion model constructed in this paper has a&#xa0;certain scientific reference value. It shows that the model constructed in this paper has a certain scientific reference value in the large-scale water area inversion. The model construction idea makes up for the shortcomings of the research on the difficulty of obtaining the spatial distribution of large-scale water quality in a short period of time. The study area is seriously affected by total nitrogen pollution, mainly due to urban pollution, and is greatly affected by industrial pollution and human activities, which proves that the study has certain feasibility in exploring the causes and functions of land use and pollution.</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

Remote Sensing Inversion Model of Water Quality Parameters Based on BP Neural Network and Spatial Distribution Analysis in the Middle Reaches of the Yangtze River Basin in China

  • 期刊论文

  • 发表日期：

    2025年3月12日

查看全文
  • Xinyao Luo,
  • Qiaozhen Guo,
  • Yaxin Tian,
  • Junwu Cao,
  • Gan Luo

摘要

Remote sensing technology has become an important way of water quality monitoring, and its characteristics such as fast, wide range and low cost are of great significance for large-scale water quality inversion. Based on Landsat 8 OLI/Landsat 9 OLI remote sensing images and the measured data provided by the national surface water quality automatic monitoring station. The middle reaches of the Yangtze River Basin were chosen as the study area. The Back Propagation (BP) neural network model was constructed to analyze turbidity, electrical conductivity, total phosphorus concentration (TP), total nitrogen concentration (TN) and gray value. The accuracy test was conducted, and the spatial distribution characteristics were analyzed. Combining with the land use and land cover graph, the causes of overall and local pollution in the study area were analyzed and deeply discussed. And a 1 km buffer zone was established for the area with a high concentration of water quality parameters in the study area to further study the correlation of land use. The results show that the BP neural network model has a good correlation with turbidity, the correlation coefficient of the turbidity’s total value is 0.798, with a mean absolute error (MAE) of 10.358, a root mean square error (RMSE) of 17.254, a mean absolute percentage error (MAPE) of 29.0%. And the BP neural network model has low MAPE with TN, the correlation coefficient of the TN’s total value is 0.536, with a MAE of 0.597, with a RMSE of 1.202, with a MAPE of 26.3%. This means that the model can predict the trend of turbidity in surface water well and the inversion of TN has a good precision. It indicates that the inversion model constructed in this paper has a certain scientific reference value. It shows that the model constructed in this paper has a certain scientific reference value in the large-scale water area inversion. The model construction idea makes up for the shortcomings of the research on the difficulty of obtaining the spatial distribution of large-scale water quality in a short period of time. The study area is seriously affected by total nitrogen pollution, mainly due to urban pollution, and is greatly affected by industrial pollution and human activities, which proves that the study has certain feasibility in exploring the causes and functions of land use and pollution.