基于Landsat 8 OLI的东圳水库水质参数反演研究  

Inversion of Water Quality Parameters in Dongzhen Reservoir Based on the Landsat 8 OLI

作　　者：何欢 陈文惠[1] 张忠婷 He Huan;Chen Wenhui;Zhang Zhongting(College of Geographical Sciences,Fujian Normal University,Fuzhou,China)

机构地区：[1]福建师范大学地理科学学院,福建福州

出　　处：《科学技术创新》2024年第1期80-84,共5页Scientific and Technological Innovation

基　　金：福建省科技计划项目(2020Y0070);福建省环保科技计划项目(2022R001)资助。

摘　　要：遥感技术是监测内陆水体水质的有效手段,东圳水库是莆田市水源地,为了对水质进行实时监测,了解其空间分布情况,本文基于Landsat 8 OLI遥感影像,结合124个采样点实测获得的Chl-a、浊度、COD浓度分别构建统计回归模型、BP神经网络模型、XGBoost模型,并采用R^(2)、MAE、RMSE进行精度检验。结果表明BP神经网络模型效果优于统计回归模型,R^(2)均大于0.9,但存在过拟合现象;XGBoost模型可以有效防止过拟合,表现出较强的拟合能力和较高的预测精度。Remote sensing technology is an effective method for monitoring the water quality of inland water.Dongzhen Reservoir serves as the water source for Putian City.In order to achieve real-time water quality monitoring and understand its spatial distribution,this paper is based on Landsat 8 OLI remote sensing images,in combination with measurements of Chl-a,Turbidity,and COD concentrations from 124 sampling points.Statistical regression models,BP neural network models,and XGBoost models were constructed and evaluated using precision tests such as R^(2),MAE,and RMSE.The results indicate that the BP neural network model outperforms the statistical regression model with an R^(2) value exceeding 0.9,albeit with some risk of overfitting.On the other hand,the XGBoost model effectively mitigates overfitting,demonstrating robust fitting capabilities and high predictive accuracy.

关 键 词：Landsat 8 OLI 水质参数 BP神经网络模型 XGBoost模型 

分 类 号：X832[环境科学与工程—环境工程]