| 基于遥感影像的木兰溪水质参数反演 |
| |
| 引用本文： | 胡晴晖,宋金玲,黄达,丁琳,林琢,胡家诚.基于遥感影像的木兰溪水质参数反演[J].中国环境监测,2023,39(3):206-214. |
| |
| 作者姓名： | 胡晴晖 宋金玲 黄达 丁琳 林琢 胡家诚 |
| |
| 作者单位： | 福建省近岸海域环境监测站, 福建 莆田 351106;河北科技师范学院数学与信息科技学院, 河北省农业数据智能感知与应用技术创新中心, 河北 秦皇岛 066004;河北科技大学经济管理学院, 河北 石家庄 050025;中国科学院空天信息创新研究院, 北京 100000;莆田市河务管理中心, 福建 莆田 351100 |
| |
| 基金项目： | 国家重点研发计划项目 (2019YFC1407903);河北省重点研发计划项目 (21370103D);河北省自然科学基金面上项目(D2019407046);2021年度河北省社会科学发展研究课题 (20210201445) |
| |
| 摘 要： | 水质监测对水环境评价及污染预防至关重要，但地面监测成本高、监测面积有限等，难以满足实时、大范围监测的要求。为了更好地解决该问题，基于遥感影像的空中监测技术越来越得到研究人员的青睐。以木兰溪为研究区，利用和地面监测数据同步的Landsat-8卫星遥感影像数据，对木兰溪的典型水质参数总磷、总氮、溶解氧、高锰酸盐指数的反演问题进行研究。首先，根据Landsat-8的水体敏感波段，分别选取总磷、总氮、溶解氧、高锰酸盐指数的反演特征波段组合为(b1-b2)/(b2+b3),(b1-b2)/(b3-b4),b2/(b1+b4),b1/b2;其次，利用反演特征波段组合分别构建总磷、总氮、溶解氧、高锰酸盐指数浓度的SVR(Support Vector Regression)反演模型，通过IPSO算法对SVR模型的参数进行优选；然后，将IPSO-SVR反演模型和统计回归反演模型、广义回归神经网络(GRNN)反演模型在验证集上进行评估，以平均绝对误差和均方根误差作为评价指标进行对比分析，结果表明IPSO-SVR反演模型的平均绝对误差和均方根误差最小，说明IPSO-SVR反演模型具有较高的精度和较好的实用性...
|
| 关 键 词： | 遥感影像 反演 水质参数 改进的粒子群优化 支持向量回归 |
| 收稿时间： | 2022/2/10 0:00:00 |
| 修稿时间： | 2022/11/14 0:00:00 |
Research on Water Quality Parameters Inversion in Mulan River Based on Remote Sensing Images |
| |
| HU Qinghui,SONG Jinling,HUANG D,DING Lin,LIN Zhuo,HU Jiacheng.Research on Water Quality Parameters Inversion in Mulan River Based on Remote Sensing Images[J].Environmental Monitoring in China,2023,39(3):206-214. |
| |
| Authors: | HU Qinghui SONG Jinling HUANG D DING Lin LIN Zhuo HU Jiacheng |
| |
| Institution: | Fujian Provincial Monitoring Station of Coastal Environment, Putian 351106, China;College of Mathematics and Information Technology, Hebei Agricultural Data Intelligent Perception and Application Technology Innovation Center, Hebei Normal University of Science & Technology, Qinhuangdao 066004, China;School of Economics and Management, Hebei University of Science and Technology, Shijiazhuang 050025, China;Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100000, China; Putian River Management Center, Putian 351100, China |
| |
| Abstract: | Water quality monitoring is very important for water environment evaluation and pollution prevention,but ground monitoring has the disadvantages of high cost and limited monitoring area,which is difficult to meet the requirements of real-time and large-scale monitoring.In order to better solve this problem,the air monitoring technology based on remote sensing image is more and more popular.The Mulan River basin is took as the research area,Landsat-8 satellite remote sensing images synchronized with the ground monitored water quality data of the basin are used,the inversion of the typical water quality parameters of total phosphorus (TP),total nitrogen (TN),dissolved oxygen (DO),permanganate index (CODMn) is studied.First,according to the water-sensitive band of Landsat-8,the inversion feature band combinations (b1-b2)/(b2+b3),(b1-b2)/(b3-b4),b2/(b1+b4),b1/b2 for TP,TN,DO,CODMn are selected.Secondly,the inversion SVR (Support Vector Regression) model for TP,TN,DO,CODMn are constructed separately using the feature band combination,in which the parameters of the SVR model are optimized by IPSO (Improved Particle Swarm Optimization) algorithm.Then,IPSO-SVR inversion model are evaluated on the verification set with statistical regression inversion model and GRNN (Generalized Regression Neural Network) inversion model,mean absolute error (MAE) and root mean square error (RMSE) are used as evaluation indexes for comparative analysis.The verification results show that the IPSO-SVR model has minimum mean absolute error and root mean square error,indicating that the IPSO-SVR model has high accuracy and good practicability.Finally,IPSO-SVR inversion model is used to analyze the spatial distribution of 4 water quality parameters in some sections of Mulan River. |
| |
| Keywords: | remote sensing image inversion water quality parameters improved particle swarm optimization support vector regression |
|
| 点击此处可从《中国环境监测》浏览原始摘要信息 |
| 点击此处可从《中国环境监测》下载免费的PDF全文 |
|
