基于香溪河中游水体总磷总氮高光谱估算模型比较  

A Comparative Study of Hyperspectral Estimation Models for Total Phosphorus and Total Nitrogen in the Midstream of the Xiangxi River

作　　者：周华冕 徐慧[2] 龙良红[1,2] 纪道斌[1,3] 韩燕星 季鑫鑫 崔玉洁 ZHOU Hua-mian;XU Hui;LONG Liang-hong;JI Dao-bin;HAN Yan-xing;JI Xin-xin;CUI Yu-jie(College of Water Conservancy and Environment,China Three Gorges University,Yichang 443002,Hubei Province,China;Engineering Research Center of Eco-Environment in Three Gorges Reservoir Area,Ministry of Education,Yichang 443002,Hubei Province,China;Hubei Field Observation and Scientific Research Stations for Water Ecosystem in Three Gorges Reservoir,Yichang 443002,Hubei Province,China)

机构地区：[1]三峡大学水利与环境学院,湖北宜昌443002 [2]三峡大学三峡库区生态环境教育部工程研究中心,湖北宜昌443002 [3]三峡水库生态系统湖北省野外科学观测研究站,湖北宜昌443002

出　　处：《中国农村水利水电》2024年第11期125-132,共8页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(52079069);湖北省自然科学基金项目(2022CFB807)。

摘　　要：研究通过地面高光谱遥感技术,针对香溪河中游的水质参数——总氮(TN)和总磷(TP)进行反演分析。研究选取160个数据样本,采用了单波段分析、一阶微分波段分析、波段比值分析、双波长差异指数、归一化差异指数以及偏最小二乘回归(PLS)分析6种方法建立反演模型。结果显示,单波段分析和一阶微分波段分析对于TP和TN的反演效果不佳。而波段比值分析在一定程度上提高了TN的反演准确性,其二次幂回归分析的R^(2)可达0.3674。在双波长差异指数的3次幂分析中,TP的预测模型达到最高R^(2)值,为0.4014。PLS回归分析在TN反演模型中表现突出,波段差值混合模型的R^(2)达到0.39,RMSEP为0.501,ARE为40.278%,展现一定的预测准确性。研究结果有助于利用高光谱遥感预测香溪河中游TN、TP长时间变化趋势,但模型预测精度仍受限于数据集和水体类型,未来研究需进一步探索和优化模型算法。This study employs ground hyperspectral remote sensing to analyze the water quality parameters,specifically total nitrogen(TN)and total phosphorus(TP),in the midstream of the Xiangxi River.Using 160 data samples,inversion models were developed through six methods:single-band analysis,first-order differential band analysis,band ratio analysis,dual-wavelength difference index,normalized difference index,and partial least squares(PLS)regression.The results show that single-band and first-order differential analyses were ineffective for TP and TN.Band ratio analysis improved TN inversion accuracy,achieving an R^(2) of 0.3674 with second-order regression analysis.The highest R^(2) of 0.4014 for TP was obtained using third-order regression of the dual-wavelength difference index.PLS regression showed an outstanding performance in TN inversion,with an R^(2) of 0.39,RMSEP of 0.501,and ARE of 40.278%,showing a certain prediction accuracy.These findings support the potential of hyperspectral remote sensing for predicting the long-term trend of TN and TP in the middle reaches of Xiangxi River,though further optimization of models is needed due to current limitations in data and water body types.

关 键 词：高光谱遥感 总氮(TN) 总磷(TP) 偏最小二乘回归(PLS) 水质监测 

分 类 号：TV121[水利工程—水文学及水资源] P237[天文地球—摄影测量与遥感] X87[天文地球—测绘科学与技术]