摘要
为明确太湖地区土壤全氮的高光谱特征,构建定量分析模型,以江苏省无锡市滨湖区为研究区域,选取地理位置跨度大、土壤质地相似的93个样品,进行土壤风干样品全氮含量测定和光谱数据采集,对光谱反射率进行一阶微分,运用相关系数峰谷值法筛选敏感波长,将敏感波长两两结合进行土壤调节光谱指数(MSASI)运算。将两两结合后敏感波段分别采用多元线性回归分析、人工神经网络分析和偏最小二乘法构建土壤全氮含量的定量高光谱分析模型。结果表明,研究区内土壤全氮含量与光谱反射率呈正相关,敏感波段包括420~444 nm和480~537 nm。基于土壤调节光谱指数的多元线性回归分析对敏感波段诊断的效果最佳(R^2=0.98、RMSE=0.04),其精度高、可靠性强,是筛选出的最佳土壤全氮含量估测模型。偏最小二乘法模型(R^2=0.70、RMSE=0.13)次之,而人工神经网络模型(R^2=0.69、RMSE=0.15)精度最低。该研究结果为太湖地区土壤全氮水平的高光谱快速估测提供了方法借鉴,可为土壤养分精准管理提供技术参考。
In order to achieve a rapid and accurate estimation of soil total nitrogen(TN)in the Taihu Lake region,the hyperspectral characteristics of TN and quantitative analysis models should be constructed to reduce potential environmental risks and provide a technical reference for precision agriculture.In this study,we acquired spectral data and TN of 93 samples with similar soil textures from a large geographical lakeshore zone in Wuxi,Jiangsu Province.Three model methods were investigated.We filtered the sensitive wavelengths using a correlation coefficient in peak value.Based on regulating soil spectral index(MSASI)arithmetic established by sensitive wavelengths,we used multivariate linear regression,artificial neural networks and partial least squares methods to construct a quantitative spectral analysis model of soil TN content.Our results indicated that total soil nitrogen content and spectral reflectance were positively correlated.We also found that sensitive wave bands,including 420~444 nm and 480~537 nm based on MSASI and multiple linear regression analysis on the sensitive wavelengths yield the best results(R^2=0.98,RMSE=0.04)with high accuracy and reliability and were the best models for predicting soil total nitrogen content.The partial least squares model(R^2=0.70,RMSE=0.13)and artificial neural network model(R2=0.69,RMSE=0.15)were the least accurate.
作者
宋雪
张民
周洪印
于小晶
刘之广
徐子云
王有良
SONG Xue;ZHANG Min;ZHOU Hong-yin;YU Xiao-jing;LIU Zhi-guang;XU Zi-yun;WANG You-liang(National Engineering Laboratory for Efficient Utilization of Soil and Fertilizer Resources,College of Recourses and Environment,Shan⁃dong Agricultural University,Tai′an 271018,China;State Key Laboratory of Nutrition Resources Integrated Utilization,Kingenta Ecolog⁃ical Engineering Group Co.,Linshu 276700,China;College of Information and Engineering,Shandong Agricultural University,Tai′an 271018,China)
出处
《农业资源与环境学报》
CAS
北大核心
2020年第1期43-50,共8页
Journal of Agricultural Resources and Environment
基金
国家重点研发计划项目(2017YFD0200706)
国家自然科学基金项目(41571236)
全国农业专业学位研究生实践教育示范基地项目(MA201601008)
山东省研究生教育创新计划项目(SDYY16043)
山东省研究生导师指导能力提升项目(SDYY18108)
山东农业大学研究生教育教学改革研究项目(YZD2018002)~~
关键词
全氮
高光谱分析
模型
快速评估
敏感波段
total nitrogen
hyperspectral analysis
estimation model
rapid assessment
sensitive wave bands