基于光谱变换的宁夏银北地区可溶性阴离子反演
作者:
作者单位:

(1.宁夏大学地理科学与规划学院,宁夏 银川 750021;2.榆林市自然资源调查与规划中心,陕西 榆林 719000;3.宁夏大学生态环境学院,宁夏 银川 750021)

作者简介:

陈睿华(1996-),硕士研究生,研究方向为精准农业与土地质量提升。E-mail:dajingyuchen@163.com。

通讯作者:

基金项目:

基金项目:宁夏自然科学基金项目(2020AAC03113)。


Retrieval of soluble anions in northern Yinchuan area of Ningxia region based on spectral transformation
Author:
Affiliation:

(1.College of Geography and Planning,Ningxia University,Yinchuan Ningxia 750021;2.Yulin Natural Resources Survey and Planning Center,Yulin Shaanxi 719000;3.School of Ecology Environment,Ningxia University,Yinchuan Ningxia 750021)

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
    摘要:

    可溶性阴离子是土壤盐分的重要组成部分,对植物生长发育有重要影响。为探讨野外实测光谱对土壤可溶性阴离子的反演精度,以宁夏银北平罗县盐渍化土壤为研究对象,对野外实测光谱选用6种常规变换[平滑R、平滑倒数1/R、平滑对数lg(R)、平滑倒数的对数lg(1/R)、平滑一阶微分R′、平滑二阶微分R″]预处理,然后分别利用相关性分析和逐步回归法筛选离子敏感特征波段,最后采用主成分回归(PCR)、偏最小二乘回归(PLSR)和支持向量机(SVM)建立土壤各阴离子反演模型。结果表明:1)研究区土壤中Cl-和SO42-含量较高,CO32-含量最低,属于氯化物-硫酸盐盐渍土。2)原始反射率经R″变换后与土壤阴离子的相关性最强,与CO32-、HCO3-、Cl-和SO42-相关系数分别达到0.572、0.741、0.802和0.545。3)与相关性分析相比,逐步回归(SR)更好地解决了光谱间共线性的问题,反演精度更高。4)与PCR、PLSR相比,SVM所建可溶性阴离子反演模型效果最佳。土壤CO32-、Cl-、SO42-反演效果最佳的模型均为R″-SR-SVM,其中对CO32-的反演模型建模决定系数(Rc2)为0.984、验证决定系数(Rp2)为0.560、相对分析误差(RPD)为6.76;SO42-的反演模型Rc2为0.970、Rp2为0.841、RPD为5.59;Cl-的反演模型Rc2为0.925、Rp2为0.940、RPD为3.62;HCO3-效果最佳的模型为R′-SR-SVM,Rc2为0.970、Rp2为0.840、RPD为5.59。研究结果可为该区域及同类地区土壤盐渍化反演提供理论依据。

    Abstract:

    Soluble anions are important parts of soil salinity and have an important impact on plant growth and development.In order to explore the inversion accuracy of the soluble anions of the soil in the measured spectrum,in this study the salinized soil in Pingluo county in the northern Yinchuan area was taken as the research object,and 6 conventional transformations[smooth R,smooth reciprocal 1/R,smooth Logarithm lg(R),smoothed reciprocal logarithm lg(1/R),smoothed first-order differential R′,smoothed second-order differential R″]in preprocess were selected,and then correlation analysis and stepwise regression were used to screen ions’ sensitivity characteristics.Finally,principal component regression(PCR),partial least square regression(PLSR)and support vector machine(SVM)were used to establish the soil anion inversion model.The results showed that:1)The content of Cl- and SO42- in the study area were higher than others,and the content of CO32- was the lowest,indicating that the soil in this region belongs to chloride saline soil.2)The original reflectance had the strongest correlation between the contents of anions and reflectance through R″ transformation,and the correlation coefficients of CO32-,HCO3-,Cl- and SO42- were 0.572,0.741,0.802 and 0.545,respectively.3)Compared to correlation analysis,stepwise regression (SR)could solve the problem of collinearity between spectra,and the inversion accuracy was higher.4)Compared to PCR and PLSR,the soluble anion inversion model built by SVM had the best effect.The best model for soil CO32-,Cl- and SO42- inversion was R″-SR-SVM.Among them,modeling coefficient of determination(Rc2),verification coefficent of determination (Rp2) and relative prediction deviation (RPD) of the inversion model of CO32- was 0.984,0.560 and 6.76;Rc2,Rp2 and RPD of the inversion model of SO42- was 0.970,0.841 and 5.59;Rc2,Rp2 and RPD of the inversion model of Cl- was 0.925,0.940 and 3.62,respectively.The best model for soil HCO3- inversion was R′-SR-SVM.Rc2,Rp2 and RPD of the inversion model of HCO3- was 0.970,0.840 and 5.59,respectively.The research results could provide a theoretical basis for the inversion of soil salinization in this area and similar areas.

    参考文献
    相似文献
    引证文献
引用本文

陈睿华,孙媛,尚天浩,张俊华.基于光谱变换的宁夏银北地区可溶性阴离子反演[J].中国土壤与肥料,2022,(8):94-103.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:2021-05-22
  • 最后修改日期:
  • 录用日期:2021-10-09
  • 在线发布日期: 2022-10-10
《中国土壤与肥料》招聘启事
关闭