基于电磁感应协同野外原位光谱的土壤盐分反演研究
作者:
作者单位:

(塔里木大学植物科学学院,新疆 阿拉尔 843300)

作者简介:

罗德芳(1994-),女,甘肃人,在读硕士,研究方向为高光谱遥感。E-mail:zkyldf@163.com。

通讯作者:

基金项目:

基金项目:国家重点研发计划(2018YFE0107000,2016FYC0501400);国家自然科学基金项目(41361048)。


The study about soil salt inversion based on electromagnetic induction and situ spectra in field
Author:
Affiliation:

(College of Plant Sciences,TarimUniversity,Alar Xinjiang843300)

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:

    快速准确监测土壤盐渍化可为土地资源合理开发利用与改良提供科学依据。利用 EM38-MK2大地电导仪和野外光谱仪测定的土壤表观电导率和光谱数据,构建表观电导率与土壤电导率的反演模型,依据相关性分析结果进行土壤盐渍化特征波段的提取,并采用反射率、反射率倒数和反射率一阶微分 3种数据变换形式构建土壤电导率的全波段与特征波段的偏最小二乘回归与主成分回归土壤盐分监测模型。研究结果表明,EM38-MK2测定的土壤表观水平电导率和表观垂直电导率相结合建立的电导率解译模型的拟合优度达到 0.89,在土壤盐渍化光谱建模中可快速提供电导率数据。全波段建模精度高于特征波段建模精度,偏最小二乘回归建模精度高于主成分回归建模精度,反射率一阶微分变换后建立的模型精度优于反射率倒数变换与反射率。研究区土壤电导率的预测模型选取经一阶微分变换后的全波段偏最小二乘回归建模方法为最佳模型,精度指标可达到 0.85,相对分析误差可达到2.56。

    Abstract:

    The soil salinization rapid and accurate monitoring can provide scientific basis for rational development andimprovement of land resources. In this study,the apparent conductivity and spectral data on soil using EM38 -MK2conductivity meter and field spectrometer are measured,the apparent conductivity and soil conductivity inversion modelis built. After that,characteristics band of the soil salinization is extracted according to the results of correlation analysis,and three kinds of data transformation form are used,including theReflectance,Reflectivity reciprocal and Reflectivityof first order differential. Finally,the model of Partial least squares regression and Principal component regression soil salt monitoring in all and characteristics bands of the soil electrical conductivity bands are structured. The results shows that the apparent horizontal conductivity and apparent vertical conductivity measured by EM38-MK2 combined has built a conductivity interpretation model with a goodness of fit R2of0.89,which can provide conductivity data quickly in soil salinization spectral modeling. The precision of full-bands modeling is superior to characteristic bandmodeling,the partial least square regressionmodeling is more accurate than principal component regression modeling,and the accuracy of the model established after the first-order differential transformation of reflectivity is better than that of the reciprocal transformation and reflectivity. The best model is full-band PLSR modeling method after the first-order differential transformation of the prediction model of soilconductivity in the study area. The precision index can reach0.85,and the relative percent deviation can reach 2.56.

    参考文献
    相似文献
    引证文献
引用本文

罗德芳,冯春晖,吴家林,殷彩云,柳维扬,彭杰 .基于电磁感应协同野外原位光谱的土壤盐分反演研究[J].中国土壤与肥料,2020,(6):107-113.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
历史
  • 收稿日期:2019-09-18
  • 最后修改日期:
  • 录用日期:2019-12-20
  • 在线发布日期: 2021-01-21