摘要
传统土壤重金属含量信息获取方式成本高、效率低,反射光谱分析技术的发展为土壤重金属研究提供了新的手段。针对以往研究多采用风干研磨后的土壤反射光谱导致的效率低下等问题,以云南省杞麓湖流域的216份农田鲜土样本作为研究对象,在实验室测试分析鲜土样本的可见—近红外反射光谱及重金属铬(Cr)等元素含量的基础上,结合不同光谱预处理方法和偏最小二乘回归(PLSR)构建农田土壤铬(Cr)含量的估算模型,并进一步采用偏相关分析探究模型构建机理。结果表明:基于均值中心化反射光谱的PLSR模型可以对Cr含量进行估算(RPD>1.4)。Cr与Fe的相关关系显著(p<0.05),并且光谱第一主成分(PC1)与Cr的偏相关系数在控制土壤有机质(SOM)前后未有变化,在控制Fe之后则由显著(p<0.05)变为不显著。因此,Cr元素可以被反射光谱估算的机理在于其与Fe元素的相关性。研究表明利用鲜土反射光谱在流域尺度上估算农田土壤重金属Cr元素具有较好的可行性。
Conventional methods for obtaining heavy metal information in soil are costly and inefficient.The development of spectral analysis technology provides new alternatives for the study of heavy metals in soil.Previous studies utilized reflectance spectra of soil samples after air drying and grinding,which is not efficient.In this study,a total of 216 farmland fresh soil samples from Qilu Lake basin of Yunnan Province were collected.Then we measured the visible-near infrared reflectance spectra and chromium(Cr)content of each soil sample.With combined partial least-squares regression(PLSR)with different spectral preprocessing methods,models were built to estimate Cr content in farmland soil.We further used the partial correlation analysis to explore the mechanism of model construction.The results showed that Cr content can be estimated based on the PLSR model of mean centered reflectance spectra(RPD>1.4).There was a significant correlation between Cr and Fe element(p<0.05).Moreover,the partial correlation coefficient of the first princapal component of spectra(PC1)and Cr did not change while controlling soil organic matter(SOM),but changed from significant(p<0.05)to insignificant after controlling for Fe,which could be the main mechanism for estimating Cr content by visible-near infrared spectroscopy.Our research suggests that it is feasible to estimate Cr content in farmland soil with reflectance spectroscopy of moist soil at a watershed scale.
作者
魏钰
陈奕云
徐璐
洪永胜
WEI Yu;CHEN Yi-yun;XU Lu;HONG Yong-sheng(School of Resource and Environmental Science,Wuhan University,Wuhan 430079,China;State Key Laboratory of Soil and Sustainable Agriculture,Nanjing 210008,China;Key Laboratory of Digital Mapping and Land Information Application,Ministry of Natural Resources,Wuhan 430079,China;Key Laboratory of Geographic Information System,Ministry of Education,Wuhan 430079,China)
出处
《国土资源科技管理》
2021年第1期118-129,共12页
Scientific and Technological Management of Land and Resources
基金
国家自然科学基金项目(41771440)。
关键词
土壤重金属
可见光—近红外光谱
偏最小二乘回归
光谱预处理
估算机理
heavy metals in soil
visible-near-infrared spectroscopy
partial least squares regression
spectral pretreatment
estimation mechanism
