摘要
对新疆渭库绿洲土壤样本的As、Hg、Pb、Cr、Zn和Cu共6种重金属含量与高光谱关系进行分析,基于逐步回归方法筛选出特征波段,采用地理加权回归(GWR)和普通最小二乘法回归(OLS)构建土壤重金属的高光谱反演模型.结果显示:1)研究区的6种重金属元素含量存在一定的差异,土壤中的平均重金属含量依次为Zn>Pb>Cr>Cu>As>Hg,均未超过国家土壤背景值.研究区的Pb平均含量高于当地(新疆)土壤背景值,即研究区表土层中的Pb元素明显富集;2)不同的光谱变换使土壤重金属的光谱特征均得到了增强,但强度有所差异,经过二阶微分变换(SD)和一阶立方根微分变换(CRFD)的土壤光谱相对于原始光谱增强最显著;3)从模型验证来看,在反演6个土壤重金属元素含量时,GWR的R^(2)高于OLS,其中Zn的R^(2)接近0.8,Cu的R^(2)接近0.6,表明模型有一定预测能力,而As、Hg、Pb和Cr的R^(2)均依然低于0.5,模型预测能力并不理想;4)两种模型预测的6种土壤重金属含量的空间分布表现出一定的空间差异性.其中,As元素空间分布差异在这两种预测模型中最大,而其余5种重金属元素Hg、Pb、Cr、Zn和Cu的分布则较为均匀.通过高光谱反射率估算土壤重金属元素含量,实现了干旱区绿洲农田土壤重金属元素含量的高效快速反演,为干旱区绿洲农田土壤重金属元素含量动态监测提供了可靠的技术支撑.
Analysis of the relationship between the content of six heavy metals(As,Hg,Pb,Cr,Zn,and Cu)in soil samples from the Ugan-Kuqa River Oasis in Xinjiang and hyperspectral data was conducted.Feature bands were selected based on the stepwise regression method,and soil heavy metal hyperspectral inversion models were constructed using Geographically Weighted Regression(GWR)and Ordinary Least Squares(OLS).1)There were differences in the content of the six heavy metal elements in the study area,with the average content of heavy metals in the soil being Zn>Pb>Cr>Cu>As>Hg,all of which did not exceed the national soil background values.The average Pb content in the study area was higher than the local(Xinjiang)soil background value,indicating significant enrichment of Pb in the surface soil layer of the study area;2)Different spectral transformations enhanced the spectral characteristics of soil heavy metals,but with some differences in intensity,with soil spectra after second-order differential transformation(SD)and first-order cube root differential transformation(CRFD)showing the most significant enhancement compared to the original spectra;3)From the perspective of model validation,when inverting the content of the six soil heavy metal elements,the R^(2) of GWR was higher than that of OLS,with the R^(2) of Zn approaching 0.8 and that of Cu approaching 0.6,indicating a certain predictive ability of the model,while the R^(2) of As,Hg,Pb,and Cr were still below 0.5,indicating that the predictive ability of the model was not ideal;4)The spatial distribution of the six soil heavy metal contents predicted by the two models exhibited some spatial differences.Among them,the spatial distribution difference of As was the largest in these two prediction models,while the distribution of the other five heavy metal elements,Hg,Pb,Cr,Zn,and Cu,was relatively uniform.By estimating soil heavy metal contents through hyperspectral reflectance,efficient and rapid inversion of soil heavy metal contents in oasis farmland in arid areas had bee
作者
买买提·沙吾提
阿不都艾尼·阿不里
胡昕
Mamat SAWUT;Abudugheni ABLIZ;HU Xin(College of Geography and Remote Sensing Sciences,Xinjiang University,Urumqi 830017,China;Xinjiang Key Laboratory of Oasis Ecology,Xinjiang University,Urumqi 830017,China;Key Laboratory of Smart City and Environment Modelling of Higher Education Institute,Xinjiang University,Urumqi 830017,China)
出处
《中国环境科学》
EI
CAS
CSCD
北大核心
2024年第4期2208-2216,共9页
China Environmental Science
基金
新疆自然科学计划(自然科学基金)联合基金资助项目(2021D01C055)
国家自然科学基金资助项目(42167058)。
