余哲修, 张超, 任伟, 等. 香蒲砷含量高光谱估算模型研究[J]. 西南林业大学学报(自然科学), 2020, 40(6): 87–96 . DOI: 10.11929/j.swfu.201902049
引用本文: 余哲修, 张超, 任伟, 等. 香蒲砷含量高光谱估算模型研究[J]. 西南林业大学学报(自然科学), 2020, 40(6): 87–96 . DOI: 10.11929/j.swfu.201902049
Zhexiu Yu, Chao Zhang, Wei Ren, Yi Zhang, Xiaohua Niu, Yan Wang, Yungen Liu. Study on Estimation Model for Arsenic Content of Typha orientalis by Hyperspectral Data[J]. Journal of Southwest Forestry University, 2020, 40(6): 87-96. DOI: 10.11929/j.swfu.201902049
Citation: Zhexiu Yu, Chao Zhang, Wei Ren, Yi Zhang, Xiaohua Niu, Yan Wang, Yungen Liu. Study on Estimation Model for Arsenic Content of Typha orientalis by Hyperspectral Data[J]. Journal of Southwest Forestry University, 2020, 40(6): 87-96. DOI: 10.11929/j.swfu.201902049

香蒲砷含量高光谱估算模型研究

Study on Estimation Model for Arsenic Content of Typha orientalis by Hyperspectral Data

  • 摘要: 选取典型湿地植物香蒲为研究对象,置于不同砷浓度梯度环境下培养,采集香蒲反射光谱数据,分析不同砷浓度条件下香蒲的反射光谱特征,通过原始光谱反射率、导数光谱和窄波段植被指数构建砷含量的估算模型。结果表明:不同梯度环境的香蒲均呈现健康绿色植被的反射光谱特征,第8和第9梯度的香蒲在1950~2450 nm范围的反射率低于其他梯度;一阶导数和二阶导数变换后,各梯度香蒲反射光谱在近红外波段间的差异明显,“三边”参数的比值变量能定量分析各梯度香蒲的光谱差异。原始光谱反射率与砷含量的相关性均为负相关,973 nm的反射率二阶微分与砷含量的相关性为最高,相关系数为−0.494,窄波段植被指数相关性最高的指数组合为FRVI (1065, 1260),相关系数为0.760。基于FRVI (1065, 1260)构建的回归模型估算效果最佳,R2为0.793,RMSE为107.457。

     

    Abstract: Typha orientalis as a typical wetland plant were cultured in different arsenic concentration gradients. The reflectance spectrum data of T. orientalis were collected, and the reflectance spectrum characteristics of T. orientalis under different arsenic concentration were analyzed. Estimation models were established by original spectral reflectance, derivative spectrum and narrow band vegetation index. The result showed that the reflectance features of T. orientalis in different gradients was similar to that of healthy green vegetation, but the reflectance of the 8th and 9th gradients were lower than that of other gradients in the band range of 1950-2450 nm. The obvious difference of near-infrared band range between different arsenic gradients after first order derivative and second order derivative transformation. Reflectance spectral difference were analyzed quantitatively by ratio variables of 'three edges' parameters between different arsenic gradients. The correlation was negative between the original spectral reflectance and arsenic content. The highest correlation between the second order derivative reflectance of 973 nm and arsenic contents, the correlation coefficient was −0.494. The highest correlation index combination was FRVI (1065, 1260) in all narrow band vegetation indices, the correlation coefficient was 0.760. The best regression model was based on FRVI (1065, 1260) in all models, R2 and RMSE were 0.793, 107.457 respectively.

     

/

返回文章
返回