|
| 渭-库绿洲耕层土壤砷含量高光谱估测研究 |
| Hyperspectral Estimation of Arsenic Content in Topsoil of Wei-Ku Oasis |
| |
| DOI： |
| 中文关键词: 砷 高光谱估测 包络线去除 一阶微分变换 连续小波变换 耕层土壤 |
| 英文关键词: Arsenic Hyperspectral estimation Envelope removal First-order differential transform Continuous wavelet transform Topsoil |
| 基金项目:国家自然科学基金资助项目（41561051）；新疆维吾尔自治区自然科学基金资助项目（2020D01A79）；新疆师范大学大学生创新创业训练计划基金资助项目（X202110762011） |
|
| 摘要点击次数: 289 |
| 全文下载次数: 77 |
| 中文摘要: |
| 以新疆渭干河库车河三角洲绿洲耕层土壤为研究对象，通过对原始光谱反射率（R）、包络线去除（CR）和一阶微分（R′）变换光谱数据结合连续小波变换生成小波系数，筛选出与砷含量具有极显著相关性的特征波段（P＜0.001），运用偏最小二乘回归、BP神经网络、随机森林回归及支持向量机回归分析方法建立耕层土壤砷含量的高光谱估测模型。结果表明，经一阶微分结合连续小波变换处理后，光谱数据与砷含量之间的相关性提升尤为明显。综合考虑各模型估测精度及稳定性，认为R′-CWT-26-SVMR模型可作为研究区耕层土壤砷含量的最佳估算模型，其训练集和验证集的决定系数（R2）分别为0.753和0.740，均方根误差（RMSE）分别为1.350mg/kg和1.819mg/kg，相对分析误差（RPD）分别为2.03和2.00。 |
| 英文摘要: |
| Taking the topsoil of the Weigan-Kuqa River delta oasis in Xinjiang as the research object，original spectral reflectance （R），envelope removal （CR） and first-order differential （R′） transform spectral data were combined with continuous wavelet transform to generate wavelet coefficient， and the characteristic bands with highly significant correlation with arsenic content were screened out （P＜0.001）. Using partial least squares regression， BP neural network，random forest regression and support vector machine regression analysis， hyperspectral estimation models for arsenic content in topsoil were established. The results showed that the correlation between spectral data and arsenic content was significantly improved after first-order differentiation combined with continuous wavelet transform processing. Considering the estimation accuracy and stability of each model，R′-CWT-26-SVMR model was considered as the best estimation model for arsenic content in topsoil of the study area， and the determination coefficients （R2）of its training set and validation set were 0.753 and 0.740，respectively. The root mean square errors （RMSE）were 1.350 mg/kg and 1.819mg/kg， respectively， and the relative analysis errors（RPD） were 2.03 and 2.00，respectively. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
