| 基于去包络线和连续投影算法的枣园土壤电导率光谱检测研究 |
| |
| 引用本文： | 王涛,喻彩丽,张楠楠,王斐,白铁成.基于去包络线和连续投影算法的枣园土壤电导率光谱检测研究[J].干旱地区农业研究,2019,37(5):193-199. |
| |
| 作者姓名： | 王涛 喻彩丽 张楠楠 王斐 白铁成 |
| |
| 作者单位： | 塔里木大学信息工程学院,新疆 阿拉尔843300;新疆南疆农业信息化研究中心,新疆 阿拉尔843300;塔里木大学信息工程学院,新疆 阿拉尔843300;Gembloux Agro-Bio Tech, University of Liège, Gembloux, Belgium 25030 |
| |
| 基金项目： | 国家自然科学基金项目（41561088，61501314）；塔里木大学校长基金项目（TDZKQN201614）；塔里木大学现代农业工程重点实验室项目（TDNG20160501） |
| |
| 摘 要： | 选取新疆阿拉尔市典型极端干旱区为研究对象，利用土壤高光谱特征对土壤电导率进行反演。为了准确快速检测土壤电导率，通过获取南疆阿拉尔市红枣种植区土壤电导率和高光谱信息，在去包络线处理基础上，分别采用相关性分析法和连续投影算法（SPA）筛选特征波长，并建立特征波长与土壤电导率的偏最小二乘回归模型，使用均方根误差(RMSE)、决定系数（R2）以及相对分析误差（RPD）对不同处理方法的模型效果进行评价。结果表明，基于原始光谱直接使用相关性分析法的预测精度RMSE=0.85566，R2=0.7479，RPD=2.7569；通过去包络线处理使用相关性分析筛选特征波长后，模型的预测精度RMSE=0.44490，R2=0.9500，RPD=6.4510；基于原始光谱使用SPA选择特征波长后，模型的预测精度RMSE=0.31178，R2=0.9707，RPD=8.4445；通过去包络线处理使用SPA选择特征波长后，模型的预测精度RMSE=0.30173，R2=0.9764，RPD=9.3215。综上,说明SPA方法具有较强的特征波长选择能力，基于去包络线处理+SPA的偏最小二乘回归反演模型的预测精度最好，可实现新疆阿拉尔地区土壤电导率的快速检测。
|
| 关 键 词： | 土壤电导率 光谱检测 连续投影算法 去包络线 预测 |
Spectral detection of electrical conductivity in jujube orchard soil based on continuum-removal and SPA |
| |
| WANG Tao,YU Cai-li,ZHANG Nan-nan,WANG Fei,BAI Tie-cheng.Spectral detection of electrical conductivity in jujube orchard soil based on continuum-removal and SPA[J].Agricultural Research in the Arid Areas,2019,37(5):193-199. |
| |
| Authors: | WANG Tao YU Cai-li ZHANG Nan-nan WANG Fei BAI Tie-cheng |
| |
| Abstract: | The typical extreme arid area of Alar City, Xinjiang was selected as the research object, and the soil electrical conductivity was inverted by using the soil hyperspectral characteristics. In order to accurately and quickly detect the soil electrical conductivity, the soil electrical conductivity and hyperspectral information of the red jujube planting area in Alar City, southern Xinjiang were obtained. Based on the continuum-removal, the correlation analysis method and the successive projections algorithm(SPA) were used to select the characteristics wavelength, and establish a partial least squares regression model of characteristic wavelength and soil electrical conductivity, using the root mean square error (RMSE), determination coefficient (R2) and relative analysis error (RPD) to evaluate the model effect of different processing methods. The results showed that the prediction accuracy based on the original spectrum directly using the correlation analysis method was RMSE=0.85566, R2=0.7479, RPD=2.7569. After the feature wavelength was selected by continuum-removal, the prediction accuracy of the model was RMSE=0.44490, R2=0.9500, RPD=6.4510; after using the SPA to select the characteristic wavelength based on the original spectrum, the prediction accuracy of the model was RMSE=0.31178, R2=0.9707, RPD=8.4445; the model was predicted by continuum-removal using SPA to select the characteristic wavelength. The accuracy was RMSE=0.303173, R2=0.9764, RPD=9.3215. In summary, the SPA method had strong feature wavelength selection ability. The prediction accuracy of partial least squares regression inversion model using SPA based on the continuum-removal was best, which could realize the rapid soil conductivity in Xinjiang Alar region detection. |
| |
| Keywords: | soil electrical conductivity spectral detection successive projections algorithm continuum removal prediction model |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《干旱地区农业研究》浏览原始摘要信息 |
| 点击此处可从《干旱地区农业研究》下载免费的PDF全文 |
|
