| GA-PLS方法提取土壤水盐光谱特征的精度分析 |
| |
| 引用本文： | 柴思跃,马维玲,刘高焕,黄翀,刘庆生.GA-PLS方法提取土壤水盐光谱特征的精度分析[J].遥感技术与应用,2015,30(4):638-644. |
| |
| 作者姓名： | 柴思跃 马维玲 刘高焕 黄翀 刘庆生 |
| |
| 作者单位： | (1.中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室,北京 100101;
2.中国科学院大学,北京 100049) |
| |
| 基金项目： | 国家自然科学基金项目“现代黄河三角洲地下水—土壤—大气相互作用模式研究”（41271407）, 海洋公益性行业科研专项项目(201105020)。 |
| |
| 摘 要： | 光谱定量遥感已成为土壤盐渍化大尺度调查的有效手段之一,但黄河三角洲地区盐渍化土壤的光谱响应特征尚未明确。以黄河三角洲野外测定土壤体积含水率、电导率为例,应用遗传偏最小二乘法(GA-PLS)在小样本集条件下提取盐渍土壤的水分-盐分的光谱响应特征,利用蒙特卡罗方法随机模拟结果表明:在不同土壤水盐含量条件下,GA-PLS方法所提取的光谱特征具有鲁棒性,含水率模型稳定在23个波段变量,即响应特征为365~425,500~515,720~740,755~765与955~965 nm;土壤电导率模型的特征集数目为20个波段变量,特征为370~385,405~425\,500~535,650~660,755~760与1 030~1 050 nm。实验在不同预处理模型下,GA-PLS算法所建立水盐光谱模型较PLSR模型均显示出更高的精度。其中,包络线预处理方法与GA-PLS算法相结合效果最优,其水分光谱模型测试集拟合精度(R2),预测残差平方和(PRSS)与残差预测方差(RPD)分别为0.88,9.36与15.80;土壤光谱模型测试集精度R2,PRSS与RPD分别为0.71,15.68与13.76。
|
| 关 键 词： | 遗传-偏最小二乘算法(GA-PLS) 土壤电导率 高光谱 黄河三角洲 |
| 收稿时间： | 2014-06-18 |
Accuracy Analysis of GA-PLS based Soil Water Salinity Hyperspectral Characteristics Mining Approach |
| |
| Chai Siyue,Ma Weiling,Liu Gaohuan,Huang Chong,Liu Qingsheng.Accuracy Analysis of GA-PLS based Soil Water Salinity Hyperspectral Characteristics Mining Approach[J].Remote Sensing Technology and Application,2015,30(4):638-644. |
| |
| Authors: | Chai Siyue Ma Weiling Liu Gaohuan Huang Chong Liu Qingsheng |
| |
| Affiliation: | (1.Institute of Geographic Sciences and Natural Resources Research,;
Chinese Academy of Sciences,Beijing 100101,China;;
2.University of Chinese Academy of Sciences,Beijing 100049,China) |
| |
| Abstract: | Hyperspectral remote sensing data is one of effective ways which can be used to retrieve salinity quantitatively in soil monitoring.But the quantitative structure-property relationship between soil salinity and soil spectral reflection characters has not been found in yellow river delta region.Genetic Algorithm with Partial Least Square kernel(GA-PLS)method is applied to mine spectral features of volumetric moisture content(V%)and Electrical Conductivity(EC)using the in-stu salinity soil sampling in Yellow River Delta region.MC simulation result shows GA-PLS method mines stable characters numbers and fitness under different of water\|salt level,which prove the robustness of the algorithm.Therefore,the spectral features of V% exist in 365~425,500~515,720~740,755~765 and 955~965 nm bands,compared with the spectral features of EC b appear in 370~385,405~425,500~535,650~660,755~760 and 1 030~1 050 nm bands.According to the experiment result,through 4 different preprocessing approaches,water content model and electric conductivity model of both PLS and GA-PLS are all evaluated by R2t,Predicted Residual Sum of Squares(PRSS)and Residual Predictive Deviation(RPD),GA-PLS models got the better point in prediction accuracy rather than PLS regression.The continuum removal approach leads to the highest prediction accuracy among all other preprocessing methods,with R2,PRSS and RPD equal 0.88,9.36 and 15.80 in soil water content model and 0.71,15.68 and 13.76 in EC model. |
| |
| Keywords: | Genetic algorithm-partial least square(GA-PLS) Soil electrical conductivity Hyperspectral Yellow river delta |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《遥感技术与应用》浏览原始摘要信息 |
|
点击此处可从《遥感技术与应用》下载全文 |
|
