| 谭先明,王仲林,张佳伟,王贝贝,杨峰,杨文钰.基于连续小波变换的干旱胁迫下玉米冠层叶绿素密度估测[J].干旱地区农业研究,2021,39(4):155~161 |
| 基于连续小波变换的干旱胁迫下玉米冠层叶绿素密度估测 |
| Estimation of maize canopy chlorophyll density under drought stress based on continuous wavelet transform |
| |
| DOI：10.7606/j.issn.1000-7601.2021.04.20 |
| 中文关键词: 干旱胁迫 叶绿素密度 连续小波变换 植被指数 冠层；玉米 |
| 英文关键词:drought stress chlorophyll density continuous wavelet transform vegetation index canopy maize |
| 基金项目:国家重点研发计划项目（2016YFD0300602） |
|
| 摘要点击次数: 1122 |
| 全文下载次数: 134 |
| 中文摘要: |
| 测定了玉米各生育时期干旱胁迫下冠层叶绿素密度及冠层光谱数据，利用原始光谱反射率与冠层叶绿素密度进行相关性分析，采用常用植被指数、波段自由组合、连续小波变换构建玉米冠层叶绿素密度估测模型，并用决定系数(R2)、均方根误差(RMSE)进行精度检验。结果表明：冠层叶绿素密度在抽雄期相较于正常对照，轻度、中度、重度干旱胁迫处理分别下降7.8%、29.5%、44.2%；波段自由组合指数RVI(555，538)、NDVI(555，538)和敏感小波系数bior5.5(26，792)、rbio2.6(22，790)、gaus6(21，791)与叶绿素密度的相关性较高，相关系数绝对值均达到0.900以上；基于敏感小波系数构建的冠层叶绿素密度估测模型验证集R2均在0.850以上，相较于其他植被指数模型R2平均提升20.6%，RMSE平均降低32.6%；最优模型为以gaus6（21,791）为自变量的一元线性回归模型，R2为0.864，RMSE为1.532。利用连续小波变换对光谱数据进行预处理，可以有效提升玉米冠层叶绿素密度估测模型的精度。 |
| 英文摘要: |
| This study measured maize canopy chlorophyll density and canopy spectrum data under drought stress in each growth period.The original spectral reflectance and canopy chlorophyll density were used for correlation analysis.Using common vegetation index, free combination of bands, continuous wavelet transform to build maize canopy chlorophyll density estimation model, and the coefficient of determination (R2) and root mean square error (RMSE) were used to test its accuracy. The results showed that: compared with the normal control the canopy chlorophyll density during the tasseling stage decreased by 7.8%, 29.5%, and 44.2% under mild, moderate, and severe drought stress, respectively; RVI(555,538), NDVI(555,538) and sensitive wavelet coefficients bior 5.5 (26,792), rbio 2.6 (22,790) and gaus 6 (21,791) had higher correlation with chlorophyll density, and the absolute correlation coefficients were all above 0.900; The validation set R2 of the canopy chlorophyll density estimation models based on sensitive wavelet coefficients were all above 0.850, comparing with other vegetation index models, the average increase was 20.6%, and the average RMSE decreased by 32.6%; The optimal model was a univariate linear regression model with gaus 6 (21,791) as the independent variable R2=0.864, RMSE=1.532. Using continuous wavelet transform to preprocess the spectral data effectively improve the accuracy of the maize canopy chlorophyll density estimation model. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| | |