| 引用本文: | 吴启侠,晏 军,朱建强,等.花后浅地下水埋深对小麦高光谱特征的影响及叶绿素估算模型[J].灌溉排水学报,2018,37(9):29-35. |
| WU Qixia,YAN Jun,ZHU Jianqiang,et al.花后浅地下水埋深对小麦高光谱特征的影响及叶绿素估算模型[J].灌溉排水学报,2018,37(9):29-35. |
|
| 摘要: |
| 江汉平原春季雨水较多，小麦中后期易受渍害。【目的】将高光谱遥感技术应用于渍害监测，为渍害监测提供一种无损、快捷的诊断方法。【方法】在小麦花后设置不同地下水埋深（0、20和40 cm）处理，分别于处理后8、17、28 d监测小麦冠层光谱反射率和旗叶叶绿素量，分析了小麦花后浅地下水埋深对冠层高光谱特征的影响，并建立了叶绿素高光谱估算模型。【结果】小麦花后0 cm、20 cm地下水埋深持续17 d左右时，小麦冠层反射光谱中蓝紫光波段与红光波段形成的2个吸收谷比40 cm的平坦，而2个吸收谷之间的反射峰变陡，红边位置发生蓝移，且地下水埋深越浅，持续时间越长，2个吸收谷越平坦，蓝移位移越大。浅地下水埋深胁迫小麦旗叶叶绿素a（Chla）、叶绿素b（Chlb）、叶绿素（Chl(a+b)）量分别与红边位置（λr）、红边偏度（Sr）以及红边峰度（Kr）呈线性、线性和一元二次曲线关系。选取λr、Sr、Kr三个特征因子作为网络输入层建立BP神经网络模型估算浅地下水埋深胁迫小麦旗叶Chla、Chlb、Chl(a+b)量，建立的模型其拟合精度高（决定系数R2分别为0.842 5、0.700 2、0.850 8、均方根误差RMSE分别为0.146、0.048、0.173）。【结论】以λr、Sr、Kr为输入层建立的BP神经网络模型可以作为估算浅地下水埋深胁迫小麦旗叶叶绿素量的高光谱估算模型。 |
| 关键词: 小麦； 浅地下水埋深； 高光谱； 红边参数； BP神经网络 |
| DOI：10.13522/j.cnki.ggps.2017.0605 |
| 分类号: |
| 基金项目: |
|
| Effects of the Depth of Shallow Groundwater Table During Post-anthesis Stage on Hyperspectral Characteristics of Winter Wheat as well as Model for Predicting Leaf Chlorophyll Content |
|
WU Qixia, YAN Jun, ZHU Jianqiang, LI Dongwei, ZHOU Xinguo, GUO Shulong
|
|
1. Farmland Irrigation Research Institute, Chinese Academy of Agricultural Sciences, Xinxiang 453002, China; 2. CAAS/Agricultural Environmental Science Observation Experiment Stations of Shangqiu, Ministry of Agriculture, Shangqiu 476001, China; 3. College of Agriculture, Yangtze University, Jingzhou 434025, China
|
| Abstract: |
| 【Objective】The monsoon spring in Jianghan Plain of China often results in waterlogging in wheat field. This study aimed to investigate the feasibility of using hyperspectral remote sensing to monitor the physiological traits of the wheat under different depth of shallow groundwater table in attempts to provide a non-destructive and rapid method to monitor waterlogged stress.【Method】 We designed three shallow groundwater depths at 0, 20 and 40 cm after the anthesis stage. In the experiment, the spectral reflectance of the wheat canopy and the flag leaf chlorophyll content were measured after 8 days, 17 days and 28 days of the onset of the experiment. The effects of the shallow groundwater depth on the hyperspectral characteristics were analyzed and a model was proposed to calculate the chlorophyll content.【Result】 When the subsurface waterlogging continued for about 17 days at groundwater depth of 0 cm and 20 cm, the spectral reflectance of the canopy in the two absorption-valleys in the blue-purple wave band and the infrared wave band increased and the reflection-valley became flat, and the peak between the two absorption valleys became steep. The shallower of groundwater table was, the longer it continued, and the rising of reflectivity in the two absorption valleys became steeper and the peak became flatter. Waterlogging caused an reduction in red absorption and the red edge “blue shifts”. The longer of duration that crop stayed under the shallow groundwater, the more obvious of “blue shifts” was. The liner and quadratic regression models were selected to stimulate the relationship between the position of red edge (λr), the skewness of red edge (Sr), the kurtosis of red edge (Kr) and the Chla, Chlb and Chl(a+b) content of wheat flag leaf under shallow groundwater depth stress. The R2 of the λr, Sr, Kr-based BP neural network model was used to estimate Chla, Chlb and Chl(a+b) content of wheat flag leaf under shallow groundwater depth stress. The R2 were 0.842 5, 0.700 2, 0.850 8, and the RMSE were 0.146, 0.048 and 0.173 respectively. 【Conclusion】 The BP neural network model can be used to estimate the dynamics of the chlorophyll content the flag leaf under the shallow groundwater stress. |
| Key words: winter wheat; shallow groundwater depth; hyperspectral; red edge parameters; BP neural network |
