| 引用本文: | 马岩川,刘浩,陈智芳,等.不同水分条件下棉花冠层含氮量高光谱监测研究[J].灌溉排水学报,0,():-. |
| Ma Yanchuan,Liu Hao,Chen Zhifang,et al.不同水分条件下棉花冠层含氮量高光谱监测研究[J].灌溉排水学报,0,():-. |
|
| 摘要: |
| 摘要：【目的】为了建立一种适用于不同水分条件的棉田氮肥监测模型。【方法】通过设置包含灌溉梯度和施氮梯度的大田水肥试验，在生育期内同步测定棉花冠层光谱反射率、冠层含氮量(Canopy nitrogen content, CNC)、冠层等效水厚度(Canopy equivalent water thickness, CEWT)等信息，综合分析棉花冠层含氮量及冠层等效水厚度与光谱指数的相关性，确定最优光谱指数并构建棉花CNC的高光谱监测模型。【结果】冠层光谱与CNC在可见光波段附近出现连续的敏感区域，其中最大相关系数|r|max为0.53，位于718 nm；在不考虑CEWT对模型精度影响时，NDSI(800,770)的建模效果最佳（R2 = 0.76），但是进入花铃后期其预测精度偏低，出现了低估现象；综合考虑CEWT的影响后，本研究选取NDSI(570,500)作为最优光谱指数，所建模型有效改善了棉花水分含量变化而造成模型精度偏低的现象（RRMSE=0.18）。【结论】本研究建立的新型水分钝感光谱指数NDSI(570,500)可以有效提升棉花CNC的估算精度，为高光谱技术在棉田氮肥监测的应用提供技术依据。 |
| 关键词: 高光谱；棉花；不同水分梯度；冠层含氮量 |
| DOI： |
| 分类号:S562 |
| 基金项目:现代农业棉花产业技术体系建设专项(CARS-15-13)、国家自然科学基金(51709262)、国家重点研发计划(2016YFC0400208)、中央科研院所基本科研业务费专项资金项目（FIRI2018-05） |
|
| Study on hyperspectral monitoring of canopy nitrogen content in cotton under different water conditions |
|
Ma Yanchuan, Liu Hao, Chen Zhifang, Zhang Kai, Wang Jinglei, Sun Jingsheng
|
|
Farmland Irrigation Research Institute, Chinese Academy of Agricultural Sciences
|
| Abstract: |
| Abstract: 【objective】To establish a CNC monitoring model of cotton whitch suitable for different water conditions.【Method】Irrigation gradient and nitrogen application gradient were set up in the field.Information such as spectral reflectance, canopy nitrogen content (CNC) and Canopy equivalent water thickness (CEWT) were measured simultaneously during the growth period. The correlation between nitrogen content and equivalent water thickness of cotton canopy and spectral index was analyzed comprehensively, and the optimal spectral index was determined and the hyperspectral monitoring model of cotton CNC was constructed.【Results】Canopy spectra and CNC have continuous sensitive regions near visible and near-infrared bands，and the maximum correlation coefficient |r| Max is 0.53, located at 718nm. NDSI (800,770) had the best modeling effect (R2 = 0.76) when the influence of CEWT on the model accuracy was not considered. However, in the late stage of flower boll, the prediction accuracy was low and the phenomenon of underestimation appeared. After comprehensively considering the influence of CEWT, NDSI(570,500) was selected as the optimal spectral index in this study，and the model can effectively improve the low accuracy of the model caused by water content (RRMSE=0.18). 【Conclusion】NDSI(570,500), a new moisture insensitive spectral index established in this study, can effectively improve the estimation accuracy of cotton CNC. It provides the technical basis for the application of hyperspectral technology in monitoring nitrogen fertilizer in cotton field. |
| Key words: Hyperspectral; Cotton; Different Water Gradients; Canopy Nitrogen Content |
