基于高光谱数据的滴灌甜菜叶绿素含量估算

Estimation of chlorophyll content in sugar beet under drip irrigation based on hyperspectral data

为明确甜菜叶绿素含量与高光谱植被遥感的定量关系,探索建立干旱区甜菜叶绿素含量估测模型,即时监测甜菜生长状况,选取新疆滴灌甜菜(Beta356)为研究对象,利用ASD野外高光谱仪在甜菜叶丛快速生长期、块根膨大期与糖分积累期采集各处理反射光谱,并同时测定叶绿素含量,分析原始光谱反射率和一阶微分光谱反射率与叶绿素含量的相关关系,并进一步建立光谱特征参数和敏感波段植被指数叶绿素含量估算模型。结果表明:原始光谱反射率在近红外区(700~1300 nm)随着氮素水平的增加呈先升高后降低趋势,红边(680~760 nm)也表现出相同趋势,原始光谱反射率在近红外区(700~1300 nm)随着运筹管理的递进呈现升高趋势,红边(680~760 nm)也表现出相同趋势;原始光谱反射率和一阶微分反射率与叶绿素含量均具有较好的相关性,其最大正相关分别位于902 nm(r=0.574,P<0.01)和676 nm(r=0.843,P<0.01)附近,最大负相关分别位于611 nm(r=-0.664,P<0.01)和1138 nm(r=-0.727,P<0.01)附近。对所建12个线性模型进行精度检验,其中差值植被指数DR676–DR446和DR676估算模型的预测值与实测值的决定系数分别达到0.774和0.781,以DR676所建立的估算模型最优。本研究为快速无损监测甜菜生长状况、制定氮素管理方案、指导甜菜氮肥管理提供支持。

In this study,a model for estimating the chlorophyll content of sugar beet in arid areas was established to clarify the quantitative relationship between chlorophyll content of sugar beet and remote sensing of hyperspectral vegetation,and to monitor the growth status of sugar beet in real time.We investigated this relationship in Xinjiang drip-irrigated sugar beet(Beta vulgaris‘Beta356’).Spectral reflectance of each treatment was collected by ASD field hyperspectral radiometer and chlorophyll content was measured in the period of rapid leaf growth,root expansion period,and sugar accumulation period.The correlations between original spectral reflectance,first-order differential spectral reflectance,and chlorophyll content were analyzed,and a hyperspectral remote sensing model for estimating chlorophyll content,as well as a sensitive band vegetation index,was established.The results showed that the original spectral reflectance in the near-infrared region(700~1300 nm)first increased,but thereafter decreased with the increase in nitrogen application rate,and the red edge(680~760 nm)showed the same trend.The original spectral reflectance in the near-infrared region increased with the change in nitrogen management mode,and the red edge(680~760 nm)showed the same trend.The original spectral reflectance and the first-order differential reflectance were correlated with chlorophyll content.Maximum positive correlation was observed near 902 nm(r=0.574,P<0.01)and 676 nm(r=0.843,P<0.01),while maximum negative correlation was observed near 611 nm(r=-0.664,P<0.01)and 1138 nm(r=-0.727,P<0.01).While testing the accuracy of 12 established linear models,the determination coefficients of real and predicted vegetation index values of DR676–DR446 and DR676 reached 0.774 and 0.781,respectively,and the estimation model established using DR676 was found to be the best.This study provides information on the rapid and non-destructive monitoring of sugar beet growth and the development and regulation of a nitrogen management program.