基于高光谱植被指数的棉花干物质积累估算模型研究

Estimating Modeling of Cotton Dry Matter Accumulation Based on Hyperspectral Vegetation Index

利用北疆8个棉花主栽品种(其中2个棉花品种设4水平种植密度)的各生育期冠层高光谱数据,经多元统计分析与光谱微分处理,建立了基于比值植被指数(RVI)和归一化植被指数(NDVI)的5种函数形式的棉花干物质积累估测模型,相关系数均达到了极显著水平(α=1%,n=96)。基于RVI和NDVI构建的估测模型,前者比后者具有更高的估测精度,指数函数、对数函数和双曲线函数形式的模型可以产生较高的预测精度;一阶微分光谱数据与棉花干物质积累量的逐步回归相关分析表明,相关系数的最高值发生在748 nm波段处(r=0.6920**),由748 nm波段处的微分数值建立的回归模型,估测精度较高,具有实际应用的潜力。

Utilizing Multivariate Regression Modeling and Spectral Derivative to establish five function pattern cotton dry matter accumulation modeling based on Ratio Vegetation Index（RVI） and Normal ized Difference Vegetation Index （NDVI）, all correlation coefficient is highly positive （α=1 %, n=96）, based on RVI modeling has a higher precision than NDVI does, hyperbolic curve fitting, expo nential curve fitting and logarithmic curve fitting have further improved analysis against cotton dry matter accumulation. Analysis of correlation between the first derivative spectra data and dry matter yield, the maximum Value of correlation coefficient （r=0. 6920^＊＊） in waveband 748 nm. Using derivaT tive spectra data at waveband 748 nm, the regression modeling of dry matter accumulation has been established, and derivative spectra has improved the prediction of cotton dry matter accumulation, so it has shown a potential capacity in application.