利用高光谱数据估测植物叶片碳氮比的可行性研究

植物碳氮比作为一个在农业、生态、全球变化等领域广泛使用的因子,如果能够利用遥感获得的高光谱数据进行估测,可以突破传统测量方法的种种弊端,具有重要的实践意义,同时对于定量遥感反演领域的拓宽也具有启示作用。利用统计分析的方法,对碳氮比遥感定量估测的可行性进行深入探讨,认为利用高光谱数据估测植物叶片碳氮比是可行的。另外还通过与氮的遥感定量研究相比较,找到一个较好的研究碳氮比遥感定量反演的切入点,并将两者分别作为因变量进行逐步回归分析,得到比较理想的结果。

Feasibility of Estimating Leaf C/N Ratio with Hyperspectral Remote Sensing Data

This paper presents a statistical analysis on the feasibility of estimating leaf C/N ratio with hyperspectral remote sensing data. Correlation analysis on the samples of NASA ACCP data set shows that the changes of correlation coefficients with wavelengths in 3 different forest sites are similar. Correlation coefficients of C/N ratio and reciprocal of nitrogen concentration with spectra from 400nm to 2500nm are compared respectively. Their correlation curves fit each other so well, which reveals a shortcut in C/N ratio researches by using the methods and validated results in previous nitrogen researches. Stepwise multiple linear regressions are employed on the 3 sites respectively. C/N ratio, nitrogen concentration and reciprocal of nitrogen concentration are calculated as dependent variables respectively, and the apparent absorption spectra at different wavelengths are calculated as independent variables. Almost all adjusted R+2 of the 3 sites are more than 0.8 with not more than 4 wavelengths being selected. The satisfactory statistical results prove that estimating leaf C/N ratio with hyperspectral remote sensing data is feasible.