光谱数据变换对玉米氮素含量反演精度的影响

Effect on Retrieval Precision for Corn N Content by Spectrum Data Transformation

通过对玉米叶片光谱数据进行6种变换,分析了变换后的光谱值与叶片氮素含量的相关关系,探讨了550 nm和680 nm两波段处不同形式光谱变量对氮素含量反演的精度。结果表明,微分处理(D(R)、D(log(R))和D(N(R)))显著改变了氮素含量与光谱值的相关性,归一化(N(R))次之,对数处理几乎无变化(R与log(R),N(R)与log(N(R)))。不同的变换形式之间,与氮素含量相关性高的,所建立的回归模型的决定系数较高,模型的精度也较高。在波段550 nm和680 nm波段处,光谱数据的归一化对数处理(log(N(R)))能显著提高回归模型对氮素含量的反演精度。

It was analyzed that correlation between the transformed spectrum value and leaf nitrogen content,further more it was researched that retrieval precision for the nitrogen content with different form spectrum variables in 550nm and 680nm wave band by carrying on 6 kind of transformations to the corn leaf spectrum data.The results showed that differential form（D（R）,D（Log（R））and D（N（R））） could significantly change the correlation between the nitrogen content and the spectrum value,and the normalized form（N（R））was next and the last was logarithm form（R and Log（R）,N（R）and log（N（R）））.In addition,determined coefficient of regression model was larger and the model precision was higher correspondingly with better correlation between different form spectrum variables and nitrogen content.In this study,the normalized logarithm form（log（N（R））） parameter of spectrum data could significantly improve the precision of regression model for retrieval nitrogen content in 550nm and 680nm wave band.