论文部分内容阅读
光谱数据变换对消除背景、噪音影响以及提取光谱特征有重要的作用,是光谱数据分析过程中的必要步骤。为了研究光谱变换处理对土壤氮素PLSR模型的影响精度,并选择最佳光谱变换处理方法,本文对原始光谱数据进行了15种典型光谱变换,通过比较不同变换光谱与土壤氮素的相关性,实现土壤氮素的PLSR精确诊断,并综合评定最佳光谱数据变换方法。结果表明,涉及微分处理后的光谱变换,尤其是先进行开方(T8、T11)、对数(T6、T12)等变换后再进行微分处理,可提高其与土壤氮素的相关性。在引入较少因子变量个数的条件下,该方法使因变量解释量达到了98%。综合考虑模型的校正、验证效果及模型复杂度(模型最佳因子变量个数),可得出光谱平方根的一阶微分变换处理(T8)为最佳的土壤光谱变换算法。该条件下的土壤氮素的校正模型表现为R2=0.985、RMSEC=0.000132、Fn=6,验证模型的表现为R2=0.9853、RMSEV=0.000162,结果表明基于T8的光谱数据变换可实现本试验条件下土壤氮素的光谱估算。另外,可以考虑将原始光谱的一阶微分(T9)、对数和对数倒数的一阶微分(T6、T7)以及平方根和对数的二阶微分(T11、T12)作为光谱数据变换方法。本文研究结果可为土壤氮素估算和光谱数据预处理提供技术参考。
Spectral data transformation plays an important role in eliminating background, noise impact and extracting spectral features, and is an essential step in spectral data analysis. In order to study the effect of spectral transformation on the soil nitrogen PLSR model and choose the best spectral transformation method, 15 typical spectral transformations were performed on the original spectral data. By comparing the correlation between different transformation spectra and soil nitrogen, To achieve accurate diagnosis of soil nitrogen PLSR, and comprehensive evaluation of the best spectral data transformation method. The results show that the spectral transformation after differential processing, especially the transformation of T8 and T11, logarithm (T6 and T12), and then differential processing can improve the correlation with soil nitrogen. Under the condition of introducing a few variables, the method can explain the dependent variable to 98%. Taking the correction of the model, the verification effect and the model complexity (the number of the best model variables), the first-order differential transformation (T8) of the square root of the spectrum can be obtained as the best soil spectral transformation algorithm. Under this condition, the calibration model of soil nitrogen showed R2 = 0.985, RMSEC = 0.000132, Fn = 6, and the validation model showed R2 = 0.9853 and RMSEV = 0.000162. The results show that the spectral data transformation based on T8 can achieve the experimental conditions Spectral Estimation of Soil Nitrogen in Soil. In addition, first-order differential (T9), first-order differential (T6, T7) of logarithms and logarithms of the original spectrum, and second-order differential of square root and logarithm (T11, T12) can be considered as spectral data conversion methods. The results of this paper can provide technical reference for soil nitrogen estimation and spectral data pretreatment.