建模法在土壤有机质的近红外光谱分析中的研究进展  

Research Progress of Modeling Method in near Infrared Spectroscopy of Soil Organic Matter (SOM)

作　　者：虞丹尼[1] 张小艳 黎司[1] 江忠远[1] 敖克厚[1] 高智席[1] 

机构地区：[1]遵义师范学院化学与化学工程学院,贵州遵义

出　　处：《分析化学进展》2016年第4期91-95,共5页Advances in Analytical Chemistry

基　　金：贵州省教育厅自然科学研究项目:黔教合KY字([2013]172号)资助;贵州省科学技术基金:黔科合J字LKZS[2014]02号、15号,黔科合LH字[2015]7037资助;遵义师范学院博士基金:2013BJ03,2013BJ02资助。

摘　　要：近红外光谱(NIRS)广泛应用于土壤有机质(SOM)研究中,但在我国,大多数研究人员在应用NIRS时一般采用仪器自带软件进行光谱预处理和数据建模来预测SOM、土壤水分、速效N和速效P的含量,这对光谱数据处理的能动性有一定的限制。本文初步论述了光谱预处理方法——小波分析、波长选择方法——遗传算法及估测土壤养分的建模方法——神经网络建模法在土壤有机质分析中的应用,并对这些方法在我国SOM的含量预测方面的应用前景进行了展望。Near infrared spectroscopy (NIRS) is widely used in the study of Soil Organic Matter (SOM). However, most researchers in China in the application of NIRS in general use its own software to carry out the spectral preprocessing and data modeling to predict SOM, soil moisture, available N and available P content. So there are some limitations in the activity of spectral data processing. In this paper, the method of spectral pretreatment—wavelet analysis, wavelength selection method— genetic algorithm, and the modeling method of estimating soil nutrient—the method of neural network modeling in the analysis of soil organic matter are discussed. Also, the application prospect of these methods in the prediction of SOM in our country is discussed.

关 键 词：近红外光谱 土壤有机质 建模 进展 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]