土壤养分的红外衰减全反射与漫反射光谱同步测量方法

发布时间：2018-06-05 22:52

本文选题：衰减全反射 + 漫反射　；参考：《光谱学与光谱分析》2017年11期

【摘要】：光谱学分析技术因具有快速、实时、无损等优势,被广泛用于土壤养分检测等各个领域,也是现代农业检测控制中广泛应用的一种有效方法。实验采集不同养分含量的土壤样本,分别获取中红外衰减全反射(ATR)和漫反射(DRIFT)光谱数据。通过不同的光谱预处理算法结合偏最小二乘法(PLS)建立土壤中全碳、全氮、碱解氮的定量分析模型,对土壤养分的ATR和DRIFT光谱分别进行独立模型和叠加模型的分析。并以土壤养分中的全碳和全氮为例,发现经过标准正态变量变换处理的两种光谱建立的PLS模型相关系数均有所提高。全碳的两种光谱模型校正集的R~2分别提高至0.826和0.919,而全氮两种模型校正集R2分别提高至0.841和0.928。我们进一步对全碳、全氮各自叠加后的光谱进行分析,模型的R~2分别提高至0.942和0.951。用同样的方法分析土壤中的碱解氮,联用两光谱后的R~2提高至0.919。研究结果表明,两种光谱的联合使用,为提高模型的相关系数和预测能力提供了一种合理有效的方法。
[Abstract]:Spectroscopic analysis technology is widely used in many fields such as soil nutrient detection because of its advantages of fast real-time and nondestructive. It is also an effective method widely used in modern agricultural detection and control. Soil samples with different nutrient contents were collected and the spectra of ATR and DRIFT were obtained respectively. The quantitative analysis models of total carbon, total nitrogen and alkali-hydrolyzed nitrogen in soil were established by using different spectral pretreatment algorithms and partial least square method. The ATR and DRIFT spectra of soil nutrients were analyzed by independent model and superposition model, respectively. Taking the total carbon and total nitrogen in soil nutrients as examples, it was found that the correlation coefficients of the two PLS models established by the standard normal variable transformation were improved. The R2 of the two spectral model correction sets for total carbon was increased to 0.826 and 0.919, respectively, while the total nitrogen model correction set R2 was increased to 0.841 and 0.928 respectively. We further analyzed the spectra of total carbon and total nitrogen after superposition, and the R2 of the model was increased to 0.942 and 0.951 respectively. The alkali-hydrolyzed nitrogen in soil was analyzed by the same method. The results show that the combined use of the two spectra provides a reasonable and effective method for improving the correlation coefficient and prediction ability of the model.
【作者单位】：山西农业大学;北京农业智能装备技术研究中心北京市农林科学院;
【基金】：重点研发计划(2016YFD0800901)资助
【分类号】：O657.33;S158

【相似文献】