论文部分内容阅读
近红外光谱(NIR)技术在线检测土壤养分含量是精细农业研究热点之一.土壤水分在近红外波段具有较高的吸收系数,处于波动状态的土壤水分,直接影响NIR技术检测土壤有机质含量的预测精度.四川、重庆地区紫色土中有机质含量较低,比土壤水分最大有一个数量级的差异.利用BRUKER公司的MPA傅里叶近红外谱仪对不同含水率的紫色土壤进行了近红外光谱扫描,分析水分对近红外光谱检测紫色土壤有机质含量模型精度影响,提出了一种消除土壤水分对紫色土壤有机质含量预测影响的方法.近红外光谱扫描显示,在同一有机质含量水平下,含水率增加,光谱吸光度呈非线性上升趋势,直接利用光谱数据建立的偏最小二乘法(PLS)模型,在较高水分条件下,模型精度低;在遗传算法选择的波数区域,用二次多项式拟合紫色土壤含水率指数对近红外扫描光谱吸光度影响,其相关系数在0.95~0.97之间;将一定含水率下的近红外光谱吸光度折算为干土条件下吸光度,再利用PLS法建立近红外光谱有机质含量预测模型,与原有模型相比预测精度得到显著提高.实验结果表明,本方法可以有效消除紫色土壤水分对近红外光谱检测其有机质含量预测影响,为不同水分条件下近红外光谱在线检测紫色土壤有机质含量提供一定的理论和技术支持.
Near-infrared spectroscopy (NIR) technology is one of the hottest topics in fine agriculture for on-line detection of soil nutrient content. Soil water has a high absorption coefficient in the near infrared band and soil moisture in fluctuating state has a direct impact on the prediction of soil organic matter content by NIR technique Accuracy.The purple soil in Sichuan and Chongqing areas had lower organic matter content and an order of magnitude difference than the soil moisture.The near-infrared spectrum of purple soil with different water content was scanned by BRUKER’s Fourier transform infrared (NIR) The effect of moisture on the prediction of organic matter content in purple soil by near infrared spectroscopy was analyzed, and a method to eliminate the effect of soil moisture on the prediction of organic matter content in purple soil was proposed. Near infrared spectroscopy showed that under the same organic matter content, The absorbency showed a non-linear upward trend. The PLS model was directly established by using spectral data. Under high water conditions, the accuracy of the model was low. In the wave number region selected by genetic algorithm, the quadratic polynomial was fitted to purple soil water The influence of rate index on the absorbance of near-infrared scanning spectroscopy has a correlation coefficient of 0.9 5 ~ 0.97; Near-infrared spectral absorbance at a certain moisture content is converted to absorbance under dry soil conditions, and PLS method is used to establish prediction model of organic matter content in near-infrared spectra, which improves the prediction accuracy significantly compared with the original model. The results show that this method can effectively eliminate the effect of purple soil moisture on the prediction of organic matter content by near infrared spectroscopy, and provide theoretical and technical support for the on-line determination of organic matter in purple soil by near infrared spectroscopy under different water conditions.