基于特征波长积分算法的土壤有机质抗水分干扰模型研究

土壤有机质是土壤的重要成分，也是农作物生长的重要营养指标。快速、准确检测土壤有机质含量对施肥管理具有重要意义。近年来，近红外光谱被广泛应用于土壤有机质的快速检测，然而土壤有机质敏感波段易受土壤水分干扰，从而会影响到土壤有机质的预测结果。在山西省境内采集了140个土壤样本，采用ASD光谱仪分别获取了不同含水率（0%，5%，10%，15%，17%）土壤样本谱图曲线（350～2 500 nm）。为了提高土壤有机质预测模型精度，提出特征波长积分算法，即通过特征波长处吸光度的积分值作为自变量进行建模的方法，建立了土壤有机质预测模型及抗水分干扰修正系数模型。结果表明：（1）使用特征波长处吸光度的积分值作为自变量建立的土壤有机质预测模型统计参数优于传统的使用特征波长处的吸光度值作为自变量的建模方法；（2）校正后的湿土光谱更加接近干土土样，在一定程度上解决了传统水分修正系数在水分含量较高时修正效果较差的问题；（3）提高了湿土样本的预测精度，预测相关系数（R_P）提升了约0.09，预测均方根误差（RMSEP）降低了约1.72 。说明该方法可以有效降低水分对土壤有机质光谱的影响，提高不同含水率土壤有机质的预测精度，可为后续仪器开发提供理论支持。

Research on Anti-Moisture Interference Soil Organic Matter Model Based on Characteristic Wavelength Integration Algorithm

As an important component in soil, soil organic matter (SOM) is a critical nutrition index in the process of crop growth. Rapid and accurate detection of SOM content is of great significance for the fertilization management. In recent years, NIR has been widely used in the rapid detection of SOM. However, soil moisture is one of the important factors that affect the prediction results of SOM. In this study, 140 soil samples were collected in Shanxi Province, and the spectral information with different water content (0%, 5%, 10%, 15%, 17%) was collected by ASD spectrometer (350~2 500 nm). In order to improve the accuracy of the SOM prediction model, a characteristic wavelength integration algorithm (taking the integral absorbance value at characteristic wavelength as the independent variable) was proposed. The results show that: (1) the statistical parameters of the SOM prediction model established by this algorithm are better than the traditional characteristic wavelength modeling method; (2) the moisture correction model established by this algorithm can eliminate the influence of moisture, and the corrected spectra of wet soil samples are closer to the corresponding dry soil samples; (3) the prediction accuracy of wet soil samples is improved. The RP increased by about 0.09 and RMSEP decreased by about 1.72. The results show that the method can effectively reduce the influence of soil moisture on the spectral characteristics of SOM, improve the prediction accuracy of SOM with different water content, and provide theoretical support for the subsequent instrument development.