Effects of different particle sizes on the spectral prediction of soil organic matter

Soil organic matter (SOM) is an important criterion for soil quality. The rapid determination of SOM can provide basic data support for the implementation of precision agriculture. Although the spectral reflectance has been used to rapidly obtain the SOM content, the method is not convenient to comparisons between different research results because the selection pretreatment criteria are not the same for soil samples. In this study, soil samples were prepared with six different standards of < 1 mm, 0.5-1 mm, 0.25-0.5 mm, 0.15-0.25 mm, 0.088-0.15 mm and < 0.088 mm. The reflectance of the different soil samples were compared, and the correlation between the reflectance of different soil samples and SOM were analyzed, as were the modeling and precision of the predictions of SOM. The results show that the soil spectral reflectance and the correlations between the uniform samples and SOM increased with decreasing particle size. Among the six kinds of soil particles, < 0.088 mm had the highest correlation with SOM, and the reciprocal correlation coefficient reached 0.84. In the process of single-variable and multivariate modeling, the mixed samples of < 1 mm and < 0.088 mm had a better prediction effect than the uniform sample. Although the prediction result of < 1 mm was slightly worse than that of < 0.088 mm, the modeling results were the same (coefficient of determination (R-2) of 0.78), and the validation set also showed a satisfactory predictive ability (R-val(2) = 0.77). The soil samples of < 1 mm, regardless of the ease of comparison with other soil preparation standards or the ability to predict SOM showed a high correlation and stable predictive ability. Therefore, when measuring the soil spectrum, < 1 mm would be ideal for the pre-treatment of soil samples.