Hyperspectral inversion of soil organic matter content in cultivated land based on wavelet transform

Soil organic matter (SOM) is one of the most important indicators of cultivated land fertility and greatly influences other soil nutrient factors and physicochemical characteristics. This study aimed to develop a universal method to detect SOM content within the plough layer of cultivated land using ground hyperspectral data. The hyperspectral data was decomposed using the wavelet transform algorithm. The sensitivity of the high-frequency information increased with the degree of the wavelet decomposition. SOM content was retrieved using the high-frequency coefficients created with the wavelet transform and random forest algorithm. The validation model showed a R-2 of 0.748 and RMSE of 0.254. The predictive accuracy of the model based on the random forest algorithm was improved by 10.2% compared to that of the math transformations. Therefore, the high-frequency information decomposed by the wavelet technology effectively enhanced the predictive accuracy of the SOM content by coupling the wavelet technology and random forest algorithm.