Monitoring available phosphorus content in soil of cultivated land based on hyperspectral technology

摘要

The study aimed to develop a universal method to monitor soil APC by hyperspectral data. The correlations between soil APC and the hyperspectrum reflectivity and its mathematical transformations were analyzed. The feature bands and its transformations were screened to develop the optimizing model of monitoring soil APC based on the method of multiple linear regression. The in-situ testing samples were used to evaluate the accuracy of the model. Results showed that the correlations of first order differential could reach around 0.6, especially for visible bands. Differential transformation was helpful for reducing the noise interference to the diagnosis ability of the target spectrum. The first order differential of reciprocal transformation was determined as the optimal inversion model for soil APC, of which the corresponding sensitive bands were 472, 967, 1062, 1211and 1402 nm. The in-situ testing samples were used to evaluate the accuracy of the model. The inversion model of soil APC in the cultivated land with the one differentiation of reciprocal transformation of hyperspectral could reach high accuracy with good stability.