Hyperspectral analysis of total nitrogen in soil using a synchronized decoloring fuzzy measured value method

Abstract Visible and near-infrared reflectance spectroscopy is a rapid, effective, and non-destructive method for monitoring the total nitrogen (TN) content of soil. However, soil color decreases spectral reflectance and severely limits the accuracy of a TN content estimate. The aim of this study was to improve the TC content estimate accuracy by rapidly decreasing the effect of soil color using a synchronized decoloring fuzzy measured value (SDFMV) method. Visible/near-infrared spectra were acquired using an ASD FieldSpec 3 spectroradiometer under laboratory conditions. Hue–saturation–intensity color values were obtained from the reflectance spectra, and the optimal model (group H; j = 6.5) for the SDFMV method was used to develop the final model, which performed better (root mean square error of validation 0.137 g/kg, mean relative error of validation 9.089 %) than a partial least squares regression model. The SDFMV method has great potential for accurately and remotely determining TN contents, and improves the range of methods available for producing spatial maps of soil nutrients.