Quantitative Prediction Study of Climate-sensitive Potassium with Hyperspectrum in Arid and Semi-arid Region of Northwestern China

Soil potassium content in arid and semi-arid region can reflect the conditions of the paleoclimate and it can be inverted by soil spectra. The relationship between soil spectra and soil potassium content was discussed in this research. Based on four reflectance transformations, single-variance analysis and multi-variances inversion model were built to invert potassium content. The results of single-wavelength inversions were very significant except for the reflectance model. The multi-variances models were good and accurately (R² (determination coefficient) >0.674 and RMSE (root-mean-square error) <0.09). Then, the sensitive wavelengths of the potassium were chosen by using the higher correlation coefficients. The results of this study showed that both methods have a great potential for predicting soil potassium content. The sensitive wavelengths of the potassium content were at 2200–2300 nm which could be illustrated by the potassium-bearing minerals spectral absorption features.