A new method of searching for concealed Au deposits by using the spectrum of arid desert plant species

With the increase of exploration depth, it is more and difficult to find Au deposits. Due limitation time cost, traditional geological methods are becoming increasingly be effectively applied. Thus, new ideas urgently needed. This study assessed feasibility effectiveness using hyperspectral technology prospect for hidden For this purpose, 48 plant (Seriphidium terrae-albae) soil (aeolian gravel desert soil) samples were first collected along a sampling line that traverses an mineralization alteration zone (Aketasi mining region in arid China) used obtain contents by chemical analysis method reflectance spectra plants obtained with Analytical Spectral Device (ASD) FieldSpec3 spectrometer. Then, corresponding relationship between content anomaly concealed deposits was investigated. Additionally, characteristic bands selected from four different methods, namely, genetic algorithm (GA), stepwise regression (STE), competitive adaptive reweighted (CARS), correlation coefficient (CC), then input into partial least squares (PLS) construct model estimating content. Finally, quantitative 15 transformation established PLS method. The results compared those based on full spectrum. indicate location can predicted geochemical information, feasible effective use spectrum estimate soil. cross-validated determination (R2) ratio performance deviation (RPD) value measured reached maximum 0.8218 2.37, respectively, minimum 6.56 µg/kg root-mean-squared error (RMSE) model. However, process modeling, crucial select appropriate as parameter Compared GA, STE, CC CARS superior band screening accuracy complexity When modeling bands, highest accuracy, R2 0.8016, RMSE 7.07 µg/kg, RPD 2.20 when 56 transformed (1/lnR)’ (where represents derivative reciprocal logarithmic spectrum) sampled inversion replace constructing proposes deposits, which may have promising application prospects because its simplicity rapidity.