Hyperspectral inversion of soil arsenic content in polymetallic mining areas based on optimized spectral index combined with PLSR
Arsenic(As)is a prominent contaminant within polymetallic mining areas in China,posing substantial threats to the environment,agriculture,and human health.Near-ground hyperspectral technology,characterized by its rapid,dynamic,non-destructive and high spectral resolution,holds significant potential for the monitoring and integrated management of soil arsenic pollution in polymetallic mining areas.However,the applicability and accuracy of hyperspectral inversion models are subject to variations influenced by factors such as contaminated areas,soil background,hyperspectral quality,and spectral inputs.This study focused on a polymetallic mining area in southern Hunan,utilizing Pearson correlation analysis in conjunction with variable projection importance(VIP)criteria,we extracted univariate spectral bands under 18 transformed spectral forms,as well as optimized spectral indices under 4 spectral index algorithms,as spectral input variables.These variables were then utilized to construct a partial least squares regression(PLSR)model to achieve the inversion of soil As content within the mining area.The results show that,there are high correlations between transformed spectral data(reciprocal(RT),logarithmic(L),square root(Sqrt),second derivative of standard normal variables(SNV_SD),etc)and As content.The optimized spectral indices reveal the spectral response characteristics of As in a two-dimensional spectral space,and the PLSR model constructed with the optimised spectral indices has better performance compared to the univariate characteristic bands.The ratio index(RI)model,whose R2c,RMSEc,R2p,RMSEp and RPD are 0.908,50.8 mg/kg,0.949,35.6 mg/kg and 4.45,respectively,emerges as the optimal model for the inversion of soil As content in the study area in this study.The combination of univariate characteristic bands with optimized spectral indices demonstrates favorable feasibility in predicting soil As content,providing a scientific foundation for the rapid monitoring of soil As pollution in polymetallic mining areas.
soil heavy metalsarsenichyperspectralremote sensingspectral transformoptimised spectral indicespartial least squares regression(PLSR)
NETL
NSTL
万方数据
