Study On Leaf Spectral Characteristics And Chlorophyll Content Retrieval Of Reclaimed Vegetation In Rare Earth Mining Area

Rare earths are mainly used in the field of cutting-edge technology and military industry,and are strategic national resources.Ganzhou is the main production area of ion adsorption rare earth ore,rare earth mining for a long time its leaching process,dissecting the original surface and injecting a large amount of leaching liquid into the soil,resulting in the destruction of surface vegetation in the mining area,causing soil erosion,soil acidification,sanding and other environmental problems,natural recovery of vegetation is difficult,mainly relying on artificial reclamation.Due to the change of soil physical and chemical properties caused by rare earth mining,the growth of reclaimed vegetation is difficult and extremely poor,which seriously restricts the ecological environment recovery of reclaimed mining areas.As the vegetation is disturbed by the outside world,the change of its own structure makes the spectral curve change locally,which can be used to judge the growth of vegetation.Therefore,hyperspectral technology is widely used to monitor the growth condition under crop stress,and has shown great potential in the monitoring of vegetation stress in reclaimed rare earth mining areas.In this study,we obtained the spectral data and chlorophyll content of reclaimed vegetation and normal vegetation in rare earth mining areas,analyzed the spectral characteristics under different environmental stresses,and extracted the spectral characteristics of subtle differences using various transformation methods,and correlated them with chlorophyll,combined with various linear and nonlinear mathematical methods to construct and compare the reclaimed vegetation chlorophyll estimation models,and screened out the best estimation model suitable for reclaimed vegetation.The results of the study are as follows.(1)Spectral control analysis of the vegetation under different environmental stresses revealed that the spectral reflectance of the reclaimed vegetation in the visible range was significantly higher than that of the normal vegetation in the original spectrum,and the reclaimed vegetation showed red shift in the "red valley" position.In the first-order derivative spectra,the changes in the derivative spectra under different environmental stresses were more or less the same,and there were only large differences at 520 nm and 700 nm wavelengths.In the wavelet transform,the wavelet coefficient values of vegetation curves under different environmental stresses appeared to intersect at the d5 scale,which could clearly distinguish the stress state.The fractional order differential treatment made the reflectance of reclaimed vegetation substantially exceed that of normal vegetation with great difference,and the difference between them gradually decreased with the increase of fractional order.(2)The correlation between each spectral data and chlorophyll content was analyzed,and sensitive spectral information was screened for six vegetation species in the reclaimed land of the mine.Except for wetland pine,the sensitive spectral characteristics parameters of other reclaimed vegetation were distributed in the original and first-order derivative spectra,fractional-order differential,and wavelet transform,and the main spectral characteristics were distributed around 550 nm,690 nm,and 710 nm,which belonged to the visible and near-infrared ranges.The similarities and differences in sensitive spectral parameters of different vegetation.(3)The vegetation indices,PLSR and random forest algorithm were used to construct models for estimating chlorophyll content of reclaimed vegetation.The five vegetation indices with the highest correlation were selected among 43 vegetation indices,and single vegetation index and multiple regression analysis estimation models were developed for each vegetation,with the highest estimation accuracy for heather and camellia.The random forest estimation model was the best model among the three estimation models for chlorophyll in reclaimed vegetation.Wetland pine had the lowest correlation and the worst prediction accuracy,which may be mainly due to the error of measurement data caused by leaf shape.