Hyperspectral Inversion Of Heavy Metals In Mining Area And Its Response To Concentration

In recent years,the contradiction between human activities and land use has become increasingly prominent.Mining activities not only promote the rapid development of Chinese economy,but also make a large number of heavy metal elements penetrated into the soil.As a result,there are many abandoned mining lands that unable to engage into production,which has intensified the contradiction between human being and land useagain.At present,the traditional method for detecting heavy metals in soil has been unable to meet the demand at this stage.In order to alleviate this contradiction,realize the quality and ecological restoration of regional soil heavy metals,and strengthen the monitoring and management of the"trinity"of the quantity,quality and ecology of land resources,it is essential to measure the soil heavy metal quickly and accuratelyHyperspectral technology has gradually become mature and has been widely used in various fields of soil science.Hyperspectral technology has become a research hotspot for the rapid detection of soil components.However,compared with other soil index inversion,hyperspectral soil heavy metal hyperspectral inversion research started late,and related technical methods still need to be further improved.On the one hand,the accuracy of the model is low,and the inversion system is not perfect.On the other hand,the influence of key factors such as iron oxide and heavy metal concentration on the inversion effect is not fully understand and considered in the inversion process.These have become a bottleneck restricting the development and application in soil heavy metal content estimation using hyperspectral technology.To address this issue,In this paper,the spectral characteristics of heavy metals were extracted by spectral transformation methods such as first-order differential(FDR)and second-order differential(SDR).Then,the modeling methods such as partial least square model(PLSR)and BP neural network(BP)were used to further refine and improve the soil hyperspectral inversion system of heavy metals.The shortcomings of the model in modeling band selection are optimized by continuous projection algorithm(SPA).Many studies have shown that iron oxide is an important factor in soil spectral inversion.In this paper,we set up the experiment of the concentration difference between iron oxide and heavy metal to explore the influence of soil iron oxide on the spectral reflectance of soil and heavy metal elements.At the same time,according to the concentration of ferric oxide,a new modeling group was formed by gradient experiment and mining area samples,in order to avoid the influence of iron oxide concentration on the hyperspectral inversion of soil heavy metals and improve the inversion effect of the modelTherefore,the gradient experiment of heavy metal and iron oxide concentration was set up to explore the influence of the spectral characteristics of heavy metals in iron oxide.Thereby,the inversion ability of the model is optimized,and the spatial prediction of regional heavy metal concentration is realized.The specific research conclusions are as follows:1.Pearson correlation coefficient(r)was used to express the degree of correlation between heavy metal elements and spectral reflectance.Then the spectral characteristic band of heavy metal was selected.The four modeling methods,including the partial least square model(PLSR),BP neural network(BP),support vector machine(SVM)and random forest method(RFR),were used to estimate the soil heavy metal content.The determination coefficient(R2),root mean square error(RMSE)and relative analysis error(RPD)were used to evaluate the prediction accuracy of the above four models.The results showed that the differences of spectral transformation methods and modeling methods in heavy metal inversion effects.The optimal inversion models of Cu,Cr and Zn were PLSR based on SDR,PLSR based on CR and PLSR based on SDR,respectively.Each model can achieve quantitative inversion of regional soil heavy metals with certain accuracy.In addition,the successive projections algorithm(SPA)was used to optimize the spectral characteristic band,thereby realizing the optimization of the heavy metal inversion model.The inversion accuracy and model stability of heavy metal spectral inversion model were enhanced after application of SPA.The R2 and RPD of the model prediction set were obviously improved with greatly reduced RMSE.2.The control variable method was used to set different gradient experiments with iron oxide and heavy metal concentrations.At the same time,the continuous removal(CR)was used to expand the spectral information,which to explore the spectral characteristic bands of iron oxide and different heavy metals,and the influence of different concentration of iron oxide on the spectral characteristics of heavy metals.The results showed that with the increase of iron oxide content,the spectral reflectance decreases from 350 nm to 1100 nm gradually,and there was no obvious rule between the spectral reflectance and iron oxide content from 1100 nm to 2500 nm.The concentration of iron oxide affected the spectral characteristics of heavy metals in the whole band(400-2500 nm).In the visible short wave near infrared band(400-1000 nm),iron oxide mainly covers the spectral reflection peak of heavy metal ions,and in the long wave near infrared band(1000?2500 nm),with the increase of iron oxide content,the change of absorption valley and reflection peak is decreased,which covers up the weak spectral transformation of heavy metals.3.The relationship between iron oxide in soil and heavy metal concentration was analyzed quantitativelyusing correlation analysis and Pearson correlation coefficient.The results showed that the correlation degree of iron oxide in the whole band(400-2500 nm)was significantly higher than that of heavy metal,indicating that iron oxide was the main influencing factor.The Pearson coefficient of iron oxide and heavy metal was partially coincident,so the concentration of iron oxide can affect the expression of heavy metal spectral characteristics.4.According to the concentration of iron oxide,the modeling group was brought into the hyperspectral inversion model of heavy metals based on the gradient experimental samples and the soil samples from the study area.Iron oxide has always been one of the main factors affecting the hyperspectral quantitative inversion of heavy metals in soil.The results show that too little or too much iron oxide can reduce the inversion effect of heavy metal model.When the content of iron oxide is too low,the aggregation and adsorption of heavy metals were weakened in soil;while the content of iron oxide is too much,the spectral characteristics of heavy metals in soil will be covered by iron oxide.Figure[20]table[13]reference[80]...