Estimation Model Of Nutrients In Mine Tailings Based On Hyperspectral Remote Sensing Information

Vegetation restoration is the core measure of ecological restoration of mining sites,and the nutrient content of mine tailings is the basis for plant colonization.Quickly obtaining information on the nutrient content of mine tailings is particularly important for guiding the restoration of mining areas.In the determination of tailings,available phosphorus(OP),available potassium(AK),alkali hydrolyzed nitrogen(AN),and organic matter(SOM)are commonly used as indicators to measure the nutrient status of tailings.In recent years,hyperspectral remote sensing has provided a new technology for tailings nutrient monitoring with its rich spectral information.This paper uses Landsat series remote sensing data to study the evolution of typical iron ore areas in Tangshan City,Hebei Province from 2000 to 2018.Samples of typical iron ore tailing ponds were collected to determine nutrient content,and to obtain restoration plant leaf spectra and GF-5 image end-member spectra.Extract characteristic bands and construct various estimation models of ASD spectral information and GF-5 spectral information respectively,evaluate and analyze the accuracy of the models,and study the estimation accuracy of tailings nutrient content by different modeling methods and different spectral data processing.The main research results are as follows:Restoration plants are greatly affected by surrounding environmental factors,and the overall leaf spectral reflectance is low;in the ASD spectral information estimation model,the first-order differential,second-order differential,and continuum removal transformation can effectively improve the spectral data and the nutrient content of tailings Among them,the continuum removal spectrum processing has the most significant improvement effect;among the ASD spectrum information estimation models,the PLSR model has the highest accuracy.The model validation set R2 is greater than 0.8,and the RPD is greater than 1.7,which can effectively restore plant leaf spectrum and tail Quantitative estimation of mineral nutrient content;the accuracy of the MLR model is the second,and the model validation set R2 is in the range of 0.7-0.8,and the quantitative estimation ability of the tailings nutrient content is poor;the PCR tailings nutrient estimation model cannot realize the effective organic matter content Estimation: In the GF-5 spectral information estimation model,the MLR and PLSR estimation model validation set models R2 are both less than 0.7,and the RPD does not exceed 1.4.It is impossible to achieve accurate quantitative estimation of the tailings nutrient content,but the PLSR model estimates the data Consistent with the measured data and the transformation trend of the ASD spectral information estimation model,it can realize the rough estimation of the nutrient content of the tailings in a large area,which is of great significance to the monitoring of the ecological restoration effect of the mining area.Figure 37;Table 49;Reference 108...