Construction Of Soil Heavy Metal Analysis Model Based On High Resolution Remote Sensing

In this study,Guangdong Province was selected as the study area,and 75 samples were selected as the research object.The contents of heavy metals As and Hg in the soil were measured by atomic fluorescence spectrometry,and the spectral reflectance rate of the soil was measured by AvaField-3 high-precision spectroscopic geo-physical spectrometer.Spectral smoothing is performed on the spectral data,and spectral indices such as first-order reflectivity,second-order reflectivity,reflectivity,reciprocal reflectivity,logarithm of reflectivity,reciprocal logarithm of reflectance,and waveband combination are constructed on the basis of spectral smoothing.Through the correlation analysis between the metalloid As and heavy metal Hg measured in the laboratory,the spectrum characteristic bands of metalloid arsenic and heavy metals mercury in the soil were obtained,and the regression model was used to establish an inversion model of spectral reflectance and soil metalloid arsenic and heavy metals mercury content.Provide technical support for the large-area,fast and accurate access to soil metalloid arsenic and heavy metals mercury.The main results of the full text are as follows:?1?The arsenic content in the study area is 1.37-68 mg·kg^-1,and the mercury content is0.026-0.310 mg·kg^-1.The level of arsenic and mercury in some areas has already exceeded the national background values,showing obvious accumulation characteristics of heavy metals.?2?The spectral reflectance of the soil after smoothing was first derivative,second derivative,reciprocal,logarithm,reciprocal logarithm,and continuum removal transformation,which effectively enhanced the correlation between spectral variation and soil arsenic and mercury content.Spectral feature bands.Among them,the correlation between soil heavy metal arsenic and mercury content and the spectral variables after first-order differential treatment was the highest,and the correlation coefficients were 0.578 and-0.745,respectively.The reflectance continuum removal was smaller for the heavy metal arsenic than the first differential treatment method,and the correlation is 0.447,while the heavy metal mercury is slightly smaller than the first-order differential treatment by a second-order differential treatment with a correlation of 0.603.?3?The metalloid arsenic and heavy metal mercury contents measured by atomic fluorescence spectrometry in laboratory were respectively used to construct heavy metal inversion model with spectral index after combined with band pass,first-order reflectivity,second-order reflectivity,reflectivity,reciprocal reflectivity,logarithm of reflectivity,reciprocal logarithm of reflectance,reflectance continuum removal.The results of metalloid arsenic model construction show that the first-order derivative of the reflectance of the spectral index is combined with the bands of R_?966.869?and R_?2842.058?to construct a ternary cubic inversion model y=0.0107x³+0.1244x²-2.0097x+10.618 in 6 Among the spectral indexes,the best effect was obtained,and the degree of fitness was 0.785.The fitting degree R² of the metalloid arsenic content predicted by this model and the actual measured arsenic content was 0.938,and the RRMSE was 0.329.The result of the heavy metal mercury model construction shows that the first index derivative of the spectral index is an exponential inversion model constructed by combining the R_?587.705?with a 700-fold expansion and the R_?1373.49?with a 250-fold expansion.y=1.4112e^-0.48x Among the six spectral indexes,the best effect was obtained,and the fitting degree was 0.620.The fitting degree R² of the heavy metal arsenic content predicted by this model and the actual measured arsenic content was 0.770,and the RRMSE was 0.184.