Study On Remote Sensing Inversion Model Of Rice Chlorophyll Under Heavy Metal Cadmium Stress

Heavy metal pollution in farmland has a serious impact on crop planting all over the world.Heavy metal cadmium stress will lead to the change of chlorophyll content in rice.Therefore,the establishment of spectral index sensitive to chlorophyll change is of great significance for remote sensing monitoring of large-area crop cadmium pollution.Although many spectral indices have been proposed to evaluate chlorophyll levels,the study of spectral indices for chlorophyll changes under cadmium pollution is still of great significance.This paper takes the late tillering rice in an agricultural area of Hunan Province as the research object.The natural environmental conditions such as temperature,humidity and climate in the study area are the same,and the content of heavy metal cadmium in soil is the main factor affecting the growth of rice.By collecting the measured rice and soil research data,analyzing the change law of rice spectral data under the influence of cadmium stress,and establishing the relationship model between spectral parameters and chlorophyll value,the quantitative inversion of rice chlorophyll by remote sensing is realized.The main conclusions of this paper are as follows:（1）The agricultural land in the study area is polluted by cadmium in varying degrees.With the increase of cadmium content in soil,the chlorophyll content in rice decreases,resulting in the change of spectral information.The green peak reflectance in the visible band increased by 7%,the spectral reflectance in the near-infrared band decreased by 37%,and the position of the red edge moved to the short wave direction by 10 nm.（2）Calculate and extract the vegetation index of measured hyperspectral and multispectral data.In this study,a new spectral index NTI（R_i,R_j,R_k）=（R_j-R_i+C）/（R_j+R_k）was established.More spectral information can be obtained from three bands to identify the subtle changes of chlorophyll content under cadmium stress.19 vegetation indexes such as CI_RE,VOG,TCARI and MTCI commonly used in the existing research are selected for comparative analysis.Among the commonly used vegetation indexes,CI_RE has the highest correlation with chlorophyll content,with a correlation coefficient of 0.801,while the correlation coefficient between the new index NTI(R₇₄₈,R₇₇₈,R₇₄₉)and chlorophyll content can reach 0.853,with a better correlation.（3）The regression model between vegetation index and chlorophyll content of rice was established.Comparing the statistical models of chlorophyll content between the new index and other vegetation indexes,it is found that the new index NTI has the best prediction ability for chlorophyll content,with R² of the quadratic model of 0.795,while the best statistical model of the existing vegetation index is CI_RE’s quadratic model,with R² of 0.688.In order to further illustrate the prediction effect of the new spectral index,four sensitive vegetation indexes CI_RE,VOG,TCARI₂ and MTCI are selected as independent variables and chlorophyll content as dependent variables to establish a random forest regression model.The fitting effect of the modeling set model R²=0.723.The results show that the R²of the estimation model established by NTI is 8.4%higher than that of the random forest model.（4）The new index is applied to Sentinel-2 multispectral remote sensing images.Through the linear correction of Sentinel-2 multispectral data through hyperspectral data,the inversion model of rice chlorophyll content of remote sensing spectrum is constructed.Compared with other indexes,NTI’s quadratic model has the best prediction ability,R² is0.582,followed by CI_RE’s power function model,R² is 0.524.Compared with the spatial prediction results of chlorophyll content of other indexes and the spatial distribution of cadmium content in the study area,the inversion effect of the new index on chlorophyll content of rice under cadmium stress in Sentinel-2 image is more accurate.The new index NTI based on hyperspectral data of rice under cadmium stress can better identify the changes of chlorophyll content of rice under cadmium pollution than other existing vegetation indexes.