Estimation Of Phosphorus Content In Corn Leaves Based On Hyperspectral Remote Sensing

Hyperspectral remote sensing technology is an important direction of precision agriculture research.High efficiency and low consumption crop nutrient growth monitoring technology is of great significance to precision crop management and rational fertilization.Hyperspectral remote sensing technology has the characteristics of high resolution,strong continuity and large amount of information,which can make real-time,rapid and nondestructive nutritional estimation of crops.In order to estimate the phosphorus content and nutrition status of maize leaves,this study took maize from Wuyuan,Inner Mongolia and Luoyang,Henan province in 2020-2021 as the research object,using hyperspectral remote sensing technology to collect leaf spectral information and corresponding agronomic parameters in the critical growth period of maize.The correlation between the optimal spectral index and the optimal wavelet function and leaf P content was analyzed.Finally,the estimation model of leaf P content of forest maize based on partial least squares regression and random was established.The phosphorus content in maize leaves increased with the increase of fertilizer application,and showed a similar pattern in different growth periods,and the phosphorus content in maize leaves began to decrease with the advance of growth periods.In the visible band,the spectral reflectance of maize leaves was not significantly different from that of different phosphorus levels.In the near infrared band range,phosphorus dosage had a significant effect on spectral reflectance curve of maize leaves.With the increase of fertilization level,spectral reflectance of maize leaves increased.The correlation between existing spectral index and optimized spectral index and phosphorus content in maize leaves was analyzed.It was found that the band optimization significantly improved the prediction ability of spectral index,and the correlation between optimized spectral index(RSI)and phosphorus content in maize leaves was the best,with the highest correlation coefficient reaching 0.61.Through the analysis of the optimal wavelet function and the correlation of corn leaf phosphorus content,continuous wavelet transform is better than the discrete wavelet transform,comprehensive consideration is based on continuous wavelet transform characteristics of corn leaf p estimating model accuracy is higher than based on optimization models for predicting the spectral index of maize leaf,The determination coefficient R~2and root mean square error RMSE of the optimal model validation are 0.67and 0.46,respectively.Compared with the optimal spectral index and the optimal continuous wavelet transform,the accuracy of the estimation model of phosphorus content in maize leaves was better by combining the original spectral reflectance with machine learning.The accuracy of the partial least squares regression model based on the original spectral reflectance is higher than that of the random forest model,and the R~2and RMSE of the optimal model are 0.76 and 0.38,respectively.By comprehensive comparison of the accuracy and error of each model,the partial least squares regression model based on the original spectral reflectance obtained better prediction results,which provided theoretical basis and technical support for the rapid and non-destructive monitoring of crop growth status,fertilization decision,yield estimation and other guidance for agricultural production.