SPAD inversion of summer maize combined with multi-source remote sensing data

Abstract

The chlorophyll content is an important indicator of corn growth and yield. In order to improve the prediction accuracy of chlorophyll content, this study combines ground hyperspectral characteristic parameters (original spectral characteristics, first-order differential, characteristic spectral position), vegetation index calculated by multispectral, and effective plant height (Canopy Height Model, CHM) of crops, etc. Through correlation analysis of sensitive characteristics of chlorophyll content, the study uses multiple linear regression (MLR), partial least squares regression (PLSR), classification and regression tree regression (CART), and random forest (RF) to construct a summer maize SPAD inversion model. Then, the accuracy of the model was evaluated through the root mean square error (RMSE) and coefficient of determination ( R 2 ). The results show that the position of the red edge and the first-order differential values within the red edge Dr , CHM, SAVI, NDVI, RDVI, GNDVI, RVI, and DVI are significantly correlated with SPAD; the MLR model under a single data source is the best, the model’s R 2 is 0.8281, RMSE is 2.136; the RF model under multi-source data is the best. The model’s R 2 and RMSE are 0.9114 and 2.3955 respectively. The accuracy of the SPAD inversion model constructed based on multi-source data is better than that of a single data source. This study shows that the random forest model based on multi-source data can invert the SPAD of summer maize better. This method can provide theoretical support for summer maize growth monitoring and fine fertilization management. Keywords: Multi-source remote sensing, summer corn, Unmanned Aerial Vehicle, chlorophyll, machine learning DOI: 10.33440/j.ijpaa.20210402.174 Citation: Pan F J, Li W H, Lan Y B, Liu X G, Miao J C, Xiao X, Xu H Y, Lu L Q, Zhao J. SPAD inversion of summer maize combined with multi-source remote sensing data. Int J Precis Agric Aviat, 2021; 4(2): 45–52.