A random forest model for estimating Canopy Chlorophyll Content in rice using hyperspectral measurements

Abstract

Accurate estimation of the canopy chlorophyll content of a crop is essential for crop production. Ground-based hyperspectral datasets were obtained under a wide range of plant and environmental conditions in Jilin using Analytical Spectral Devices(ASD) spectroradiometers, and canopy chlorophyll content in canopy were measured by Soil and Plant Analyzer Development(SPAD)-502. The objective of this study is to determine the most suitable input variables to estimate the canopy chlorophyll content by Random Forest model. On the basis of a comprehensive analysis of the spectral data, the RF model is explored to provide an accurate and robust assessment of Canopy Chlorophyll Content(CCC). The correlation coefficient (R2) of the second RF model between the measured chlorophyll content and the predicated chlorophyll content is 0.82, and the root mean square error (RMSE) is 12.5738, which is better than the first RF model and the other indexes.