Analytic Models to Predict Root Structure Depth

摘要

Plant root systems absorb water and minerals and synthesize organic matter. The investigation and prediction of root system architecture (R.SA) can provide significant information that is potentially beneficial for promoting plant growth and reproduction. Existing approaches use manual sampling, which involves digging up the plant and examining the root. This process is destructive and time-consuming. Ground-penetrating radar has been used for exploring root structures of large plants, such as trees, but not for small plants due to resolution limitations. For this study, a finite element analysis (FEA) model was built to investigate the feasibility of using infrared imaging to predict root depth given the amount of heat flux required to obtain an image, the image acquisition time, and the thickness of the plant container. Polynomial regression, support vector machine, and artificial neural network models were designed to predict root structure depth based on the thermal profile of the structure over time derived from FEA model. Analysis results suggest that infrared imaging can be used to provide depth information of root structures. However, the thickness and complexity of the root structure impact prediction accuracy. Future directions include (1) development of image enhancement algorithms to improve detection capability and accuracy, and (2) conducting experiments to confirm the findings from the simulation.