3D characterization of crop water use and the rooting system in field agronomic research

Abstract

A new approach to produce 3D characterizations of crop water use and root activity in large on-farm GxExM sorghum experimentation was developed using electromagnetic induction (EMI) instrument coupled with an inversion algorithm. A DUALEM-21S sensor was applied to collect time-lapse and continuous apparent soil electrical conductivity (ECa) data across the study field cropped with 144 G´E´M combinations in three replicates. An artificial neural networks (ANN) model was developed to predict dynamic 3D soil moisture (qv) using the inverted depth-specific true soil electrical conductivity (σ, mS m-1) from the ECa. 3D soil water dynamics and crop water use were successfully predicted with good prediction agreement (LCCC = 0.81) and low prediction error (RMSE = 0.03 cm3 cm-3). The results indicate that the root activity was significantly affected by the depths, growing stages, water levels and plant population and their interactions. The approach appeared to be able to capture GxExM interactions based on the analysis of linear mixed models and recursive partitioning trees. In conclusion, the approach could provide a rapid and cost-efficient opportunity to characterise differences in crop water use and root activity across large field experiments. We also argue that the improved understanding of the 3D crop water use dynamics can help inform complex GxExM interactions in field research.