PastureCast

PastureCast is a cyber-physical system that integrates proximal sensing, remote sensing, and data analytics for biomass estimation on grazing lands. PastureCast is inspired by the University of Missouri’s (MU’s) PaddockTrac, a proof-of-concept device, and mobile application linked to a website interface (MU Grazing Wedge), which measures plant height using ultrasonic sensors and provides biomass estimates through height-to-yield regression equations.

The MU Pasture Precision team has conducted beta testing of PaddockTrac with farmers across 15 states. These beta-testers have recognized the significant potential of a Cyber-Physical System (CPS) in grazing land management.

The PastureCast CPS proposed here transforms the existing system in two ways. First, it advances proximal sensing by exploring the addition of RADAR, VCSEL (vertical-cavity surface-emitting laser), and LiDAR sensors and rectifying the errors induced by vehicle dynamics to build a full ground-to-canopy-surface biomass model. This advanced equipment – to be tested at multiple locations and calibrated with actual cut-and-weigh data – will give the world's most refined vision of pasture biomass.

Second, PastureCast merges these proximal sensing data with satellite images into a data cube. PastureCast then explores the data cube using machine learning (and similar techniques) to create an accurate and dynamic prediction of pasture biomass. Data from proximal sensors are employed to calibrate a machine-learning model for estimating forage biomass from satellite imagery, as demonstrated in Figure 2. The outcome is a system that permits farmers to obtain forage biomass estimates more accurately and efficiently, eliminating the frequent need for ATV monitoring. Notably, PastureCast expands these predictions to pastures where on-the-ground sensing or sampling is challenging or not possible, thereby enhancing the scalability of biomass monitoring across diverse terrains.