Toward A Structural Description Of Row Crops Using Uas-Based Lidar Point Clouds

The determination of structural description of crops could contribute to precision agriculture applications, such as yield assessments. However, an efficient and reliable automatic system for evaluating key structural descriptors is still lacking. Unmanned aerial systems (UAS)-based light detection and ranging (LiDAR) offers relatively affordable high spatial and temporal resolution 3D data. In this study, we used the UAS-LiDAR system to collect 3D point clouds for 24 plots of snap bean across different seasonal growth stages. We extracted a digital elevation model (DEM) from ground return points, and then introduced a parameter to calibrate data from different flights. Based on the segmentation results, key structural descriptors, including canopy height, width, and leaf area index (LAI) were calculated and compared with in situ measurements. While canopy width showed the most uncertainty, the height evaluation was fairly accurate with a RMSE = 0.04m and R-2 = 0.72. LAI assessments also showed promise with a RMSE = 0.45 and R-2 = 0.43. These results bode well for extension to yield modeling and within-season management interventions.