A buckwheat experiment compared broadcast sowing with four drill spacings and RGB-UAV imagery at 30, 50, and 60 m. Vegetation indices correlated strongly with biomass and yield under selected conditions, but no index, altitude, or growth stage was universally optimal; 50 m was most stable in this experiment.
Key findings
- Drilling at 12.5 cm produced the highest fresh weight; seed weight was higher for every drill treatment than broadcast sowing, while seeds per plant peaked at 40 cm, showing a density trade-off. VI-ground correlations were r=0.82-0.98 but varied by stage, sowing, and altitude. The 50 m altitude was most stable, and ExG/GLI were more consistent than NGRDI.
Why this matters globally
The results support lower-cost RGB precision agriculture and warn that models transferred across fields require adjustment for canopy structure and management.
Thai researcher contribution
Kasetsart agronomy and genetics researchers linked sowing design with remote phenotyping in an international collaboration; the abstract does not identify the experimental field as being in Thailand.
Limitations to consider
One cultivar and limited conditions constrain generalization. Within-experiment correlations are not independent predictive validation and may partly reflect treatment separation. Three altitudes, three indices, and two stages do not cover varying illumination, sensors, soils, seasons, or cultivars.