Aerial multispectral imaging for crop hail damage assessment in potato

Abstract

Crop loss assessment after an event of hailstorm can be inaccurate, subjective, and time consuming with the conventional method. Low-altitude, high-resolution aerial imaging using an unmanned aerial vehicle can be utilized for rapid assessment of target crops in a large scale, which can potentially improve the evaluation procedure. The goal of this study was to evaluate the feasibility of rapid and accurate assessment of crop damage due to simulated hailstorms using aerial multispectral imaging. Field experiments were conducted during two seasons in two potato varieties (Russet Norkotah, Ranger Russet) with three levels of mechanical defoliation (33%, 66%, and 99%) at three growth stages (tuber initiation, early bulk, and late bulk). All defoliation treatments were compared to the non-treated control plots (0% defoliation). Aerial multispectral images were collected between 77 and 108days after planting (0–60days after damage). Vegetation indices such as green normalized difference vegetation index (GNDVI), normalized difference vegetation index, and soil-adjusted vegetation index were calculated from replicate plots of different treatments. Results from two seasons showed similar trends in GNDVI values, with maximum effect of hail damage observed in early bulk stages. The mean GNDVI value was significantly lower in crops with the severe damage (99% defoliation) than others upon hail damage at the early bulk stage, with imaging after 10days after damage. The difference in GNDVI for crops with 33–66% damage could be detected within 10days after damage, and crop regrowth after that time period removed the effects of defoliation. The 99% defoliation damage at the early bulk stage also affected the crop yield significantly. Correlation analysis between vegetation indices and yield data indicated a strong relationship (r=0.77–0.90) for damage at the early bulk stage than other stages.