Calibration of an Unmanned Aerial Vehicle for Prediction of Herbage Mass in Temperate Pasture

Abstract

Accurate estimation of herbage mass is crucial for managing pastoral livestock systems. The Normalized Difference Vegetation Index (NDVI) from Unmanned Aerial Vehicle (UAV) sensors shows promise for high-resolution estimations of pasture herbage mass, but it is still unknown how this method differs among forage species, seasons, and pasture management practices. A commercial sensor was calibrated to predict herbage mass using NDVI. Additionally, the effect of different forage species, days of regrowth, and nitrogen (N) status on the relationship between NDVI and herbage mass was evaluated. Two pastures of tall wheatgrass (Thinopyrum ponticum) and tall fescue (Festuca arundinacea), divided into 30 and 72 plots, respectively, were assessed during spring and autumn regrowth over two years in Balcarce, Argentina. Doses of 0, 50, and 100 kg N ha−1 were applied to tall wheatgrass, and 0, 50, 100, 200, 400, and 600 kg N ha−1 were applied to tall fescue to create variability in herbage mass and N status. Exponential regression models of herbage mass (y) fitted against NDVI (x) showed an average R2 of 0.83 ± 0.04 and a mean absolute error of 170 ± 60 kg DM ha−1. The relationship between NDVI and herbage mass differed (p ≤ 0.05) between species, seasons, and regrowth stage, but was not influenced by N status (p > 0.05). Results suggest that accurate predictions of herbage mass using NDVI measurements by an UAV require frequent model recalibrations to account for observed differences among forage species, days of regrowth, and years.