Ground based digital imagery for grassland biomass estimation

Abstract

Aboveground biomass was estimated on the shortgrass prairie of eastern Colorado using ground based conventional (RGB) digital camera imagery. The accuracy and efficiency of image‐based estimates were compared with clipped biomass measurements. Field measurements of aboveground biomass were obtained on three grazing treatments and three sample dates (phenological status). Grazing treatments did not significantly affect (p>0.10) estimates of clipped green biomass taken from digital images. However, plant phenology, green biomass estimates from images, and the interaction of plant phenology and green biomass estimates from images significantly affected clipped green biomass measurements (p≤0.04). Analysed images provided fair estimates of total clipped green biomass (R2 = 0.55) and clipped green biomass without cactus (R2 = 0.73) when plant phenological status was included in the models. When plant phenology was removed from the models, the variability explained by green biomass estimates from images declined to 25% for clipped green biomass, and 32% for clipped green biomass without cactus. Thus, results showed that plant phenological status was the most important variable in the prediction of green vegetation biomass. Results indicated that the usefulness of RGB digital camera imagery for green biomass estimation is limited for the shortgrass prairie.