A comparison of visual and photographic estimates of weed biomass and weed control

The objective of this study was to compare the consistency and accuracy of visually estimated weed biomass and weed control data to data obtained through image analysis. Weed biomass and weed control were evaluated in soybean herbicide efficacy trials conducted at the University of Nebraska-Lincoln during 1992 and 1993. Measurements were based on visual estimates and on aerial. photographs taken at a height of 3.5 m above the soil surface. Photographs were digitized and classified, producing pixel values for broadleaf weeds, grass weeds, soybean, and soil. Percent weed cover was calculated in relation to the crop canopy, based on the respective number of pixels per image. Visual and photographic ratings of weed biomass and of weed control were not closely correlated. In the first year the visual method discriminated between more treatments than the photographic method, but the opposite occurred in the second year. The photographic method predicted yield more closely than the visual estimates. We concluded that visual estimates were less consistent and more subject to observer bias than measurements obtained with the photographic method.