Weed Detection in Field Corn Using High Resolution Multispectral Digital Imagery and Field Scouting
Weed species occur in non-uniform Patches across agricultural fields with the amount of patchiness differing among weed species and field. This patchiness complicates herbicide recommendations; however, herbicide applications can be targeted to specific areas by identifying the locations and weed species in the field. The size, shape, and location of weed infestations can be determined by intensive field scouting, but this approach is expensive. Remote sensing information, combined with ground-truth data, may provide a useful method to solve this problem. A study was conducted to determine the feasibility of using remote sensing techniques to detect weed populations at several stages of corn growth. A charge-coupled device (CCD) with four spectral filters mounted in an airplane was used to obtain several near digital images with 1 m * 1 m resolution of a 65 ha no-till corn field from May through September 1997. The CCD sensor contained four spectral filters sensitive in the blue, green, red, and near infrared (NIR) wavelengths. Latitude and longitude coordinates of the field perimeter were integrated into a geographical information system so that coordinate of anomalous areas of the field could be identified. Using the coordinates of the anomalous areas, ground scouting from the aerial images was conducted to define the nature of the anomaly. Comparing the aerial images to the ground-truthed data indicated that the NIR wavelength showed the greatest differences between corn and weeds. A normalized difference vegetative index was generated using the image corresponding with maximum green canopy cover and was highly correlated with corn yield, indicating that future yield predictions may be possible using remote sensing, however, field scouting was necessary to distinguish weed species and densities. Remote sensing combined with ground scouting provided an excellent method to determine the location of weed infestations over an entire field, to create a database for site specific herbicide management, and to monitor changes in weed species density over time.
DOI of Published Version
American Society of Agronomy
Broulik, Brian L.; Dalsted, K. J.; Clay, S. A.; Clay, D. E.; Carlson, C. G.; Ellsbury, M. M.; and Malo, D. D., "Weed Detection in Field Corn Using High Resolution Multispectral Digital Imagery and Field Scouting" (1999). Agronomy, Horticulture and Plant Science Faculty Publications. 227.