Aerial Identification of Individual Cotton Plants

Authors: Westbrook, John; Eyster, Ritchie; Yang, Chenghai; Suh, Charles

Submitted to: National Cotton Council Beltwide Cotton Conference
Publication Type: Proceedings
Publication Acceptance Date: 2/10/2017
Publication Date: 5/16/2017
Citation: Westbrook, J.K., Eyster, R.S., Yang, C., Suh, C.P. 2017. Aerial identification of individual cotton plants. Proceedings of the 2017 Beltwide Cotton Conferences, January 4-6, 2017, Dallas, Texas. p. 547-553.

Interpretive Summary: Boll weevils can infest individual cotton plants as well as cotton fields. However, cotton plants may grow where seed has scattered along roadsides and waterways, and in fields following rotation from cotton production. Timely detection of these potential host plants across large regions is critically needed to expedite boll weevil eradication and prevent re-infestation in southern Texas. We acquired a temporal sequence of airborne multispectral images of individual cotton plants from vegetative to full maturity stages. Analysis (probability guided unmixing) of multispectral reflectance images revealed that the likely contribution of cotton plants to spectral reflectance values increased throughout the growing season, and exceeded 50% when plant height and plant width exceeded 50 cm (corresponding to the early fruiting stage of boll development). Application of a percent threshold of likely contribution of cotton plants to pixels of georeferenced spectral reflectance images will yield likelihood maps of estimated distribution of cotton plants. Timely identification of volunteer and regrowth cotton plants will aid eradication program personnel in locating and destroying these hostable plants.

Technical Abstract: Boll weevils can infest individual cotton plants as well as cotton fields. However, volunteer and regrowth plants may grow where seed has scattered along roadsides and waterways, and in fields following rotation from cotton production. Timely areawide detection of these potential host plants is critically needed to expedite boll weevil eradication and prevent re-infestation in southern Texas. We acquired a temporal sequence of airborne multispectral images of individual cotton plants from vegetative to open boll stages. Probability guided unmixing of multispectral reflectance images revealed that the likely contribution of cotton plants to spectral reflectance values increased throughout the growing season, and exceeded 50% when plant height and plant width exceeded 50 cm (corresponding to the third-grown square growth stage). Application of a percent threshold of likely contribution of cotton plants to georeferenced spectral reflectance images will yield likelihood maps of estimated distribution of cotton plants. Timely identification of volunteer and regrowth cotton plants will aid eradication program personnel in locating and destroying these hostable plants.