Comparison of airborne multispectral and hyperspectral imagery for mapping cotton root rot

Abstract

Cotton root rot, caused by the soilborne fungus Phymatotrichum omnivorum, is a major cotton disease in the south western and south central United States. Accurate delineation of root rot infestations is necessary for site-specific management of the disease. The objective of this study was to compare airborne multispectral and hyperspectral imagery for detecting and mapping root rot areas in cotton fields. Two centre-pivot irrigated fields in south Texas were selected for this study. Airborne 3-band multispectral imagery and 128-band hyperspectral imagery were taken from the two fields shortly before harvest when infested areas were fully expressed for the season. Both types of imagery and the principal component images derived from the hyperspectral imagery were classified into 2-10 spectral classes using unsupervised classification techniques. The individual spectral classes were then grouped into infested and non-infested zones. Accuracy assessment on the two-zone classification maps showed that both types of imagery as well as the principal component imagery equally accurately identified root rot areas within the fields. These results indicate that both airborne multispectral and hyperspectral imagery can be successfully used for assessing root rot infestations within cotton fields. The classification maps and buffer maps as well as the image processing procedures presented in this study can be readily used by cotton growers and consultants for the site-specific cultural and chemical management of this disease.