Abstract
Previous studies have indicated that crown morphology and within-crown color variations, rather than between-crown color differences, are principal keys to the accurate aerial detection of Dutch elm disease (DED). Because within-crown color variations often tend to be very subtle on conventional infrared photographs, microdensitometry and subsequent digital enhancement were evaluated as potential means of improving detection accuracy.
Vertical 70 mm color infrared photographs at scales of 1:6,000 and 1:12,000 were taken on several dates during 1979-80 over a St. Paul, Minnesota study area containing high incidence of DED. Scanning microdensitometer data from the photographs were processed using three enhancement techniques: color contrast stretch, two-band spectral ratioing, and multiple discriminant analysis. The numerically processed data were subsequently reconverted to image form using a high resolution color film recorder. Experienced interpreters analyzed both the original imagery and the enhanced products.
The computer enhanced images dramatically increased the expression of subtle within-crown and between-crown color variations, but only nominally improved detection accuracy. Thus, the factor limiting detection from vertical photography appears to be the inconsistency, not the subtletly, in which DED is manifested as crown color variation.
Date of this Version
1981