924
mays L.), sorghum (Sorghum bicolor L.), soybean (Glycine max Merr.), and wheat (Triticum aesth'um L.) plus 40
agricultural soils. As crop residues decomposed, their fluorescence values approached the fluorescence of the soils.
Moisture quenched fluorescence, but the relative difference in fluorescence between crop residues and soil remained
fairly constant. Greater than 90% of the crop residues less than 2 years old could be discriminated from 35 of the
40 dry soils and 39 of 40 wet soils using fluorescence. Daughtry et. al. (1993) concluded that fluorescence
techniques were better suited for discriminating soils and residues than reflectance techniques.
Our previous research involved non-imaging techniques to measure fluorescence. Advances in low light
imaging technology convinced us that it is feasible be possible to capture an image of the fluorescence from crop
residues. Video imaging of crop residue fluorescence could provide an intuitive understanding as to the amount of
residue cover which was not possible using traditional non-imaging techniques. Our objective was to demonstrate
the feasibility of fluorescence imaging to quantify crop residue cover.
2 - MATERIALS AND METHODS
Video images were acquired with a Xybion Intensified Multispectral Camera 1 (Model 201, Xybion Electronics Corp..
Cedar Knoll, New Jersey, USA). This camera has a filter wheel, located in front of the intensified camera assembly
that contains six 1-inch (25 mm) optical filters. The filter wheel can run synchronously with the video scan rate to
provide a repeating 6-image sequence or the wheel can be locked at any one filter position to provide continuous
imaging in a selected spectral region. For the fluorescence images, we operated the camera in the locked filter mode
and integrated over 6-120 video fields (0.1 -2.0 sec). For the reflectance images, integration was not necessary' and
we operated either in the run mode or the locked mode depending on our objectives. Video images were grabbed
using the Xybion Image Capture and Analysis System (X1CAS) on a PC microcomputer.
The filter wheel contained the following band pass interference filters: 453-488 nm, 535-570 nm. 650-685
nm, 735-750 nm. 775-795 nm, and 840-870 nm. Although the 453-488 nm band was not optimal for measuring the
broad band fluorescence of crop residues (Daughtry et al. 1993), it was satisfactory for this proof-of-concept
demonstration.
The video camera was mounted on a frame 0.7 m
above the surface (Fig 1). The ultraviolet illumination
source was four 12-volt, 6-Watt, longwave UV lamps
(Model ML49, UVP, Inc., San Gabriel, California, USA)
arranged in a square pattern 0.3 m above the soil surface.
The lamps were angled toward the center to provide nearly
uniform illumination over the scene. Radiation from the
lamps was filtered with Schott UG-1 glass to minim ize
radiation greater than 400 nm that would interfere with
measurements of fluorescence. Visible and near infrared
illumination was provided by two 120-volt, 60-Watt
incandescent lamps positioned to provide nearly uniform
illumination over the scene.
To calibrate the video data, three aluminum disks
of known area that were painted white were placed on a
gray background. The number of pixels occupied by the
disks were determined by e xaminin g a histogram of the
video scene. This distribution was distincdy bimodal, i.e.,
bright disks on a dark background. The area per pixel was
calculated and later used to determine the fraction of the
scene covered by residue.
A series of test scenes was created using two
distincdy different agricultural soils and weathered soybean
stems collected 8 months after harvest. The soybean stems
were mottled with dark-colored areas caused by microbial
colonization and were selected as a representative of crop
residues that fluoresce moderately (Daughtry et al., 1993).
The projected area of the soybean stems was measured with
an area meter (LI-3100, LICOR, Inc., Lincoln, Nebraska,
Figure 1 Xybion intensified multispectral camera
mounted over a tray of soil and soybean residue.
Four UV lamps with filters illuminated the scene.
1 Company and trade names are given for the benefit of the reader and do not imply any endorsement of
the product or company.