Tyson B. Raper, Derrick M. Oosterhuis, Upton Siddons, Larry C. Purcell, and Morteza Mozaffari
University of Arkansas, Fayetteville, AR 72701
OBJECTIVE
The objective of this research was to examine the
effectiveness of DGCI derived from standard digital
photographs and commercially available image-analysis
software to determine cotton N status and to compare
sensitivities of calculated DGCI from laboratory, field
nadir, and field off-nadir photographs with changes in leaf
N and chlorophyll meter measurements.
METHODS (cont.)
Field nadir and field off-nadir (approximately 60 from
nadir) pictures were taken of the canopy with an
inexpensive digital camera (Canon PowerShot SD450,
Lake Success, NY) against a neutral color board that
included yellow and green disks which served as interval
color standards (Fig. 2). Chlorophyll meter (Minolta
SPAD-502, Konica Minolta Sensing, Inc., Tokyo, Japan)
measurements and pictures of two most recently
matured, fully expanded leaves 4-6 nodes from the
terminal were taken under fluorescent lighting against a
standardized color board (Fig. 2).
INTRODUCTION
Inadequate or excessive applications of fertilizer N in
cotton are financially and environmentally costly. Timely
in-season N determination in cotton (Gossypium hirsutum
L.) can help producers combat these negative effects;
however, current methods of determination are often time
consuming and/or expensive. More instantaneous, accurate
methods of determining N status which utilize equipment
already in the possession of the producer are of particular
interest. Recent work utilizing an inexpensive digital
camera and image processing software to calculate the dark
green color index (DGCI) has resulted in successful
determination of corn and turf N status (Rorie et al., 2011;
Karcher et al., 2003). This new method has the potential
to provide producers with accurate, precise measurements
of cotton N status in a quick enough manner to influence
yield- impacting management decisions.
RESULTS
Response of leaf N to applied fertilizer N was moderate
(r22=0.55, visual differences evident in Fig. 2).
Field nadir and field off-nadir DGCIs did not correlate
as strongly to leaf N as laboratory DGCIs
(Fig. 3).
Laboratory DGCIs and SPAD measurements responded
similarly to changes in leaf N (r22=0.603, r22=0.561,
respectively).
Laboratory DGCIs were very strongly correlated to
SPAD readings (r22=0.914, Fig. 3).
Field nadir and field off-nadir DGCIs were moderate and
strongly correlated to SPAD readings (r 22=0.680, r22=0.818,
Fig. 3).
1.0
SPECTRAL REFLECTANCE RESPONSE TO N
0.6
N DEFICIENT
N SUFFICIENT
0.8
r =0.603
n=28
0.7
r =0.914
n=28
0.6
0.5
0.4
0.9
0.2
0.0
500
600
700
800
WAVELENGTH, nm
NADIR DGCI
REFLECTANCE, %
0.8
LAB DGCI
0.9
0.8
0.7
r =0.440
n=28
0.6
r =0.680
n=28
0.5
ACKNOWLEDGEMENTS
Special thanks to Upton Siddons for his assistance in data
analysis and to Dr. Morteza Mozaffari for allowing data
collection on his field.
OFF NADIR DGCI
0.4
METHODS
A field trial was conducted in 2011 at the Lon Mann Cotton
Research Station near Marianna, AR. Fertilizer N rates
included 0, 30, 60, 90, 120, and 150 lb N/acre applied as
urea in a single, pre-plant application to create a wide range
of plant N status. Treatments were replicated four times.
Leaf sampling, chlorophyll meter readings and digital
pictures were taken at the third week of flowering. Leaf
samples were dried and ground to pass a 20 mesh sieve and
leaf N concentration of the ground sample was determined
by the Agricultural Diagnostic Laboratory at the University
of Arkansas in Fayetteville, AR.
CONCLUSIONS
Digital image analysis is a practical and inexpensive
method sensitive to cotton N status which could replace
chlorophyll meters.
Laboratory DGCIs correlate most strongly to changes in
leaf N and SPAD readings, but field off-nadir DGCIs
appear to be the most practical option for producers.
An effective extension program could be set up to accept
emailed or picture messaged off-nadir images of the crop of
interest with a standardized color board for instantaneous
determination of cotton N status.
0.9
0.8
REFERENCES
r =0.480
n=28
0.7
r =0.818
n=28
0.6
0.5
0.4
1.5
2.0
2.5
3.0
LEAF N, %
3.5
4.0
35
40
45
SPAD READING
50
Bell, P. F., D. J. Boquet, E. Millhollon, S. Moore, W. Ebelhar, C. C.
Mitchell, J. Varco, E. R. Funderburg, C. Kennedy, G. A.
Breitenbeck, C. Craig, M. Holman, W. Baker, and J. S.
McConnell. 2003. Relationships between leaf-blade nitrogen and
relative seedcotton yields. Crop Sci. 43:1367-1374.
Karcher, D. E., and M. D. Richardson. 2003. Quantifying turfgrass
color using digital image analysis. Crop Sci. 45: 943-951.
Rorie, R. L., L. C. Purcell, M. Mozaffari, D. E. Karcher, C. A. King,
M. C. Marsh, and D. E. Longer. 2011. Association of
Greenness in corn with yield and leaf nitrogen. Agron. J.
103:
529-535.
