Surveying Thermally-defoliated Cotton Plots with Color-infrared Aerial Photography

Search PMN

Reginald S. Fletcher, Research Soil Scientist, USDA-ARS-IFNRRU, 2413 East Highway 83, Weslaco, TX 78596; Allan T. Showler, Research Entomologist, USDA-ARS-IFNRRU, 2413 East Highway 83, Weslaco, TX 78596; and Paul A. Funk, Research Agricultural Engineer, USDA-ARS-SW Cotton Ginning Research Lab, P.O. Box 578, Mesilla Park, NM 88047

Abstract

Thermal defoliation is a nonchemical alternative for terminating cotton (Gossypium hirsutum L.) growth and preparing it for harvest, making the technique ideal for cotton grown in sustainable systems. For large cotton fields, growers need assistance in examining the effectiveness of thermal defoliation because green leaves remaining in the field reduces the price grade of the cotton and increase the time needed to harvest the fields. We qualitatively and quantitatively evaluated aerial color-infrared photographic transparencies of two study sites to determine the feasibility of using color-infrared aerial photography as a tool for surveying cotton plots subjected to thermal defoliation. Treated canopies appeared brown and in mixtures of brown and red tones on the color-infrared transparencies, leading to their separation from the control plant canopies (appearing in a red color). Quantitative analysis indicated that the green, red, and/or near-infrared light sensitive layers of the film significantly contributed to the color differences observed between the thermally-defoliated plots and the control plots. The results indicate that color-infrared aerial photography has high potential as a tool for assisting growers in surveying thermally-defoliated cotton fields. After visiting suspect areas identified on the photography, producers may opt to reapply the treatment.

Introduction

Thermal defoliation has shown promise as a nonchemical alternative for terminating cotton growth and defoliating cotton canopies (4,5,11), making it an ideal harvest aid for cotton grown under organic farming systems, which prohibits the use of synthetic harvest-aid chemicals (4). This procedure uses propane gas to generate heat that is applied directly to the plant canopies for rapidly killing the leaves, allowing growers to harvest cotton fields after twenty-four hours (5,11). Unlike conventional defoliation methods (10), thermal defoliation does not require optimal weather conditions to apply the treatment. Producers can also use it in conventional systems for preparing fields threatened by severe weather, increasing their ability to harvest fields before the arrival of the inclement weather (11). Thermal defoliation has also shown potential in controlling late-season sucking insects (5).

Cotton harvested from areas where leaf kill was incomplete may receive lower price grades because chlorophyll in green leaves stains the fibers during harvest, leading to a reduction in the price grade. The juices remaining in the green leaves also cause excessive gum build-up in the spindles of the pickers, resulting in stoppage of the harvest to clean the spindles and thus increasing the harvest time per field. In small fields, growers can easily determine the effectiveness of the treatment by driving around the fields or by conducting ground surveys to locate areas that may need further treatment; but in large fields, this surveillance is tedious and time consuming. Identifying areas not responding to the treatment can lead to producers re-treating these areas, which would maintain the price grade of the cotton throughout treated fields and reduce the time required to harvest the fields.

Photography acquired from an airborne platform may provide the information needed to survey thermally-defoliated fields because it provides a "bird’s eye" view of the area of interest. Over the years, scientist and natural resources managers have used color-infrared aerial photography as a survey tool to assess and monitor a variety of crop stresses (1,2,9,10,13). Healthy vegetation strongly reflects near-infrared light, absorbs red light, and moderately reflects green light (6), all of which are detected by color-infrared aerial film. On color-infrared positive transparencies and prints, healthy vegetation appears in red tones and stressed vegetation appears in various false colors including pink, magenta, yellow, white, blue, and green (3,6). The color of healthy or stressed vegetation is dependent on plant type, time of exposure to the stress, and stress type (1). Because the thermal treatment leads to quick leaf death with desiccated leaves remaining on the plant for an extended period of time (11), we hypothesized that color-infrared aerial photography would provide adequate information for evaluating cotton plots subjected to thermal defoliation. The objective of this study was to determine the feasibility of using color-infrared photography as a tool for surveying cotton plots subjected to thermal defoliation.

Evaluating A Cotton Plot Survey Tool

Data were colleted from two on-going experimental studies located near Weslaco, TX (26°09′N 97°57′W). Site one (»1 acre) was located at the United States Department of Agriculture-Agricultural Research Service (ARS)-Kika de la Garza Subtropical Agricultural Research Center; site two (» 2 acre) was located at the Texas A&M University South Research Farm. Cotton cultivar DP-5415-RR was planted in the experimental plots on 2 March 2004 on 40-inch rows. A randomized complete block design consisting of six blocks and three treatments was employed at both sites. The treatments were control, thermal defoliation, and chemical defoliation (Def 6, Bayer Cropscience, Kansas City, MO) (11). The objective of the current study was to compare the differences between the thermally-defoliated cotton and the control cotton; therefore, subsequent analyses focused on these two treatments. The thermal treatment was applied to the plots on 21 July 2004 using a two-row prototype thermal defoliator (5,11). The thermal treatment consisted of forcing air with an average temperature of 380°F through the cotton canopies. The tractor was traveling at 1 mph, resulting in a dwell time of 10 seconds.

On 27 July 2004, color-infrared aerial photographs were taken of the plots with a Fairchild Reconnaissance camera (E. Coyote Enterprises Incorporated, Mineral Wells, TX) having a 12-inch focal length lens. Cloudy skies hindered ARS personnel from acquiring the photography at an earlier date. The camera was loaded with Kodak 1443 series 9 × 9-inch color-infrared film (Eastman Kodak Company, Rochester, NY). A Kodak Wratten gelatin filter number 15 (Eastman Kodak Company, Rochester, NY) was placed in front of the camera lens to block blue light because the film layers were sensitive to blue light; not blocking this light would lead to contamination of the layers by blue light. The aperture setting used on the camera was f11 with an exposure time of 1/500th of a second. The camera was mounted in a hole located in the belly of a fixed-winged aircraft. The photographs were acquired at an altitude of 1500 ft above ground level. The color-infrared film was sent to a commercial laboratory for processing. We requested for the laboratory to provide us with positive transparencies (land cover types on the positive transparencies have the colors described earlier in the Introduction).

For qualitative and quantitative analyses, the transparencies of the study sites were digitized at 1000 dots per inch with an Epson desktop scanner Model 1600XL (Epson America Inc., Long Beach, CA) and the Adobe Photoshop software, version 6.0 (Adobe Systems Inc., San Jose, CA), resulting in a pixel resolution (smallest area observable on the ground) of 0.125 ft. The end product was an image (digitized color-infrared transparency) having similar colors to the color-infrared transparency. The digitized transparencies were qualitatively assessed to determine differences between the thermally-defoliated and the control cotton plots. To better understand the colors shown on the digitized transparency, we used the Adobe Photoshop software to separate the transparencies into their individual light sensitive layers: near-infrared, red, and green.

Within each light sensitive layer, the marquee function and the histogram option of the Adobe Photoshop software were used to extract numerical information from the two center rows in the control plots, the thermally-defoliated plots, and the chemically-treated plots. Data were extracted from the chemically treated plot because it was needed for the analysis of variance, a statistical analysis used to evaluate data of a randomized complete block design (12). At site one, a 7.3 × 25.0-ft (average for each plot based on 11,600 pixels per plot) area was extracted from the thermally-defoliated plots, the chemically-defoliated plots, and the control plots for further analysis; at site two a 23.8 × 7.0-ft (average for each plot based on 10,640 pixels per plot) area was extracted from the thermally-defoliated plots, the chemically-treated plots, and the control plots for further analysis. Occasionally, the beginning and or ending sections of the chemically-treated plots and the control plots located in blocks adjacent to thermally-treated plots were affected by the heat released from the equipment, leading to the different sizes of the extraction areas used for sites one and two. The area of interest within each plot included the cotton plants, the soil, the soil-dead leaves complex, and the shadows, meaning there was no separation of soil, shadow, and foliage pixels.

For sites one and two, the digital data extracted from the treated plots (thermal and chemical) and the control plots for each light sensitive layer were transferred to the Statistix8 software version 8 (Analytical Software, Tallahassee, FL). The data were subjected to the analysis of variance. If the f-values indicated significant differences at α = 0.05 for the treatments, then the data were subjected to the Dunnett’s (α = 0.05) test. As indicated earlier, the primary objective of this study was to compare the thermally-defoliated treatment to the control treatment; the Dunnett’s test is the appropriate statistical analysis to use for this comparison (12).

Feasibility of Color-Infrared Photography for Surveying Thermal Defoliation

On the color-infrared transparencies (Figs. 1 and 2), cotton plants subjected to thermal defoliation appeared in brown tones or in mixtures of brown to magenta tones, the control plants appeared in red tones, the soil background appeared in medium gray to light gray tones, the soil-dead leaves complex appeared in brownish-gray tones, and the shadows appeared black. The green, red, and near-infrared light sensitive layers of the film contributed to the colors observed on the transparencies (Table 1).


	Fig. 1. (A) Scanned color-infrared transparency and its (B) near-infrared, (C) red, and (D) green components showing thermally-defoliated (arrow 1) and control (arrow 2) cotton plots at site one. Photography was acquired on 27 July 2004 (6 days after treatment).		Fig. 2. (A) Scanned color-infrared transparency and its (B) near-infrared, (C) red, and (D) green components showing thermally-defoliated (arrow 1) and control (arrow 2) cotton plots at site two. Photography was acquired on 27 July 2004 (6 days after treatment).

Table 1. Average digital count values ± standard errors (n=6) for the control and the thermally-defoliated cotton plots. Data were extracted from the green, red, and near-infrared sensitive layers of the color-infrared transparencies.

Site	Treatment	Film Sensitive Layers
Site	Treatment	Green	Red	Near-infrared
One	Control	19.6 ± 2.1	14.3 ± 2.2	47.5 ± 1.5
One	Thermal	43.2 ± 2.2	51.7 ± 2.5	67.4 ± 1.5
Two	Control	—^y	4.6 ± 0.2	36.9 ± 0.9
Two	Thermal	—	12.2 ± 1.1	33.5 ± 1.6

^x Values within the same column for a particular location were significantly different at α = 0.05 (Dunnett’s test).

^y Green film sensitive layer for site two was not subjected to the Dunnett’s test because the analysis of variance indicated that the differences among treatments were not significant.

At site one, green, red, and near-infrared data extracted from the thermally-defoliated plots were significantly higher (α = 0.05) than green, red, and near-infrared data obtained from the control plots (Table 1). At site two, the near-infrared data obtained from the thermally-defoliated plots were significantly lower than the near-infrared data extracted from the control plots, and the red brightness values of the thermally-defoliated plots were significantly higher than the red brightness values of the control plots (Table 1).

The heat applied to the cotton canopies destroyed the chlorophyll in the plant leaves, resulting in leaf death, and it affected the leaf architecture by causing some of the leaves to curl. Chlorophyll strongly absorbs red light and moderately reflects green light (7). When leaves loose chlorophyll, increases in their green and red light reflectance occurs (Table 1 and Figs. 1C, 1D, and 2C) (7). Near-infrared reflectance of plant canopies are often altered by changes in leaf architecture. However, in this study, it is believed that the architecture of the leaves were not solely responsible for the differences observed between the thermally-treated canopies and the control canopies on this light sensitive layer (Table 1 and Figs. 1B and 2B). Its peak sensitivity occurs in the near-infrared region of the light spectrum; nevertheless, this film layer also has some sensitivity to green and red light (blue light blocked by filter). It is believed that this sensitivity to visible light also contributed to the differences observed between the thermally-defoliated and the control cotton plants.

The soil background appeared in brighter gray tones than the vegetation in the thermally-defoliated plots and the control plots for all of the light sensitive layers for both sites (Fig. 1 B-D and Fig. 2 B-D). It was also observed in the thermally-defoliated plots that approximately twenty-five percent of the dead leaves fell to the ground, forming a soil-dead leaves complex (11). This complex also appeared brighter than the dead and/or green foliage remaining on the thermally-defoliated and control plants for the three light sensitive layers.

Shadows appearing black at sites one and two occurred at the end of the rows, in open areas between plants, within the canopies, and/or on the furrows (Fig. 1 B-D and Fig. 2 B-D).

At both sites, the significantly higher red data values observed for the thermally-defoliated plots were attributed to the combined effects of the plant canopy, the soil background, and the soil-dead leaves complex. The considerably higher near-infrared brightness values recorded for the thermally-defoliated plots at site one were primarily attributed to the plant canopy, the soil background, and the soil-dead leaves complex. At site two, the significantly lower near-infrared brightness values observed for the thermally-defoliated plots were primarily attributed to shadows covering the soil background and soil-dead leaves complex, resulting in a reduction of the near-infrared light reflected by these components. This decrease caused a decline in the overall near-infrared digital count values for the thermally-defoliated plots. The shadow effect at site two was increased because of row orientation. At this site, the rows were planted from north to south leading to the casting of shadows between the rows.

On the color-infrared transparencies, it is quite evident that cotton canopies subjected to thermal defoliation (appearing brown tones) can be separated from the control canopies (red tones) and that plants partially affected by the treatment can also be detected. Plants partially responding to the treatment appeared in mixtures of brown and red tones on the color-infrared transparency. At the time of image acquisition, leaf kill ranged from 80% to 98% within the thermally-treated plots (11). Reference 11 suggested that insufficient exposure to the hot air could have attributed to the thermal treatment not completely killing all of the leaves on the cotton plants. To increase leaf kill, the operator of the equipment can increase the temperature or can reduce the speed of the thermal defoliator, allowing the plant canopies to be subjected to heat for a longer period of time. Our results concurred with others indicating that color-infrared photography can be used as tool to detect differences between stressed and green vegetation and that it can be employed as a tool to survey agricultural fields (2,10,13).

Recommendations and Conclusions

To put this study into perspective, several points have to be clarified for potential users of this technology. The quantitative analysis of the film layers was used for scientific and statistical purposes only. In real word situations, qualitative assessment of the transparency should suffice for field surveys, reducing the burden of extracting data from the photographic print or transparency. For comparisons, users can achieve the best results when a control strip is included within the field. Also, evaluators of the photography must know that shadow, soil background, and the soil-dead leaves complex coupled with image resolution can have an effect on the appearance of the control plots and plots subjected to thermal defoliation. The understanding of these effects will lead to better interpretation of the photography. In addition, weather conditions can hinder the acquisition of the aerial photography, resulting in a delay of field surveillance.

The overall findings of this study indicate that color-infrared aerial photography has high potential as a tool for surveying thermally-defoliated cotton fields. Growers, extension agents, and consultants can use the photography to assist them in identifying areas that may need further treatment. After visiting the suspect areas, growers may opt to reapply the treatment.

Acknowledgments

The authors thank Veronica Abrigo, Fermin Alvarado, Raúl Cantú, Isabel Cavazos, Jaime Cavazos, Andy Cruz, Paul DelGado, Rene Davis, Armando Eldape, Jim Forward, Martín Galvan, Fred Gomez, Jaime Luna, Jaime Melendrez, and Kirk Zivkovich for their assistance.

Literature Cited

1. Avery, T. E., and Berlin, G. L. 1992. Fundamentals of Remote Sensing and Airphoto Interpretation, 5th Ed. Macmillan Publishing Company, New York.

2. Blazquez, C. H., and Horn, F. W., Jr. 1980. Aerial color infrared photography: applications in citruculture. NASA Reference Publication 1067.

3. Eastman Kodak Company. 2001. Kodak aerochrome III infrared film 1443. Kodak aerochrome III infrared NP film SO-734. Kodak Publication AS-77.

4. Funk, P. A., Armijo, C. B., McAlister, D. D., and Lewis, B. E. 2004. Experimental thermal defoliator trials. J. Cotton Sci. 8:230-242.

5. Funk, P. A., Armijo, C. B., Brashears, A. D., Hutmacher, R. B., Godfrey, L. D., McGuire, M. R., Bancroft, J. S., Showler, A. T., and Fletcher, R. S. 2005. Thermal Defoliation in 2004. 2005 Proc., Beltwide Cotton Conf., New Orleans, Louisiana. National Cotton Council, Memphis, TN.

6. Jensen, J. R. 2000. Remote Sensing of Environment: An Earth Resource Perspective. Prentice Hall, Upper Saddle River, NJ.

7. Knipling, E. B. 1970. Physical and physiological basis for the reflectance of visible and near- infrared radiation from vegetation. Remote Sens. Environ. 1:155-159.

8. Logan, J., and Gwathmey, C. O. 2002. Effects of weather on cotton responses to harvest-aid chemicals. J. Cotton Sci. 6:1-12.

9. Menges, R. M., Nixon, P. R., and Richardson, A. J. 1985. Light reflectance and remote sensing of weeds in agronomic and horticultural crops. Weed Sci. 53:569-581.

10. Philpotts, L. E., and Wallen, V. R. 1969. IR color for crop disease identification. Photogramm Eng. 35:1116-1125.

11. Showler, A. T., Funk, P. A., and Armijo, C. B. 2005. Effect of thermal defoliation on cotton leaf desiccation, senescence, post-harvest regrowth, and lint quality. J. Cotton. Sci. 9:229- 237.

12. Steel, R. G. D., and Torrie, J. H. 1960. Principles and Procedures of Statistics. McGraw-Hill Book Company, Inc., New York.

13. Toler, R. W., Smith, B. D., and Harlan, J. C. 1981. Use of aerial color infrared photography to evaluate crop disease. Plant Dis. 65:25-31.