Crop Science Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (20)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Purcell, L. C.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Purcell, L. C.
Agricola
Right arrow Articles by Purcell, L. C.
Crop Science 40:834-837 (2000)
© 2000 Crop Science Society of America

NOTES

Soybean canopy coverage and light interception measurements using digital imagery

Larry C. Purcell

Univ. of Arkansas, Dep. of Crops, Soils, and Environmental Sciences, 276 Altheimer Drive, Fayetteville, AR 72704 USA

lpurcell{at}comp.uark.edu


   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 Materials and Methods
 Results and Discussion
 REFERENCES
 
Canopy light interception (LI) is important for yield and crop growth, but LI is often not measured because measurements must be made close to solar noon in unobstructed, direct-beam sunlight. Digital imagery may allow measurements of canopy coverage that are independent of solar radiation and solar angle restrictions. The objectives of this research were to determine the proportion of ground area covered by a soybean (Glycine max L. [Merr.]) canopy from digital images taken from above, and to compare the canopy-coverage measurements with LI measurements. Software limited the scanned areas of digital images to leaves, which allowed calculation of fractional canopy coverage. Similar values of canopy coverage were obtained throughout the day using digital imagery. Furthermore, comparisons of canopy coverage values with LI measured near solar noon indicated that there was a one-to-one relationship. Digital imagery, coupled with appropriate software, offers a simple and effective method of determining canopy coverage and LI.

Abbreviations: JPEG, joint photographic experts group • LAI, leaf area index • LI, light interception • PAR, photosynthetically active radiation


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
Materials and Methods
 Results and Discussion
 REFERENCES
 
THE INTERCEPTION OF LIGHT by a canopy is a fundamental requirement for crop growth. A mechanistic understanding of light interception and the relationship to crop growth have been important concepts applicable to virtually all crops (Monteith, 1977). Despite the recognized importance of light interception, measurements of this important parameter are often neglected because of the difficulty of obtaining accurate estimates.

One problem is that measurements should be made when sunlight is unobstructed (e.g., Board et al., 1992; Egli, 1994; Flenet et al., 1996). Additionally, light interception generally refers to measurements made close to solar noon (e.g., Board et al., 1992; Egli, 1994) when the sun is near its highest point above the horizon. Measurements made at other times of day are meaningful if the solar angle is also known. To make measurements within an hour of solar noon leaves only 2 h per day for measurements, and these conditions must be in conditions of unobstructed sunlight.

The fraction of total solar radiation intercepted by a canopy (LI) was described as an analog of Beer's Law by Monsi and Saeki (1953):

(1)

In this equation, only two variables determine LI: the extinction coefficient, k, and the leaf area index (LAI). The extinction coefficient describes the angle of leaves to the sun and varies between 1 (completely perpendicular to the sun) and 0 (completely vertical to the sun). As defined, the angle between the sun and leaves depends upon the angle of leaves to the horizon and the angle of the sun to the horizon. The angle of the sun to the horizon ({alpha}), of course, changes over the course of the day and with season of the year and latitude. To correct for the angle of the sun, k may be divided by the sin ({propto}), or measurements may be made near solar noon when the sin ({propto}) is approximately 1.

The most common method of determining LI is to measure photosynthetically active radiation (PAR) above a canopy and beneath a canopy near solar noon when the light is unobstructed by cloud cover (Board et al., 1992; Egli, 1994; Flenet et al., 1996):

(2)

Line quantum sensors are available commercially that integrate PAR along a 1-m length. The sensor may be placed perpendicular to the row (Egli, 1994). If the sensor cannot be placed evenly from the middle of one row to another, then a portion of the sensor may be covered with a material that blocks light. Alternatively, the sensor may be placed parallel to the row beneath the canopy, and multiple measurements may be made between rows and averaged (Board et al., 1992).

Digital images taken from above a crop offer the possibility of estimating the potential of a canopy to intercept light provided that (i) the soil background can be distinguished from leaves, (ii) light transmission of leaves is small relative to light absorption, and (iii) that the angle of the camera to the horizon approximates the solar angle. If these assumptions are met, then the fraction of ground area covered by leaves (canopy coverage) should be similar to LI measurements made in unobstructed light, as described by Eq. [1].

One advantage of a digital imagery system for determining canopy coverage is that the camera angle is constant so that measurements can be made at any time of day, regardless of cloud cover. Digital cameras have also become advanced and affordable, image quality for most cameras is excellent, and there are powerful software applications that allow analysis of images based upon intensity and/or specific spectral bands. For example, the public-domain software developed by Ewing and Horton (1999) provides a quantitative color-image analysis system that they used to measure canopy coverage in maize (Zea mays L.).

The first objective of this research was to develop a method of determining the proportion of ground area covered by green leaves using digital photographs and commercially available software. A second objective was to evaluate the relationship of fractional canopy coverage, as determined by digital imagery, with LI, as measured with a 1-m length light sensor.


   Materials and Methods
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
Results and Discussion
 REFERENCES
 
Several soybean cultivars, ranging from maturity group 00 through IV, were sown as part of a plant population experiment at Fayetteville, AR (36° 5' N latitude) on 22 May 1999 and 15 July 1999. The soil was a Captina silt loam (fine-silty, mesic Typic Fragiudult). A plot consisted of seven rows, spaced 0.19 m apart, 7 m in length. After complete emergence, seedling counts were made, and plant populations ranged from 20 to 100 plants m-2.

On 29 June 1999 and 10 Sept. 1999 for the early- and late-sowing dates, respectively, light interception measurements were made and digital images were recorded for a wide range of plant populations. Plants were in vegetative stages of development on the June date and early stages of pod formation on the September date. Light interception measurements were made in full sunlight within an hour of solar noon with a 1 m line-quantum sensor (Model CI-150, CID, Inc., Vancouver, WA). PAR was measured above the plant canopy, and three PAR measurements were made beneath the canopy perpendicular to the row direction. Row direction was north–south and east–west for measurements made on 29 June and 10 September, respectively. The PAR beneath the canopy were averaged, and LI was calculated as described by Eq. [2]. Additionally, light interception measurements were made and digital images were recorded on 15 and 16 September at 0730, 0930, 1200, 1400, and 1630 h, for three plots with large visual differences in canopy coverage. Row direction for these plots was north–south. The purpose of these measurements was to evaluate the stability of canopy-coverage estimates during the course of a day compared with the stability of light interception.

Digital images were made from the center of each plot using an Olympus D-500L digital camera (Olympus America, Inc., Melville, NY). The camera was mounted 1.5 m above the canopy and inclined 70° from the horizon. This inclination prevented the camera mount from being included in the image and resulted in a trapezoid-shaped measurement area. The field of view was 0.79 m wide in the foreground and 1.12 m in the background with a total area of 1.62 m2. Image size was 640 by 480 pixels, and the image was stored in JPEG (joint photographic experts group) file format, which required 74 kb of memory per image.

After digital photographs were transferred to a computer, they were analyzed individually by SigmaScan Pro (v. 4.0, SPSS, Inc., Chicago, IL). The software has selectable options to define the hue and saturation values to be included in the analysis. Software settings include full-scale ranges for hue from 0 to 255 and for saturation from 0 to 100. Hue settings from 25 to 130 and saturation values from 10 to 75 selectively included green leaves in the scanned images. In some cases, hue and saturation settings were adjusted ±5% of the standard values if the scanned area was obviously not including leaves. For the diurnal measurements, in particular, some adjustments in hue and saturation values were required to separate shadows from leaves. Within a measurement time, however, no additional adjustments were required.

After scanning the green area of each image, the total pixel count in the field of view was determined. The total pixel count per image was 304 279, which reflects an approximate 1% loss of the total pixel count during image processing. The fractional canopy coverage was defined as the number of scanned pixels divided by the total number of pixels per frame.


   Results and Discussion
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
REFERENCES
 
Several examples illustrating the ability of the software to distinguish leaves from the soil background are shown in Fig. 1 . Images in the right column of Fig. 1 are the unaltered digital photographs taken from above the soybean canopy. The green portion of the images in the left cloumn of Fig. 1 shows the portion that was recognized by the software on the basis of user-defined hue and saturation values. From very sparse plant populations to near full canopy coverage, the scanning parameters successfully separated leaves from soil and from shadows. By knowing the sum of the areas highlighted in the left column of Fig. 1 and the total area per frame, the fractional canopy coverage could be calculated. The images with canopy coverage values of 0.13 and 0.49 were made immediately after a rain under cloudy conditions. The images with canopy coverage values of 0.65 and 0.93 were made when the soil surface was dry in unobstructed sunlight. Although soil color changed with soil moisture conditions this had no effect on the ability to distinguish leaf area from the soil. High light levels tended to increase frequency of false positives being detected by the image analysis, but these errors were small relative to the leaf area of the crop.



View larger version (91K):
[in this window]
[in a new window]
  		Fig. 1 Digital images of soybean canopies representing different plant populations in the field (right) and after (left) scanning. The area highlighted in green in the left column of the figure was divided by the total area of the field of view, and this value was used to calculate the fractional canopy coverage values that are shown

 
Light interception changed during the course of the day (Fig. 2A) . When the sun was near the horizon, LI by a sparse canopy approached 1.00. As the sun moved overhead, near solar noon, LI reached a minimum, and then increased in the afternoon as the sun moved toward the horizon. In contrast to the LI measurements, canopy-coverage measurements using digital imagery were independent of the time of day that measurements were made (Fig. 2B). Similar canopy-coverage values were also obtained at night with a flash on the digital camera (data not shown). The constant values of canopy coverage would be expected in that the camera inclination was held constant, whereas the solar angle changed throughout the day.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 2 Fractional light interception (A) and fractional canopy-coverage (B) versus time of day. Light interception measurements were made with a line quantum sensor and canopy-coverage measurements were made using digital imagery. Measurements were made on three plots on two consecutive days. Values were averaged over days for each time. Error bars are ± standard deviation of the mean

 
Light interception, measured near solar noon with a line-quantum sensor, was compared with canopy coverage estimates that were made at the same time with digital imagery (Fig. 3) . Canopy coverage values from near 0.0 to 1.00 fell very close to the one-to-one line for LI values, and this relationship was true across several cultivars (data not shown). From this analysis on soybean during vegetative development, canopy-coverage measurements accurately predicted LI.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 3 Fractional canopy coverage versus fractional light interception. Light interception measurements were made with a line quantum sensor within an hour of solar noon in full sunlight, and canopy-coverage measurements were made at the same time using digital imagery. The line represents the one-to-one relationship between canopy coverage and light interception

 
Although data for Fig. 3 were taken within an hour of solar noon, at a latitude of approximately 36°, the maximum solar elevation angle on 29 June and 10 September was 77 and 59°, respectively (Muneer, 1997). The close agreement between LI and canopy coverage (Fig. 3) may be due to the camera inclination approximating that of the solar elevation angle at midday. The difference in inclination between the maximum solar angle and the camera angle was -7° for 29 June and -11° for 10 September. Assuming an extinction coefficient of 0.55 for soybean grown in 0.19-m rows (Flenet et al., 1996), canopy coverage using digital imagery at a LAI of 3 would over estimate LI by 0.01 for 29 June and by 0.04 for 10 September (Eq. [1]). Closer agreement between LI and canopy coverage could be achieved by adjusting camera inclination to the maximum solar elevation angle for a given latitude and day of year. Nevertheless, for crops with LAI > 1.0, error in prediction of LI from canopy-coverage measurements should be small (<0.08) provided that the difference in maximum solar elevation angle and the camera inclination is less than 25° (Eq. [1]).

Recording digital images required approximately 30 s per plot in the field. In the laboratory, determining the percentage green area from each image required approximately 1 min. This amount of time is comparable to that required for taking LI measurements using a line quantum sensor.

The ease with which canopy-coverage measurements were made makes this a desirable and powerful means of determining potential LI by a canopy near solar noon. Canopy-coverage measurements using digital imagery overcame the major limitations of using a line-quantum sensor for LI. Canopy-coverage measurements could be made at any time during daylight hours, in the absence of direct beam radiation (i.e., under overcast conditions), and these values were similar to LI measurements made near solar noon in full sunlight. An additional advantage of this technique is the multiple uses that a digital camera and scanning software offer a laboratory (Ewing and Horton, 1999) compared with the very specialized function and limited use of a line-quantum sensor.

By combining canopy-coverage measurements with other spectral qualities of the canopy, it may be possible to separate green leaves from senescing leaves. Adamsen et al. (1999) used ratios of green to red color bands in digital images of wheat (Triticum aestivum L.) and found that the green to red ratio correlated well with other measures of leaf senescence. The digital imagery technique for measuring canopy coverage, as described in this manuscript, could be adapted easily for similar applications, such as measuring the rate of turf establishment and the regrowth of weeds following herbicide or tillage treatments.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 REFERENCES
 
This paper is published with the approval of the director of the Arkansas Agric. Exp. Stn. (manuscript number 99-107).

Received for publication September 27, 1999.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Materials and Methods
 Results and Discussion
 REFERENCES
 




This article has been cited by other articles:


Home page
 Agron. J.Home page
G. Stevens, A. Wrather, M. Rhine, E. Vories, and D. Dunn
Predicting Rice Yield Response to Midseason Nitrogen with Plant Area Measurements
Agron. J., February 26, 2008; 100(2): 387 - 392.
[Abstract] [Full Text] [PDF]

Home page
 ANN BOT (LOND)Home page
P. Dutilleul, L. Han, and D. L. Smith
Plant Light Interception Can Be Explained via Computed Tomography Scanning: Demonstration with Pyramidal Cedar (Thuja occidentalis, Fastigiata)
Ann. Bot., January 1, 2008; 101(1): 19 - 23.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
G. Stevens, E. Vories, M. Mulesky, M. Rhine, and D. Dunn
Irrigation to Maximize Vaccine Antigen Production in Genetically Modified Tobacco
Agron. J., September 10, 2007; 99(5): 1271 - 1277.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
L. J. Krutz, C. H. Koger III, M. A. Locke, and R. W. Steinriede Jr.
Reduced Surface Runoff Losses of Metolachlor in Narrow-Row Compared to Wide-Row Soybean
J. Environ. Qual., July 17, 2007; 36(5): 1331 - 1337.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
B. V. Ottis and R. E. Talbert
Rice Yield Components as Affected by Cultivar and Seeding Rate
Agron. J., November 17, 2005; 97(6): 1622 - 1625.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
J. T. Edwards, L. C. Purcell, and D. E. Karcher
Soybean Yield and Biomass Responses to Increasing Plant Population among Diverse Maturity Groups: II. Light Interception and Utilization
Crop Sci., August 1, 2005; 45(5): 1778 - 1785.
[Abstract] [Full Text] [PDF]

Home page
 Agron. J.Home page
J. T. Edwards, L. C. Purcell, and E. D. Vories
Light Interception and Yield Potential of Short-Season Maize (Zea mays L.) Hybrids in the Midsouth
Agron. J., January 1, 2005; 97(1): 225 - 234.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
D. E. Karcher and M. D. Richardson
Quantifying Turfgrass Color Using Digital Image Analysis
Crop Sci., May 1, 2003; 43(3): 943 - 951.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
L. C. Purcell, R. A. Ball, J. D. Reaper III, and E. D. Vories
Radiation Use Efficiency and Biomass Production in Soybean at Different Plant Population Densities
Crop Sci., January 1, 2002; 42(1): 172 - 177.
[Abstract] [Full Text] [PDF]

Home page
 Crop Sci.Home page
M. D. Richardson, D. E. Karcher, and L. C. Purcell
Quantifying Turfgrass Cover Using Digital Image Analysis
Crop Sci., November 1, 2001; 41(6): 1884 - 1888.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (20)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Purcell, L. C.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Purcell, L. C.
Agricola
Right arrow Articles by Purcell, L. C.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Vadose Zone Journal
Journal of Natural Resources
and Life Sciences Education		 Soil Science Society of America Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome