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


     

Published online 18 May 2006
Published in Crop Sci 46:1564-1569 (2006)
© 2006 Crop Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
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 (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.
Agricola
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.

CROP ECOLOGY, MANAGEMENT & QUALITY

Implementation of Hyperspectral Radiometry in Irrigation Management of Creeping Bentgrass Putting Greens

K. C. Huttoa,*, R. L. Kingb, J. D. Byrd, Jr.a and D. R. Shawa

a Dep. of Plant and Soil Sciences, Mississippi State Univ., Mississippi State, MS 39762
b Dep. of Electrical and Computer Engineering, Mississippi State University, Mississippi State, MS 39762

* Corresponding author (khutto{at}ifas.ufl.edu)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Field research conducted in 2003 and 2004 evaluated hyperspectral radiometry as a tool to detect localized dry spots on creeping bentgrass (Agrostis stolonifera L.) putting greens. Discriminant analysis identified 14 wavebands between 861 and 887 nm that distinguished drought severity. An overall accuracy of 85% was achieved using these wavebands with low-stressed greens correctly classified 91%. High-stressed greens were correctly classified 100% of the time with individual wavebands between 905 and 992 nm using 2003 data as a training model to cross-validate 2004 data. These results suggest these individual wavebands are highly sensitive to early season drought-stress in creeping bentgrass greens and could be considered a more stable portion of the spectrum over time concerning high-stressed creeping bentgrass putting greens. A significant relationship (p = 0.034) was established between average reflectance values of wavebands between 1480 and 1530 nm and volumetric soil water content (VWC). Although not highly correlated, this relationship demonstrated that as VWC increased, reflectance decreased. Data collected in late spring/early summer may be the optimal time to detect areas of putting greens most susceptible to drought stress before unfavorable growth conditions occur.

Abbreviations: NIR, near infrared • PCC, partial correlation coefficient • SWIR, short wave infrared • VWC, volumetric soil water content


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
THE fine leaf texture and consistent ball roll characteristics of creeping bentgrass make this turfgrass the premier putting surface in golf. The optimum growth temperature range for this cool-season grass is 15 to 24°C for shoots and 10 to 18°C for roots (Beard, 1973, p.658). High summer air temperatures combined with high relative humidity in the southeastern United States do not favor creeping bentgrass growth (Dernoeden, 2000). However, the development of heat-tolerant cultivars has made this turfgrass more popular in the Southeast. Unfortunately, even with these cultivars heat and drought stress is still a major concern. Superintendents are constantly monitoring their creeping bentgrass greens for localized dry spots or "hot spots" formed by hydrophobic soil. These spots cause the turfgrass to wilt and potentially desiccate if not detected in a timely manner. There are several possible causes of localized dry spots, including excessive thatch buildup, compacted soil, poor irrigation coverage, steep sloping grade, or the development of hydrophobic soil (Tucker et al., 1990). Some common practices used by superintendents to manage localized dry spots are syringing (DiPaola, 1984), raising mowing height (Beard and Sifers, 1997), fans adjacent to greens oriented to blow across the turf surface (Taylor et al., 1993), and subsurface cooling and aeration (Camberato et al., 1999; Dodd et al., 1999). Applying wetting agents (Miyamoto, 1985) or the addition of small amounts of clay and silt combined with coring practices (Bond, 1978) could improve water infiltration and retention.

The use of a sand-based root zone in golf course putting greens has caused frequent occurrences of localized dry spots caused by hydrophobic soils (Wilkinson and Miller, 1978; Tucker et al., 1990). These soils tend to develop in the upper rhizosphere where root density is greatest (Wilkinson and Miller, 1978). Tucker et al. (1990) observed that hydrophobic conditions were localized to the upper 50 mm of soil. Hydrophobic soils develop because of an organic coating that surrounds soil particles in the artificial soils of putting greens (Wilkinson and Miller, 1978; Tucker et al., 1990). This organic coating mainly consists of humic (Roberts and Carson, 1972) and fulvic acids (Miller and Wilkinson, 1977).

High soil or air temperatures coupled with these localized dry spots increase the difficulty of creeping bentgrass greens management. Leaf injury under high soil temperatures has been associated with many factors that affect normal plant growth and development such as root growth inhibition (Xu and Huang, 2000a), hormone synthesis and transport (Udomprasert et al., 1995), water uptake (Graves et al., 1991; Huang et al., 1991), and nutrient uptake (Gur and Shulman, 1979; Huang and Xu, 2000). High temperatures may cause an imbalance between photosynthesis and respiration processes and carbohydrate depletion (Xu and Huang, 2000b). Additionally, low mowing heights coupled with high temperatures can add stress by removing leaf matter used for photosynthesis, while respiration continues (Huang et al., 1998). Environmental stresses can produce free radical molecules such as superoxide, hydrogen peroxide, hydroxyl radical, and singlet oxygen, which can damage lipids, proteins and nucleic acids in plant cells (Smirnoff, 1993; Foyer et al., 1994). Plants exposed to high temperature stress may lose the ability to minimize damage from these free radical species due to a dysfunctional active oxygen scavenging system (Price et al., 1989; Bowler et al., 1992; Zhang and Kirkham, 1994). Physiological stresses lead to increased reflectance in the visible region (400 to 700 nm) (Carter, 1993). Carter and Miller (1994) detected increases in reflectance in the visible region due to herbicide stress using a reflectance ratio of 694 nm/700 nm.

Currently, management practices during the summer months are targeted to keeping soil temperatures as close to the optimal growth temperature range as possible (Dernoeden, 2000). Lowering soil temperatures while being exposed to high air temperatures increased canopy net photosynthetic rate and carbohydrate content and reduced the carbon consumption to production ratio, which suggests that roots are more active in mediating carbohydrate responses than shoots when subjected to high temperatures (Xu and Huang, 2000b). Lowering soil temperature improved creeping bentgrass quality and shoot and root growth that had been exposed to high air temperature.

Remote sensing applications have been made in various crops in an effort to detect stress (Ripple, 1986; Hoque et al., 1992; Carter and Miller, 1994). Remote sensing methods that use the near infrared (NIR) and mid-infrared (MIR) regions of the spectrum may delineate stressed areas from non-stressed areas (Hunt and Rock, 1989). Rock et al. (1985, 1986) used a ratio of MIR to NIR, known as the moisture stress index (MSI) to detect stressed areas in coniferous forests. This damage was attributed to natural old age or stress caused by acid deposition or air pollution.

Assessing crop water stress in the field with remote sensing is based on two parameters: leaf temperatures measured with hand-held infrared thermometers, and the relationship between changes in reflectance patterns to crop variables that are responsive to the development of water stress (Mahey et al., 1991). The infrared red radiance ratio and normalized difference vegetative index (NDVI) were larger for irrigated crops than for non-irrigated crops. Plants under moisture stress have changes in reflectance in the visible, NIR and MIR, or shortwave infrared (SWIR), regions due to a decrease in chlorophyll (Carter, 1993) and leaf water content (Ripple, 1986). A decrease in chlorophyll causes an increase in red reflectance and a decrease in NIR reflectance. A decrease in leaf water content causes an increase in MIR reflectance. It is these areas in the spectral reflectance pattern that can be used to distinguish a stressed plant from a non-stressed plant.

Remote sensing has been shown to be an effective tool in detecting certain turfgrass stressors. Previous research found that wavebands in the visible and NIR regions could be used to distinguish drought stress (Fenstermaker-Shaulis et al., 1997), wear and traffic stress (Trenholm et al., 1999; Guertal and Shaw, 2004), herbicide injury (Bell et al., 2000), and disease severity (Nutter et al., 1993; Green and Burpee, 1997; Raikes and Burpee, 1998) from non-stressed turfgrass. However, no reports could be found by the authors where remote sensing technologies have been used to specifically detect localized dry spot on creeping bentgrass putting greens.

Creeping bentgrass putting greens require a high level of maintenance to achieve the high quality putting surface that it brings to the golf industry. Localized dry spots cause problems in irrigation management and sometimes are not detected before the plant begins to wilt. Remote sensing technologies could have the potential to detect these areas before wilting symptoms occur. Therefore, the objective of this study was to evaluate hyperspectral radiometry in detecting localized dry spots on creeping bentgrass putting greens in an effort to improve irrigation management practices.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Experimental Site Description
Reflectance data were collected in the summers of 2003 and 2004 from nine ‘Crenshaw’ creeping bentgrass putting greens at Old Waverly Golf Course in West Point, MS. Three stress categories were assigned to the greens, chosen based on their historical behavior. This information was provided by the golf course superintendent. Each stress category was assigned to three greens. These categories were high, intermediate, or low stress, where high stress is described as a green needing extra irrigation during the summer months and low stress is described as a green performing well under the normal irrigation practices. On each green, nine points were marked 5 to 6 m apart, depending on the size of the green, with a global positioning system (GPS) unit (Trimble AgGPS 114; Trimble, Sunnyvale, CA) for consistent sampling over time. Boundaries of all greens were created using a GPS unit. At each sampling point, reflectance, VWC, and canopy temperature data (2004 only) were collected. Canopy temperature data were collected in 2004 only because of newly acquired equipment after the 2003 data collection season. VWC is defined as the volume of water with a given volume of soil. It can also be expressed as the depth of water for a specific unit depth of soil (Brady and Weil, 1999). A hand-held hyperspectral spectroradiometer (FieldSpec Pro; Analytical Spectral Devices, Inc., Boulder, CO) was used to collect all reflectance data. The radiometer had a spectral range of 350 to 2500 nm with a 23° foreoptic. Reflectance readings were collected from shoulder height providing a 0.65 m field of view. Data were collected between the hours of 9:30 a.m. and 12:00 p.m. The radiometer was calibrated using a barium sulfate Lambertian surface every 15 to 20 min. A time-domain reflectometry soil moisture meter (TDR 300 FieldScout; Spectrum Technologies, Inc., Plainfield, IL) was used to collect VWC data at a soil depth of 10 cm. Canopy temperatures were measured using a thermal infrared thermometer (Raytek MX2 Infrared Thermometer; Raytek Corp., Santa Cruz, CA) held 15 cm above the canopy. All data were collected every 2 wk, depending on weather conditions and ongoing management practices at the golf course. Data were collected four times from May 28, 2003 to August 11, 2003 and three times from June 21, 2004 to October 1, 2004.

Creeping Bentgrass Green Management
The creeping bentgrass greens were constructed according to the specifications of the United States Golf Association (Ferguson, 1963). The greens were maintained at a mowing height of 3.5 mm. Greens were mowed 6 times per week throughout the spring, summer, and fall, usually in the morning. During the months of April to October the golf course implemented a preventative disease management plan, which consisted of fungicide applications every 7 to 10 d depending on environmental conditions. Fertilizer applications were made on a monthly basis from March to December. Nitrogen was applied in March (3 applications), May, June, October (2 applications), November, and December. Potassium was applied in March (2 applications), April (2 applications), June–September (2 application in September), and December. Phosphorus was applied in May and July–September. Gypsum was applied in March, June–October, and December. Greens were cultivated eight times during the year using needle tine aerification. All greens were topdressed in the spring and fall of the year, aerified 3 times per year, and verti-cut twice per year. Deep irrigation was applied every 4 d with the exception of late summer. During this time, deep irrigation was applied every 3 d. Green profiles were flushed with deep irrigation every 21 d by allowing each irrigation head to turn for 3 h.

Data Analysis
Data were collected in a completely randomized design with repeated measures replicated three times. The raw reflectance data from each point was used in all analyses. Reflectance data were analyzed using stepwise discriminant analysis to identify specific wavebands that could be used to differentiate between low, intermediate and high stress levels. Stepwise discriminant analysis uses a model to determine if a variable significantly contributes to discriminating between groups. If the variable fails to meet the criterion it is eliminated from the model and the process is continued until all variables that can be used to discriminate between groups have been identified. Once this is accomplished, the analysis stops (SAS, 2004). The spectral ranges of 350 to 1000 nm and 1480 to 1750 nm were analyzed. The 350 to 1000 nm spectral range was selected for extensive analyses for practical purposes, meaning it would be less expensive and more feasible for a turfgrass manager to use a device collecting reflectance data from this spectral range compared to data collected in the SWIR region. The SWIR region was selected to validate drought stress detected by wavebands in the 350 to 1000 nm range. Significant wavebands were used to obtain classification accuracies using discriminant analysis. Discriminant analysis provided classification accuracies for the reflectance data collected from each stress. Discriminant analysis forms a training set from the variables and uses these data to classify each value (SAS, 2004). For example, categories that define the stress levels for each putting green were assigned to reflectance data collected from the points on their respective greens. Discriminant analysis uses this information to form the training model, and each reflectance value is tested against the model and classified as either low, intermediate, or high stress based on how closely it fits the values from the training model. The final results provide the percentage of times reflectance values were correctly classified and misclassified. Analysis of variance was performed using the general linear means procedure (SAS, 2004) to determine those wavebands with stress by time interactions in an attempt to identify which wavebands could be used to consistently distinguish between stress levels. Broad band widths ranging from 10 to 50 nm were analyzed for consistency over time. Discriminant analysis was used to validate the 2003 reflectance data model with the 2004 reflectance data to measure to consistency of the model over time.

Canopy temperature and VWC data were tested against reflectance data to establish correlations using multivariate analysis of variance, which provided partial correlation coefficient values.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
Reflectance data were combined over years and analyzed using stepwise discriminant analysis to classify stress levels of greens. Numerous wavebands between 350 and 1000 nm were identified as significant by discriminant analysis. However, 33 individual wavebands between 719 and 799 nm provided the highest overall accuracy (99%) of all spectral regions with high-stressed greens being correctly classified 98% of the time (Table 1). Discriminant analysis identified 14 wavebands between 861 and 887 nm that had no stress by year interaction and were used to distinguish among treatments (listed in Table 1 footnote). An overall accuracy of 85% was achieved using these wavebands with the low-stressed greens being correctly classified 91% for both years (Table 1). Individual wavebands in the SWIR (141 wavebands between 1481 and 1750 nm in 2003; 38 wavebands between 1482 and 1737 nm in 2004) maintained overall classification accuracies at least 80% for both years. Overall accuracies decreased as individual dates were analyzed using these wavebands. However, as the season progressed the high-stressed greens were correctly classified between 70 and 76%. Broad band widths or vegetative indices were not successful in differentiating between stresses (data not shown).


View this table:
[in this window]
[in a new window]
  		Table 1. Classification accuracies of drought-stressed creeping bentgrass greens using NIR or SWIR wavebands.

 
Stepwise discriminant analysis identified 98 individual wavebands between 350 and 992 nm from the 2003 data to differentiate between stress treatments (listed in Table 2). These individual wavebands provided an overall accuracy of 99%, with the high-stressed greens being correctly classified 98% of the time (Table 3).


View this table:
[in this window]
[in a new window]
  		Table 2. Individual wavebands used to differentiate between creeping bentgrass stress categories.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Validation accuracies of creeping bentgrass stress categories from reflectance data collected in 2004 using 2003 reflectance data as a training model.

 
The high number of wavebands used in this analysis might not be practical from an application standpoint. Therefore, individual wavebands were divided into 5 groups according to their spectral regions. For example, wavebands in the visible region were analyzed in two groups (350 to 391 nm and 406 to 498 nm). Individual wavebands in the NIR region were divided into 3 groups (701 to 799 nm; 800 to 896 nm; 905 to 992 nm). The 2003 reflectance data from these groups were used to cross-validate the 2004 reflectance data. The 800 nm group provided an overall accuracy of 85% for reflectance data collected in 2003 with the high-stressed greens being correctly classified 84% of the time (Table 3). Validation of the 2004 data provided an overall accuracy of 42%, with the high-stressed greens being correctly classified 80% of the time (Table 3). This same waveband group correctly classified high-stressed greens 100% of the time early in the 2003 season. The low and intermediate stresses were correctly classified 89 and 78% of the time, respectively (Table 4).


View this table:
[in this window]
[in a new window]
  		Table 4. Validation accuracies of creeping bentgrass stress categories from reflectance data collected June 21, 2004 using May 28, 2003 reflectance data as a training model.

 
Higher classification accuracies were achieved when the 900 nm group was used to classify reflectance data early in the 2003 season. Reflectance data collected on May 28, 2003 provided an overall accuracy of 88% with high-stressed greens being correctly classified 85% of the time (Table 4). Validation of the May 28, 2003 model using individual wavebands provided an overall accuracy of 33%, with the high-stressed greens being correctly classified 89% of the time for June 21, 2004. Validation accuracies for the 800 nm group decreased with high-stressed greens being correctly classified 56% of the time (Table 4). The overall accuracy decreased from validation methods, but high-stressed greens were correctly classified 100% of the time. These results suggest these individual wavebands between 905 and 992 nm are highly sensitive to early season drought-stress in creeping bentgrass greens and could be considered a more stable portion of the spectrum over time concerning high-stressed creeping bentgrass putting greens. The remaining three spectral groups (300, 400, and 700 nm groups) as well as 10 nm band widths did not perform well in differentiating between stress levels (data not shown).

Correlations were analyzed for wavebands between 350 and 1000 nm, wavebands in the SWIR, and either canopy temperature or VWC. No significant relationships were found between canopy temperature and reflectance data from either portions of the spectrum. A significant relationship was established between the average reflectance values of wavebands between 1480 and 1530 nm and VWC (PCC = –0.144). This region of the spectrum is sensitive to changes in leaf water content, and could explain the correlation between VWC and SWIR values. Greater classification accuracies were achieved using NIR wavebands compared to SWIR wavebands. However, no significant correlations were achieved between reflectance values and VWC. One possibility is because the NIR region is measuring the effect of stress on the plant, while the SWIR region is more affected by the amount of water in the plant. Although not highly correlated, this relationship demonstrated that as VWC increased, reflectance decreased (Fig. 1). This relationship illustrates the effect of leaf water content on reflectance values in the SWIR. As the amount of water in the system increases, reflectance decreases.


Figure 1
View larger version (10K):
[in this window]
[in a new window]
  		Fig. 1. Relationship between average reflectance from 1480 to 1530 nm and volumetric soil water content in 2004. Abbreviations: PCC, partial correlation coefficient; VWC, volumetric soil water content.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
Changes in environmental conditions contributed to variability in VWC and reflectance data among collection dates; therefore specific locations on individual golf greens could not be identified as localized dry spots. However, the results of this study provided some promising conclusions. Classification accuracies of high-stressed greens began to decrease after the initial collection dates. This indicates that reflectance data collected early in the season could be useful to accurately distinguish high-stressed greens from low-stressed greens. Even though overall accuracies decreased, the results of validation for the 900-nm group showed that high-stressed greens could be reliably differentiated from low or intermediate-stressed greens (model classification: 100%; test data classification: 85%) using this group of wavebands. Figure 2 illustrates the seasonal changes in reflectance for a high-stressed green where changes in NIR (900 nm group) and SWIR regions were observed. Previous research has documented that as a plant becomes more drought-stressed, changes in the red, NIR, and SWIR reflectance are observed. Results from this research coincide with the results found by Ripple (1986) where reflectance values changed in the NIR and SWIR due to changes in leaf water content.


Figure 2
View larger version (19K):
[in this window]
[in a new window]
  		Fig. 2. Seasonal changes in reflectance of a high-stress green in 2004.

 
Average canopy temperatures are shown in Fig. 3. Low-stressed greens had higher canopy temperatures in June and August compared to high-stressed greens. This could be attributed to less irrigation being applied to the low-stressed greens compared to high-stressed greens in managing stress, which could be seen in VWC data (Fig. 4). As mentioned previously, the high-stressed greens received additional irrigation cycles outside the normal irrigation cycle.


Figure 3
View larger version (11K):
[in this window]
[in a new window]
  		Fig. 3. Seasonal changes for average canopy temperature for all stress categories.

 

Figure 4
View larger version (11K):
[in this window]
[in a new window]
  		Fig. 4. Seasonal changes of average volumetric soil water content for all stress categories. Abbreviations: VWC, volumetric soil water content.

 
As air and soil temperatures elevate in July and August creeping bentgrass becomes more stressed. On creeping bentgrass, management practices such as aerification and top dressing are performed during the cooler portions of the year because temperatures are ideal for optimal growth. If greens with potential for high stress could be detected using either an equipment-mounted radiometer or by using aerial imagery before unfavorable growth conditions arise, intense management practices, such as more aggressive, deeper aerification, could be done when the turfgrass is more vigorous and has better recuperative potential. These practices could be applied to specific greens, thus preparing these potential high-stressed greens in advance to establish greater vigor going into the hot summer months when air/soil temperatures are well above the optimum growth temperature for creeping bentgrass. Further research needs to be performed to determine a more exact threshold to measure how early or how late in the spring/summer season reflectance data can be collected and reliably detect high-stressed greens.

Received for publication January 19, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 




This article has been cited by other articles:


Home page
 Crop Sci.Home page
Y. Jiang and R. N. Carrow
Broadband Spectral Reflectance Models of Turfgrass Species and Cultivars to Drought Stress
Crop Sci., July 30, 2007; 47(4): 1611 - 1618.
[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 (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.
Agricola
Right arrow Articles by Hutto, K. C.
Right arrow Articles by Shaw, D. R.

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