HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Jiang, Y. Articles by Carrow, R. N. Search for Related Content PubMed Articles by Jiang, Y. Articles by Carrow, R. N. Agricola Articles by Jiang, Y. Articles by Carrow, R. N. Related Collections Turfgrass Remote Sensing Site-Specific Analysis

TURFGRASS SCIENCE

Broadband Spectral Reflectance Models of Turfgrass Species and Cultivars to Drought Stress

^a Dep. of Agronomy, Purdue Univ., West Lafayette, IN 47907-2054
^b Dep. of Crop and Soil Sciences, Georgia Experiment Station, Univ. of Georgia, Griffin, GA 30223-1797

^* Corresponding author (rcarrow{at}uga.edu).

	ABSTRACT

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
The objective of this study was to assess canopy broadband spectral reflectance for turfgrasses under drought stress. Optimum turf quality (TQ) and leaf firing (LF) models were developed and compared based on two, three, and five wavelength bands. Sods of bermudagrass (Cynodon dactylon L. x C. transvaalensis Burtt-Davy), seashore paspalum (Paspalum vaginatum Swartz), zoysiagrass (Zoysia japonica Steud.), and St. Augustinegrass [Stenotaphrum secundatum (Walt.) Kuntze], and seeded tall fescue (Festuca arundinacea Schreb.) were used in this study with three cultivars each of bermudagrass, seashore paspalum, and tall fescue. Traditional vegetation indices (VIs) based on two bands within 660 to 950 nm were not as sensitive as three to five broadband models using a wider band range of 660 to 1480 nm. Optimum models were cultivar specific models, even within a species. The broadband wavelength at R900 and R1200 should be considered in drought sensitive spectral models since they were most often observed and exhibited high partial R² values. These results suggest that mobile broadband spectral devices to map turfgrass responses to drought stress would benefit by the availability of three to five broadbands that could be user selected for optimum, cultivar specific models.

Abbreviations: B, bermudagrass • IR/R, reflectance in the near-infrared radiation divided by reflectance in the red range • LAI, leaf area index • LF, leaf firing • NDVI, normalized difference vegetation index • NIR, near-infrared • SA, St. Augustinegrass • SP, seashore paspalum • TF, tall fescue • TQ, turf quality • VI, vegetation index • ZZ, zoysiagrass

	INTRODUCTION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PRECISION TURFGRASS management must deal with obtaining information relative to the inherent spatial and temporal variability on a site for improved efficiency in management. Site-specific management requires specific information on the site characteristics. Vehicle-mounted spectral sensing devices are commercially available that have been used on turfgrass situations or may be adapted to provide field mapping of turfgrass characteristics (Hansen and Jørgensen, 2001; Bell et al., 2002; Reusch et al., 2002; GreenSeeker, 2006). Field-based spectral mapping must rely on spectral models predictive of specific turfgrass characteristics for either real-time use such as in variable-rate applications or for incorporation into global positioning system or geographic information system maps that are then used for management decisions. Spectral reflectance response models based on one or two broadbands have been reported for turfgrass characteristics such as turfgrass quality, color, degree of cover, drought stress, wear, and diseases (Nutter et al., 1993; Fenstermaker-Shaulis et al., 1997; Trenholm et al., 1999, 2000; Bell et al., 2002).

Spectral relationships to specific plant parameters or responses may involve individual hyperspectral or broadbands, ratios as indices, normalized indices, derivatives, or regression equations (Broge and Leblanc, 2001; Thenkabail et al., 2000; Poss et al., 2006). Numerous vegetative indices (VIs) based on use of two wavelength bands have been reported in the literature using various hyperspectral (<4 nm band width) or broadband (10–50 nm width) wavelength bands (Poss et al., 2006). When hyperspectral bands or broadbands are used for calculation of VIs based on two wavelength bands or when more than two bands are used to develop multiple regression models that relate to a plant parameter such as leaf area index (LAI) or biomass, these spectral models have been found to be species or possibly genotype specific on other crops. Thenkabail et al. (2000) demonstrated that 12 common spectral band regions provided the optimum relationships for crop biophysical characteristics across five crops. These spectral regions exhibit commonality across plants because of their relationships to specific physical and physiological plant properties (Knipling, 1970), However, when optimum models were developed using one to four wavelength variables, the models differed for crop (Thenkabail et al., 2000). They noted the same general spectral bands are involved in VIs or regression models for all plants, but species may differ in reflectance response at certain wavelengths within a general band region and, thereby, result in different best model among species. Additionally, spectral reflectance data have also been used to discriminate between species by using the same multiple wavelengths in regression models, but identifying different spectral responses among species (Clark et al., 2005). Major et al. (2003) noted that cellulose, lignin, and protein absorption in the near-infrared (NIR) region could result in significant differences among species and perhaps genotypes.

Hansen and Jørgensen (2001) and Reusch et al.(2002) described field-mapping equipment capable of using up to five user-selected wavelength bands. Flexibility in band selection (wavelength bands used and number of bands) could be important if the optimum bands differed with species and genotype; using more than two bands would allow for potentially more accurate models. Recently, best models were reported using hyperspectral bands for turfgrass drought stress that were based on using more than two wavelength bands at the species level (Hutto et al., 2006) and among species and cultivars (Jiang and Carrow, 2005). Leaf firing and turfgrass quality relationships to spectral responses were somewhat stronger when more than two bands were used, which supports the results of Thenkabail et al. (2000) for other crops and plant parameter relationships where inclusion of more bands improved the model. Additionally, the specific bands exhibiting best correlations have been found to vary with species (Trenholm et al., 2000; broadband data, two species) or species and genotype (Jiang and Carrow, 2005; hyperspectral band data).

Due to the potential for more accurate or flexible field spectral mapping models using two or more user-selected bands and the possibility for optimum models to be inter- or intraspecific rather than a uniform model for all grasses, the objective of this study was to assess canopy broadband spectral reflectance models (two, three, and five broadband models) for turf quality (TQ) and leaf firing (LF) responses to drought stress across turfgrass species and within species. Species included bermudagrass (Cynodon dactylon L. x C. transvaalensis Burtt-Davy), seashore paspalum (Paspalum vaginatum Swartz), zoysiagrass (Zoysia japonica Steud.), and St. Augustinegrass [Stenotaphrum secundatum (Walt.) Kuntze], which are commonly used warm-season turfgrasses, and tall fescue (Festuca arundinacea Schreb.) a cool-season turfgrass that can grow in the upper to midsouthern climates of the USA.

	MATERIALS AND METHODS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Plant Materials
Sods were collected from mature field plots on 24 Apr. 2001 at the Griffin Campus of the University of Georgia including bermudagrass (B; ‘TifSport’, ‘TifEagle’, ‘Tifway’), seashore paspalum (SP; ‘Sea Isle 1’, ‘Sea Isle 2000’, ‘Temple’), zoysiagrass (ZZ; ‘Meyer’), and St. Augustinegrass (SA; ‘Palmetto’), and seeded tall fescue (TF; ‘Plantation’, ‘Greystone’, ‘TULSA II’). All soil was removed from sods and grasses were grown in barrels filled with Profile (Profile Products LLC, Buffalo Grove, IL), a calcined illite clay and amorphous silica material, to provide a uniform root zone media based on the work of van Bavel et al. (1978). For seeded grasses, the quantity of seed was 10 g per barrel. Barrels were 0.65-m diameter by 0.90-m depth with surface area of 0.24 m² with a volume of 206 L and with holes in the bottom for drainage. All grasses were mowed twice weekly using a hand clipper at 2.54 cm (B, ZZ, SP) or 5.08 cm (SA, TF) with clippings removed. Granular fertilizer (N–P–K) applied to the barrels (in terms of kg N ha^–1) was 97.6 N on 2 April (6–2–0), 48.8 N on 28 April (15–5–15), 24.4 N on 9 May (15–5–15), 24.4 N on 16 May (15–5–15), 48.8 N on 25 May (6–2–0), and 24.4 N on 31 May (13–0–13). Irrigation was applied as necessary to maintain healthy turfgrass and in sufficient volume to allow drainage.

Drought Stress and Spectral Reflectance
Drought stress was initiated after bringing all containers to well-watered conditions and then withholding water during dry-down periods. A sensor-controlled, removable rainout shelter facility (12.8 by 30.5 m) was used to maintain dry-down conditions. The rainout shelter was placed in its original position and was able to move automatically to cover all barrels when it rained and to move back after raining. The study consisted of three dry-downs with each period from 11 to 15 d. After each dry-down, TQ was allowed to fully recover to a nonstress level before the next dry-down started. The dry-down periods were 10–25 July, 20–31 Aug., and 10–21 Sept. 2001. The duration of each dry-down was determined by how rapidly TQ declined in response to drought stress, which allowed a range of drought stress symptoms occurring among different grasses for collection of TQ, LF, and canopy spectral reflectance data with all grasses exhibiting LF of at least >10% at the end of each dry-down. Data for each grass for drought stress responses in terms of TQ and LF are presented by Jiang and Carrow (2005), with LF at the end of dry-downs ranging from 12.3% (Temple SP) to 55.2% (Tulsa II TF).

Spectral responses were related to TQ and LF as the plant parameters. Turf quality was rated visually based on color, shoot density, and uniformity, where 1 = brown, dead turf and 9 = ideal dark-green color, density, and uniformity for the species. Leaf firing refers to leaf chlorosis starting at leaf tips and margins. Initial injury is a yellowing but often progresses into a tan or brown color with death of the tan or brown areas. Leaf firing rating was based on percentage of leaves exhibiting the above symptoms (Carrow and Duncan, 2003).

Canopy spectral reflectance was collected with a CropScan Multispectral Radiometer (CROPSCAN, Inc., Rochester, MN.). This broadband spectral device determined reflectance in a 10 to 15 nm band width centered around 660, 710, 760, 810, 900, 950, 1200, and 1480 nm with matched upward and downward sensor arrays to minimize solar radiation effects with the incident radiation used as a baseline for the reflected radiation in the same band. The 950-, 1200-, and 1480-nm bands were chosen since they represent water bands. The radiometer was held at a height of 1 m above canopy and measured about a 0.12-m² area inside barrel. The reflectance readings were taken at 1300 h to minimize background noise. All measurements were taken daily during each dry-down period. The following indices related to canopy status were developed based on reflectance:

Normalized difference vegetation index 1

(NDVI1) = (R760 – R710)/(R760 + R710) (modified from Gamon and Surfus, 1999)

Normalized difference vegetation index 2

(NDVI2) = (R950 – R660)/(R950 + R660) (modified from Trenholm et al., 1999)

Stress index 1 = R710/R760

Stress index 2 = R710/R810 (modified, Trenholm et al., 1999)

Leaf area index (reflectance in the near-infrared radiation divided by reflectance in the red range [IR/R]) = R950/R660 (modified from Trenholm et al., 1999)

Water band index = R700/R950 (modified from Penuelas et al., 1993)

Experiment Design and Data Analysis
The experiment was a randomized complete block design with repeated dry-down cycles. A total of 11 grasses (five species and selected cultivars or ecotypes for some species) was used with each individual grass replicated three times (three barrels) within each single dry-down. Correlation coefficients (r), coefficient of determination (R²), and general linear model procedures were determined using Statistical Analysis System (SAS Institute, 1987) to develop the relationship between canopy characteristics and reflectance and models for TQ against reflectance. The optimum models were then summarized for each individual grass, each species, and all grasses combined. The partial coefficient of determination (R²) for individual wavelength bands exhibited in the model was analyzed to determine relative importance of a given band in a model.

Data of TQ, LF, and canopy spectral reflectance within each treatment day were averaged for a single grass (cultivar) across the three replicates, and the averaged data were combined across three dry-downs and used for statistical analysis and model development for each individual grass. For the general model of each species, data from individual grass within the species were combined. Data from all species were combined to generate a model over all 11 grasses. Spectral data collected on low incident radiation days (<300 W/m²) were not included in the analysis due to limited reliability of crop reflectance and instrument limitations under low radiation.

	RESULTS

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
All grasses exhibited a substantial decline in TQ and increased LF under the three drought stress periods with considerable differences among grasses. The detailed results of TQ and LF for each grass under drought stress have been described (Jiang and Carrow, 2005). Differences in TQ (6.8–4.9 with 9.0 ideal) and LF (12.3–55.2%) for all grasses under drought stress provided a range of responses for broadband spectral reflectance relationships with these plant responses.

Vegetation Indices
In assessment of two broadband, wavelength VIs for correlation with TQ and LF, only four grasses exhibited a significant VI for TQ and none for LF (Table 1). The best VI differed for all four grasses which included three species with Stress 2 (TifSport B, r = –0.56), IR/R (Sea Isle 1 SP, r = 0.65), NDVI1 (Temple SP, r = 0.62), and Stress 1 (Palmetto SA, r = 0.46). Broadband wavelengths represented in the significant VI relationships for TQ were R710 and R810 nm (Stress 2), R660 and R950 nm (IR/R), and R710 and R760 nm (Stress 1 and NDVI1).

	DISCUSSION

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The most common spectral models for relating to plant characteristics are VIs. Common characteristics of VIs are (i) whether broadband or narrow band-based, they are almost always based on two spectral bands, and (ii) they use bands between 430 and 1100 nm (Poss et al., 2006). Normalized difference vegetation index, determined by broadband spectral reflectance under stresses other than drought, is a commonly used indicator for green biomass, leaf area, and stress status in plants (Daughtry et al., 1992; Gamon et al., 1995; Penuelas et al., 1997) and was found correlated to TQ and turf color under various nondrought stresses (Trenholm et al., 1999; Bell et al., 2002; Jiang et al., 2003) or biomass (Fenstermaker-Shaulis et al., 1997). The VI IR/R, associated with shoot biomass, and the Stress 1 and 2 VIs have also been related to TQ (Daughtry et al., 1992; Trenholm et al., 1999, 2000).

The broadband-based VIs used in our drought stress study demonstrated a significant relationship with TQ for only four grasses and no relationship with LF (Table 1). The best VI for the four grasses differed with each grass. The most common broadband involved was R710 in three of the VIs. Broadband spectral reflectance data and turfgrass TQ or LF as affected by drought as the specific stress are limited. On drought stressed tall fescue, Fenstermaker-Shaulis et al. (1997) observed significant relationships for broadband-based NDVI (R600–650; R800–890) versus biomass (r = 0.55), canopy temperature (r = 0.54), and tissue moisture content (r = 0.90). Hutto et al. (2006:1566) on drought-stressed bentgrass (Agrostis stolonifera L.) reported that "broadband widths or vegetative indices were not successful in differentiating between stresses." Our results and the above reports suggest that broadband VIs based on two bands may be better for some species or cultivars than for others for estimating TQ and drought stress relationships; when LF is the plant parameter of interest, VIs may be less accurate than for TQ; the best VI to use may be dependent on the cultivar; and there is no consistent drought-sensitive VI, but that the R710 band may be useful, which is a red-edge band sensitive to vegetation stress and a band region common in optimum hyperspectral models of many plants (Thenkabail et al., 2004).

As the number of broadband wavelengths increased from two, three, and five bands (with the three and five band models developed by regression) the number of significant models for TQ out of the 11 grasses increased from 4, 11, and 11, respectively (Table 1, 2, and 6). Likewise, the R² or partial R² value increased for all grasses, as the number of broadband wavelengths in the model increased, except Temple SP where NDVI1 had the highest R². For TQ spectral relationships of grasses under drought stress, each grass had a unique optimum model for the two (VIs), three, and five broadband wavelength models (Tables 1, 2, and 6).

For LF and spectral relationships of grasses exposed to drought stress, the number of significant models observed as the number of broadband wavelengths increased was two bands (0), three bands (10), and five bands (6). While inclusion of five bands resulted in only six significant models, four of these grasses showed a substantial increase in partial R² compared to the three band model (Tables 1, 4, 8). The lower sensitivity of spectral models in relationship with LF versus TQ may be due to the observation that as drought was imposed, TQ was affected earlier in the dry-down than LF.

This research demonstrated that for TQ and LF broadband spectral models in response to drought stress, a stronger model generally resulted by inclusion of more wavelength bands and the optimum model was specific to the grass. Our findings were in agreement with Thenkabail et al. (2000) for five crop species where the biophysical characteristics measured were wet biomass, LAI, plant height, and yield. They reported that models differed with crop and that the four band models performed marginally better than two band models. In a later study, Thenkabail et al. (2004) reported that five bands accounted for a high percentage of the variability in models to separate species and separation accuracies of greater that 90% occurred when using 13 to 22 bands. The wavelength bands in their study were normally about 10 nm in width since they included several adjacent hyperspectral bands. Bell et al. (2002) and Trenholm et al.(2000), using a C₃ and C₄ grass species, found that the R² for NDVI versus turf color, percentage of plant cover, or TQ varied considerably with species.

While Hutto et al. (2006) did not find any significant broadband spectral relationships of turfgrass to drought stress, they were able when using hyperspectral reflectance data in a discriminate analysis approach to identify stressed areas at >98% of the time by using 14 wavebands between 861 and 887 nm or 33 bands between 719 and 799 nm. Jiang and Carrow (2005) focused on hyperspectral approaches, in a companion paper to this one, and obtained significant spectral reflectance models for TQ on seven out of 11 grasses (R² range of 0.32–0.70) and for LF on eight grasses (R² range of 0.27–0.71). The models differed with cultivar and included two to three narrow band wavelengths selected from the 10 wavelengths between 400 and 1100 nm with the best correlations of reflectance to TQ or LF.

The VIs we used were based on broadband wavelengths between R660 to R950 nm, which is within the spectral range of R430 to R1100 nm used for the various VIs published in the literature (Poss et al., 2006). The three and five band regression models for TQ and LF, however, included all broadbands from R660 to R950 plus the two water bands of R1200 and R1440 nm. Based on the three and five band models for TQ spectral relationships, the most prevalent bands in the models were the water band at R1200 nm and the R900-nm band (Table 3 and 7). Broadband wavelengths most common in the LF spectral models using three or five bands were R1200, R900, and R950 nm, where R950 nm is another water band sensitive to tissue moisture content (Tables 5 and 9) (Thenkabail et al., 2000). Other studies also found that reflectance at the 950- to 970-nm region were an indicator of plant water status (Penuelas et al., 1993). Thenkabail et al. (2000, 2004) noted that the 915-nm band was one of the bands often observed in optimal hyperspectral models and represents the NIR reflectance peak that is sensitive to total chlorophyll, biomass, LAI, and protein. The R900 band range was not included in the bands used by Trenholm et al. (1999, 2000) or in other turfgrass studies involving broadband reflectance (Fenstermaker-Shaulis et al., 1997; Bell et al., 2002). Further investigation of this band in VIs or multiple regression models for relationship to drought stress or other turfgrass parameters may be worthwhile.

This research indicates that broadband-based spectral data can be used to monitor TQ and LF plant responses to increasing drought stress, but (i) traditional VIs based on two bands within 660 to 950 nm may not be as sensitive as three to five band models using a wider band range of 660 to 1480 nm, (ii) cultivar specific models, even within a species, should be considered to obtain highest sensitivity, and (iii) the bands at R900 and R1200 should be considered in drought-sensitive spectral models. Additionally, mobile broadband spectral devices to map turfgrass responses to drought stress would benefit by the availability of three to five broadbands that could be user selected for optimum, cultivar specific models. To develop the optimum band combination, a site could be mapped using all sensors and then the optimum model determined based on multiple regression methods.

	ACKNOWLEDGMENTS

 
Funding for this research from The Toro Company is gratefully acknowledged.

	NOTES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher. Permission for printing and for reprinting the material contained herein has been obtained by the publisher.

Received for publication October 6, 2006.

	REFERENCES

TOP NOTES ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

• Bell, G.E., D.L. Martin, S.G. Wiese, D.D. Dobson, M.W. Smith, M.L. Stone, and J.B. Solie. 2002. Vehicle-mounted optical sensing: An objective means for evaluating turf quality. Crop Sci. 42:197–201.[Abstract/Free Full Text]
• Broge, N.H., and E. Leblanc. 2001. Comparing prediction power and stability of broadband and hyperspectral vegetation indices for estimation of green leaf area index and canopy chlorophyll density. Remote Sens. Environ. 76:156–172.[CrossRef]
• Carrow, R.N., and R.R. Duncan. 2003. Improving drought resistant and persistence in turf-type tall fescues. Crop Sci. 43:978–984.[Abstract/Free Full Text]
• Clark, M.L., D.A. Roberts, and D.B. Clark. 2005. Hyperspectral discrimination of tropical rain forest tree species at leaf to crown scales. Remote Sens. Environ. 96:375–398.[CrossRef]
• Daughtry, C.T.S., K.P. Gallon, N.S. Goward, S.D. Prince, and W.P. Kustas. 1992. Spectral estimates of absorbed radiation and phytomass production in corn and soybean canopies. Remote Sens. Environ. 39:141–152.[CrossRef]
• Fenstermaker-Shaulis, L.K., A. Leskys, and D.A. Devitt. 1997. Utilization of remotely sensed data to map and evaluate turfgrass stress associated with drought. J. Turfgrass Manage. 2:65–80.
• Gamon, J.A., C.B. Field, M.L. Goulden, K.L. Griffin, A.E. Hartley, G. Joel, J. Penuelas, and R. Valentini. 1995. Relationship between NDVI, canopy structure, and photosynthesis in three California vegetation types. Ecol. Applic. 5:28–41.[CrossRef]
• Gamon, J.A., and J.S. Surfus. 1999. Assessing leaf pigment content and activity with a reflectometer. New Phytol. 143:105–117.[CrossRef][Web of Science]
• GreenSeeker. 2006. GreenSeeker sensor. Available at www.ntechindustries.com (verified 6 June 2007). NTech Industries, Ukiah, CA.
• Hansen, P.M., and R.N. Jørgensen. 2001. The RISØ cropassessor: An idea to a low-cost, robust, simple, and modular measuring device based on existing technology for monitoring the spatial field crop variation. p. 37–42. In G. Grenier and S. Blackmore (ed.) ECPA 2001. Proc. of the 3rd Eur. Conf. on Precision Agriculture, Montpellier, France. 18–20 June 2001. Agro Montpellier, Montpellier, France.
• Hutto, K.C., R.L. King, J.D. Byrd, and D.R. Shaw. 2006. Implementation of hyerspectral radiometry in irrigation management of creeping bentgrass putting greens. Crop Sci. 46:1564–1569.[Abstract/Free Full Text]
• Jiang, Y.W., and R.N. Carrow. 2005. Assessment of narrow-band canopy spectral reflectance and turfgrass performance under drought stress. HortScience 40:242–243.[Abstract/Free Full Text]
• Jiang, Y.W., R.N. Carrow, and R.R. Duncan. 2003. Correlation analysis procedures for canopy spectral reflectance data of seashore paspalum under traffic stress. J. Am. Soc. Hortic. Sci. 128:343–348.
• Knipling, E.B. 1970. Physical and physiological bases for the reflectance of visible and near-infrared radiation from vegetation. Remote Sens. Environ. 1:155–159.[CrossRef]
• Major, D.J., R. Baumeister, A. Touré, and S. Zhao. 2003. Methods of measuring and characterizing the effects of stresses on leaf and canopy signatures. In T. VanToai et al. (ed.) Digital imaging and spectral techniques: Applications to precision agriculture and crop physiology. ASA Spec. Publ. 66. ASA, CSSA, and SSSA, Madison, WI.
• Nutter, F.W., Jr., M.L. Gleason, J.H. Jenco, and N.C. Christians. 1993. Assessing the accuracy, inter-rater repeatability, and inter-rater reliability of disease assessment systems. Phytopathology 83:806–812.[CrossRef][Web of Science]
• Penuelas, J., I. Filella, C. Biel, L. Serrano, and R. Save. 1993. The reflectance at the 950–970 nm region as an indicator of plant water status. Int. J. Remote Sens. 14:1887–1905.[CrossRef]
• Penuelas, J., R. Isla, I. Filella, and J.L. Araus. 1997. Visible and near-infrared reflectance assessment of salinity effects on barley. Crop Sci. 37:198–202.[Abstract/Free Full Text]
• Poss, J.A., W.B. Russell, and C.M. Grieve. 2006. Estimating yields of salt- and water-stressed forages with remote sensing in the visible and near infrared. J. Environ. Qual. 35:1060–1071.[Abstract/Free Full Text]
• Reusch, S., A. Link, and J. Lammel. 2002. Tractor-mounted multispectral scanner for remote field investigation. In P.C. Roberts (ed.) Proc. of the 6th Int. Conf. on Precision Agriculture and other Precision Resources Management, Minneapolis, MN. 14–17 July 2002. ASA, CSSA, and SSSA, Madison, WI.
• SAS Institute. 1987. SAS user's guide: Statistics. 6th ed. SAS Inst., Cary, NC.
• Thenkabail, P.S., E.A. Enclona, M.S. Ashton, and B.V.D. Meer. 2004. Accuracy assessments of hyperspectral waveband performance for vegetation analysis applications. Remote Sens. Environ. 91:354–376.[CrossRef]
• Thenkabail, P.S., R.B. Smith, and E. De Pauw. 2000. Hyperspectral vegetation indices and their relationships to agriculture and crop characteristics. Remote Sens. Environ. 71:158–182.[CrossRef]
• Trenholm, L.E., R.N. Carrow, and R.R. Duncan. 1999. Relationship of multispectral radiometry data qualitative data in turfgrass research. Crop Sci. 39:763–769.[Abstract/Free Full Text]
• Trenholm, L.E., M.J. Schlossberg, G. Lee, W. Parks, and S.A. Geer. 2000. An evaluation of multi-spectral responses on selected turfgrass species. Int. J. Remote Sens. 21:709–721.[CrossRef]
• van Bavel, C.H.M., R. Lascano, and D.R. Wilson. 1978. Water relations of fritted clay. Soil Sci. Soc. Am. J. 42:657–659.[Abstract/Free Full Text]

This article has been cited by other articles:

				A. R. Johnsen, B. P. Horgan, B. S. Hulke, and V. Cline Evaluation of Remote Sensing to Measure Plant Stress in Creeping Bentgrass (Agrostis stolonifera L.) Fairways Crop Sci., October 22, 2009; 49(6): 2261 - 2274. [Abstract] [Full Text] [PDF]

				C. Chen, S. Lu, Y. Chen, Z. Wang, Y. Niu, and Z. Guo A Gamma-ray-Induced Dwarf Mutant from Seeded Bermudagrass and Its Physiological Responses to Drought Stress J. Amer. Soc. Hort. Sci., January 1, 2009; 134(1): 22 - 30. [Abstract] [Full Text] [PDF]

				S. Lu, Z. Wang, Y. Niu, Z. Guo, and B. Huang Antioxidant Responses of Radiation-induced Dwarf Mutants of Bermudagrass to Drought Stress J. Amer. Soc. Hort. Sci., May 1, 2008; 133(3): 360 - 366. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) Free Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Jiang, Y. Articles by Carrow, R. N. Search for Related Content PubMed Articles by Jiang, Y. Articles by Carrow, R. N. Agricola Articles by Jiang, Y. Articles by Carrow, R. N. Related Collections Turfgrass Remote Sensing Site-Specific Analysis