Published online 8 January 2009
Published in Agron J 101:226-231 (2009)
DOI: 10.2134/agronj2007.0167
© 2009 American Society of Agronomy
677 S. Segoe Rd., Madison, WI 53711 USA
Nondestructive Measurement of Grapevine Leaf Area by Ground Normalized Difference Vegetation Index
Rachid Drissia,*,
Jean-Pascal Goutoulya,
Dominique Forgetb and
Jean-Pierre Gaudillerea
a ECAV, UMR EGFV, Institut des Sciences de la Vigne et du Vin de Bordeaux, Domaine INRA de la Grande Ferrade, BP 81, 33883 Villenave d'Ornon Cedex, France
b Chateau Couhins, Chemin de la Gravette, BP 81, 33883 Villenave d'Ornon Cedex, France
* Corresponding author (drissirachid{at}yahoo.fr).
 |
ABSTRACT
|
|---|
Vine leaf area index has a great impact on berry quality. This study was conducted to determine whether vine leaf area index could be estimated, and mapped through normalized difference vegetation index (NDVI) ground-based measurements. The NDVI measurements were performed using a Greenseeker (N-Tech Industires, Ukiah, CA and Oklahoma State Univ., Stillwater), pointed sideways at the vertical shoot positioned vines [Vitis vinifera (L.)] at Bordeaux, France. Canopy gap fraction and vertical leaf area index (VLAI) measurements were also performed. Plot NDVI maps were obtained by linking the GreenSeeker to a GPS during measurements. The NDVI delivered by the sensor was sensitive to the variations of vertical leaf area index and gap fraction of the canopy, that is, vine vigor. The GreenSeeker was successfully used to carry out a follow-up of the foliar growth of the vine, but with many precautions. The maps obtained showed relative variations of vigor at an intraplot level, enabling access to relevant information for better vineyard management.
Abbreviations: LAI, leaf area index • NDVI, normalized difference vegetation index • PW, pruning weight • VLAI, vertical leaf area index
Received for publication May 17, 2007.
|
INTRODUCTION
|
|---|
Vine leaf area is an important agronomical parameter as it is related to photosynthetic capacity, water use, and grape microclimate. Berry maturation being related to photosynthesis, Murisier and Zufferey (1997) showed that the leaf area to fruit weight ratio is a relevant indicator of fruit sugar content of the berry at ripeness. Vineyard water balance is closely related to sunlight interception by the canopy (Riou et al., 1994). In addition, it has been shown that enological potential of berries is related to light environment within grapevine canopies, which itself is influenced by the distribution and the foliar density (Smart, 1985; Schultz, 1995; Dokoozlian and Kliewer, 1996; Mabrouk and Sinoquet, 1998). Therefore, a great number of cultural practices tend to regulate density and foliar distribution in the canopy so as to produce wines of quality.
Leaf area can be estimated by several techniques. To take into account local variability of this parameter, two categories of methods of leaf area estimate are defined: direct and indirect. The shoot-pruned weight allows an estimate of the foliar biomass, yet it provides information after harvest. For seasonal growth, among the direct methods, a particular technique consists in measuring length and diameter of vine shoots (Castelan-Estrada et al., 2002). Although reliable, this technique is time-consuming when implemented at a plot scale (Tregoat et al., 2001). Direct destructive methods, because of their nature, offer limited interest for growers, and even for scientists wishing to follow vine growth.
Indirect methods use optical measurements based on the relation between canopy structure and the radiation interception. These techniques measure the gap fraction, which is the proportion of transmitted light which is not blocked by foliage in a range of azimuthal directions. Leaf area is then considered starting from models including the gap fraction as a parameter, with a calibration performed locally. Grantz and Williams (1993), Sommer and Lang (1994), and Ollat et al. (1998) tested various commercially available sensors to measure the gap fraction within a vineyard. These methods are limited by measuring conditions constraints, as they require diffuse light conditions and avoidance of direct solar illumination (Ollat et al., 1998). Moreover, leaf area index (LAI) mapping on greater scales results only from a few measurements. Alternative methods are nowadays necessary to consider the leaf area at various spatial scales, reasonable cost, and in a precise way.
Remote-sensing imagery is based on the reflected electromagnetic radiation by the leaves. Most commonly employed techniques collect multispectral visual data by plane or satellite. Data is then converted into measurements of canopy density, using a vegetation index. The calculation of this index is based on the fact that photosynthetic active vegetation is characterized by a strong absorptance of the incidental light in the visible red wavelengths, and a strong reflectance in the near-infrared wavelengths. In recent years many remote-sensing studies have shown that the airborne and satellite NDVI is correlated with vertical shoot positioned (VSP) grapevine LAI (Baldy et al., 1996; Johnson et al., 1996; Dobrowski et al., 2002; Johnson et al., 2003). However, airborne sensors view canopy from above, and image treatment reveals its limits for vineyards where grass, weeds, or cover crops can interfere with vine NDVI.
Recently, N-Tech Industries, Ukiah, CA, and Oklahoma State University, Stillwater, developed a ground-based apparatus: the GreenSeeker. It measures the NDVI calculated from reflectances:
where NIR is the reflectance at near-infrared wavelengths and R is the reflectance at red wavelengths (Rouse et al., 1974). The GreenSeeker is an active sensor, so varying light conditions have very little influence on measurements (Jones, 2004). The GreenSeeker is efficiently used on cereals by producers, to improve N fertilization (Mullen et al., 2003). It is pointed toward the ground; what is needed for grapevine management is a sensor that is able to estimate the vertical distribution of leaves. The aim of this study was to evaluate the relevance of the NDVI delivered by the GreenSeeker to evaluate vine vigor considering two parameters, canopy gap fraction and VLAI. In this study, we refer to canopy gap fraction as the percentage of view unobstructed by canopy in a perpendicular direction to trellis plan, and define VLAI as leaf area per unit of vertical (trellis) area. We will also approach the aspects related to NDVI mapping using the GreenSeeker.
|
MATERIALS AND METHODS
|
|---|
Field Site
The NDVI measurements were conducted near Bordeaux in three different plots from the Institut National de la Recherche Agronomique (INRA) experimental vineyards. Besides the differences of clones, root-stocks, and planting density, these plots were identically trained: cane-pruned, vertical-shoot positioned Vitis vinifera L. cultivar Merlot topped at 150 cm above the ground, with a 50-cm trunk and a north-south row orientation.
Plot 1 was part of a first vineyard (V 1). It had a gravelly soil, planted with two different Merlot clones, 347 and 181, grafted to Gravesac and 101–14 MG, respectively. The planting density was 6600 vines ha–1 with a row by vine spacing of 150 by 100 cm. Plot 2 and plot 3 were part of a second vineyard (V 2) of 15 ha approximately. Both were planted with Merlot (clone 181), grafted to Fercal (plot 2) and to 161–49 C (plot 3). The planting density was 6200 vines ha–1 with a row by vine spacing of 160 by 100 cm.
Normalized Difference Vegetation Index Measurements
The NDVI measurements were performed using a GreenSeeker RT 100 System. This hand-held unit optical sensor uses high intensity light emitting diodes (LEDs) at 660 nm (R) and 770 nm (NIR), pulsed at high frequency. The magnitude of the light reflected off the target is measured by a photodiode detector. Electronic filters remove all background illumination (Solie et al., 2002). The sensor is temperature stable (Jones, 2004), and scans a 61 by 1 cm area. The NDVI measurements offer very little variation when the target is located between 85 and 115 cm from the sensor (Jones, 2004). Measurements occur with a frequency of 100 Hz and are averaged at 10 Hz. However, the only data delivered by the sensor are the NDVI and R/NIR ratio, with no access to information concerning reflectance values.
Unlike standard remote sensing and standard use of the GreenSeeker for which the aiming is vertical toward the ground, it was oriented horizontally toward the canopy at 1 m distance for all experiments. Thus, a grass cover below the vines could not interfere with the NDVI measurements. The GreenSeeker was mounted on a straddler apparatus for experiments in the first vineyard, and on a straddler-tractor in the second vineyard, allowing the mounting of a screen on the other side of the vine row where measurements where executed. The use of a screen placed behind the canopy was necessary to differentiate canopy from background interference. After testing different color screens, a white color screen with a low NDVI value (
0.08) was selected for use along with the sensor for all NDVI measurements.
Experiments
Experiment 1
Merlot leaves were brought together to set up two strips of a single layer of leaves: using young leaves for the first strip (mean of sum of length of the two superior lateral veins = 14.5 cm, SD = 1.45) and older leaves for the second (sum of length of the two superior lateral veins = 23.4, SD = 1.44). The GreenSeeker was placed 1 m away from a white screen in a dark room so as to visualize the light beam, and each strip was vertically down-inserted progressively into the surface analyzed by the GreenSeeker (in front of the white screen) until occupying the entire surface scanned by the sensor. The NDVI measurements were performed at each step of insertion.
Experiment 2
On 26 July 2004 a progressive and consistent defoliation of a vine was performed on plot 1 (Merlot 347) in three steps, removing approximately a third of the total leaf area at each step, and at last the green clusters (fourth step). Space considered occupied by the canopy was 1 by 1 m. To encompass the entire canopy without overlapping, the GreenSeeker was placed at two measurement heights (H and L in Fig. 1
). Each 100 by 61 cm strip (L, H) was divided into five rectangular areas of 20 by 61 cm. The NDVI measurements were completed every centimeter in vine row direction. At each step removed leaves were placed in separate bags corresponding to the rectangular area of the canopy in which they were comprised. Pictures of the vine were taken with a red screen set as a background consecutively to NDVI measurements.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 1. Schematic protocol to collect normalized difference vegetation index (NDVI) and vertical leaf area index (VLAI) measurements in Exp. 2 and 3. The VLAI is estimated on the leaves present in each of the 10 0.2 by 0.61 m rectangular areas. For the same area NDVI measurements are performed each centimeter: for this size area NDVI is averaged on 20 measurements. L: low canopy area scanned by the GreenSeeker; M: medium area scanned by the GreenSeeker; H: high canopy area scanned by the GreenSeeker.
|
|
Experiment 3
The NDVI measurements were performed during growth period in 2004 (12, 17, and 24 May; 3, 10, and 23 June; 12 July) on nine vines on plot 1 (Merlot 181). Leaf area was systematically determined per 20 by 61 cm rectangular areas (10 per vine). The NDVI measurements were executed every centimeter in row direction.
Canopy Gap Fraction Measurements
Canopy pictures were taken using a digital camera (Exp. 2). A red screen was placed vertically at the background of the vine row to facilitate image treatment. Pictures were performed at solar noon to avoid projection of vine shade on the red screen. Picture treatment was then performed using ImageJ (source code is available at http://rsb.info.nih.gov/ij/), extracting the (red screen) background and thresholding the corresponding pixels. Their enumeration led to canopy gap determination.
Vertical Leaf Area Index Measurements
Two different direct leaf area measurements were performed. For Exp. 2, leaves were thinned, then placed in a cooler in a specific bag corresponding to each rectangular area of the canopy (Fig. 1). One hour after NDVI measurements, they were weighed. For Exp. 3, the length of the two superior lateral veins was measured for each leaf (Carbonneau, 1976) present in each given rectangular area of the canopy (Fig. 1). Leaves located in between two rectangular areas were attributed to one only, after visually determining the area where the major part of the leaf was located.
Leaf area was calculated by using regression equations between the leaf area (determined by planimetry) and the fresh weight of a subsample of leaves (Exp. 1) and between the leaf area (determined by planimetry) and the sum of lengths of the two superior lateral veins of a subsample of leaves (Exp. 2). The leaves used to establish the regression equations were sampled the day before each experiment, on nearby vines.
Pruning Weights Measurements
Pruning weights were determined for 96 experimental plots of three plants each, regularly spaced on plot 2. Vines were pruned and then pruned vine shoots were weighed using a hand-held balance.
|
RESULTS AND DISCUSSION
|
|---|
Experiment 1
This experiment tested NDVI response to varying gap fraction for one layer of vertical-positioned leaves. The surface scanned by the sensor can be assimilated to a unit area of 0.01 by 0.61 m. Consequently, the maximum NDVI value was reached for a 0% gap fraction, which is determined by the spectral characteristics of the object filling the scanned surface. For gap fraction values >0%, the pixel is composite. As shown by Fig. 2
, NDVI response to varying gap fraction follows a sigmoid trend.
View larger version (9K):
[in this window]
[in a new window]
|
Fig. 2. Relationship between normalized difference vegetation index (NDVI) and measured gap fraction per 0.01 by 0.61 m rectangular area, using a single layer of vine leaves (young in white, older in black).
|
|
This trend reveals a lower sensitivity of the GreenSeeker when gap fraction is >80%. It also appears that for a low gap fraction value ( <20%), NDVI signal reaches a plateau. Color difference between young and older leaves appears to have little impact on NDVI, and only for gap fraction below 35%. Maximum difference observed (for 0% gap fraction) is of 0.07. This protocol shows that canopy gap is a relevant parameter considering NDVI response, however the relationship observed is not linear.
Experiment 2
This experiment tested NDVI response to varying gap fraction and VLAI following grapevine defoliations; the Merlot vines were defoliated in three steps to change both gap fraction and VLAI. The correlation between NDVI and gap fraction (NDVI = 9 x 10–5 x Gap fraction2– 0.0179 x Gap fraction + 0.9717, r2 = 0.94, P < 0.01, Fig. 3A
) is more significant than between NDVI and VLAI (r2 = 0.85, P < 0.01, Fig. 3B). The relationship between VLAI and NDVI occurs as a result of foliar architecture within grapevine canopy, which correlate VLAI and gap fraction (in this experiment, Gap fraction (%) = 0.0393 x VLAI2 – 0.4115 x VLAI + 1.0797, r2 = 0.89, P < 0.01).
View larger version (17K):
[in this window]
[in a new window]
|
Fig. 3. (A) Relationship between mean normalized difference vegetation index (NDVI) and measured gap fraction per 0.2 by 0.61 m rectangular area. (B) Relationship between mean NDVI and measured vertical leaf area index (VLAI) per 0.2 by 0.61 m rectangular area.
|
|
The NDVI tends to saturate for VLAI values above 4, which corresponds to very low gap fraction values (<20%). It also appears that VLAI values below 1 are poorly characterized, which is explained by the fact that the sensor is not sensitive to high gap fraction values (above 80%, Fig. 2 and 3A). Before ripening, green grape clusters and shoots can contribute to an overestimation of VLAI based on NDVI, as they are responsible for NDVI values comprised between 0.08 and 0.26 in this experiment.
Local foliar density is defined as leaf area per unit of vertical (trellis) area occupied by leaves, when viewing the canopy from aside. This formula was developed to characterize canopy depth more precisely:
where LFD is the local foliar density per 20 by 61 cm rectangular area. Multiple regression at the 5% level showed a very high significance for gap fraction (P < 0.001) and local foliar density (P < 0.001) contributing to NDVI (r2 = 0.954). It also appeared that gap fraction contribution (t = –23.785) was greater than that of local foliar density (t = 4.920). As a result of both experiments, it is concluded that the sensor is an effective tool for characterizing vine vigor, that is, VLAI and canopy gap.
Experiment 3
This experiment followed vine growth over the season using NDVI measurements. The relation between VLAI and NDVI was researched in field conditions (Fig. 4
) during vegetative growth period of vines in the plot 1 (Merlot 181). At 20 by 61 cm scale, VLAI is not very precisely determined, and leaf area and NDVI measurements were integrated on a larger area (1 by 0.61 m). The response curve is very similar to that obtained by leaf removal (Fig. 3A). Again, NDVI is saturated for VLAI values above 4.
View larger version (11K):
[in this window]
[in a new window]
|
Fig. 4. Relationship between mean normalized difference vegetation index (NDVI) and measured vertical leaf area index (VLAI) per 1 by 0.61 m rectangular area, including all the data collected from 12 May to 12 July 2004 corresponding to the lower part of the canopy.
|
|
This experiment also points out the fact that a vine-growth follow-up using the GreenSeeker is limited in time. With time the lower part of the canopy saturated the response and when the height of the shoots was >0.61 m, the sensor view had to be changed to scan the growing zone. The signals from the two areas are not additive, and as the lower part of the canopy is saturated, a major part of the higher one fills <20% of the GreenSeeker field of view. In addition, each canopy trimming induced a change in the relationship, as more laterals thrived, that is, the relation between VLAI and gap fraction evolved. Furthermore, these varying relationships showed lower correlation coefficients as NDVI values ranged between 0.08 and 0.35.
Normalized Difference Vegetation Index Mapping Using the GreenSeeker Application
The GreenSeeker can be used for NDVI mapping. Contrary to cereals for which soil is covered by a continuous vegetation layer, grapevines are planted in definite rows, making it possible to separate measurements corresponding to each row. The NDVI measurements were performed employing a straddler-tractor. For each row, measurement positions were calculated based on the hypothesis of a steady speed of the tractor, which was verified within a range of ±0.2 km h–1 while advancing at 3.6 km h–1, except for a few variations reaching ±0.5 km h–1.
Figure 5
provides an example of NDVI mapping using the GreenSeeker (plot 2) when mounted at different heights (to scan L, M, and H strips as shown in Fig. 1) on a straddler-tractor (28 July 2004). The NDVI mean measurements range between 0.67 and 0.86 for L map and between 0.19 and 0.42 for H map. In the first case (L map), in most areas of the plot, NDVI signal is saturated and the map can only help detect poor vigor areas (white encircled). In the second case (H map), most areas of the plot present low mean NDVI measurement values, with interference (Fig. 3B), and the map can only help detect greater vigor areas (black encircled). To summarize, the most relevant map is obtained when NDVI values range in the linear portion of the curve representing NDVI as a function of VLAI or gap fraction. Another map (M map) was obtained by scanning the area located between canopy topping height and 61 cm under (as shown in Fig. 1). Figure 5 shows low and high vigor areas, as well as medium vigor areas.
View larger version (72K):
[in this window]
[in a new window]
|
Fig. 5. Three normalized difference vegetation index (NDVI) maps (2004) of plot 2 obtained by Ordinary Kriging. The name of each map (L, M, and H) refers to the strip scanned with the GreenSeeker as indicated in Fig. 1. The NDVI measurements were performed every row (North-South oriented). Lower vigor areas (white encircled, L map) and high vigor (black encircled, M map) are both found when measuring the medium part of the canopy (M map), as representative of the whole.
|
|
Spots of high and low vigor determined by NDVI measurements visually match with those of pruning weight (PW) (Fig. 6
). Moreover, comparison between 2004 and 2005 vintages suggests a rather steadiness of plot variability. However, when it comes to crossing data considering NDVI values per meter of row with PW of 96 experimental plots distributed throughout plot 2, low correlation was found (NDVI = 0.0002 PW + 0.399, r2 = 0.362, P < 0.01). Tractor speed variations appeared to have an impact on data positioning imprecision. In addition, entangled vine shoots from contiguous vines which were not taken into account for PW determination interfered with NDVI measurements. Moreover, NDVI measurements were performed in the M strip (Fig. 1) whereas PWs represent the whole plant. Working with a lower resolution and using a Differential GPS (DGPS) would increase precision in data positioning, and also facilitate data processing. Nevertheless, these results show that since the sensor does not encompass the entire canopy, and although its measurements are reliable, one must be careful when interpreting such maps.
View larger version (112K):
[in this window]
[in a new window]
|
Fig. 6. Plot 2 normalized difference vegetation index (NDVI) and pruning weight maps obtained by ordinary Kriging. Pruning weight map is based on mean pruning weight of 96 experimental plots, whereas NDVI maps are based on continuous measurements in each row.
|
|
The GreenSeeker was used in 2005 to map vineyard 2 NDVI. This vineyard appeared to be very heterogeneous, and despite the relative small size of the plots, a few showed areas of variability, large enough to be of interest for reconsidering vineyard management. Plot 3, located at the North-East of vineyard 2, was segmented into two blocks (Eastern and Western), according to the differences of intraplot vigor (i.e., NDVI values, Fig. 7
) observed to carry out separate berry maturing analysis. Table 1
shows the gap of ripeness between the two blocks. The western block, with a lower vigor, was more qualitative, with a higher "sugars-to-total acidity" ratio, a higher content in anthocyanins, and smaller berries (i.e., a higher "surface-to-volume" ratio of the berries). Surprisingly this same block presented a higher assimilable N content. This example shows the relevance of NDVI mapping as a tool for redefining block segmentation within a vineyard, with an optimization of the quantity of premium wine that can be produced.
View larger version (92K):
[in this window]
[in a new window]
|
Fig. 7. Plot 3 normalized difference vegetation index (NDVI) map (June 2005). This plot showed two areas of distinct vigor, and was thus segmented in two (dotted line) for maturing controls (results in Table 1). The vertical line segments represent the vine rows.
|
|
View this table:
[in this window]
[in a new window]
|
Table 1. Plot 3 maturing controls results (31 Aug. 2005), with different grape characteristics: in this example the area with lower vigor has a higher potential for quality red wine.
|
|
Compiling data using the GreenSeeker is rather time consuming in comparison to remote imagers (it took approximately 5 d for V 2, given its planting density). However, it can be of great advantage because it can be done by the viticulturist on his own and for specific plots. In addition, time spent compiling data could possibly be coupled with a vineyard operation (such as hedging or plowing).
To summarize, NDVI measured on the growing zone of a canopy informs about its vigor (gap fraction, VLAI). This vigor can be related to environmental traits and local variation of the individual vines' growth capacity. The NDVI reflects a sum of assessments: radiative, mineral, and hydric. It can also help detect pest infestation (Johnson et al., 1996). Complex models are needed to explain vine vigor variation to point out which parameters play a significant role in each given situation of soil and climate. Moreover, cultural practices also have to be considered. In this way, NDVI helps growers gain better knowledge of their vineyard, giving them the option of either smoothing variability (by local change in soil organic matter, irrigation, trimming, etc.) or promoting it by adapting cultural practices to each given terroir conditions with segmented block harvesting (Bramley and Hamilton, 2004; Bramley, 2005). Nevertheless, it is relevant to study the relationship NDVI variability to plot berry quality indicators such as sugar content, phenolics, and total acidity. The question that remains is how vegetative growth variability contributes to the lower qualitative value of the harvest.
CONCLUSIONS
In this study, NDVI ground-based measurements using the GreenSeeker proved to be satisfactory to estimate vinegrape canopy gap. The relation adopts a sigmoid trend, allowing accurate gap fraction characterization within a range of values comprised between 20 and 80%. As an indirect consequence of foliar distribution within the canopy, that is, the relation between gap fraction and VLAI, NDVI allows VLAI estimation for VLAI values under 4. Also, when performing NDVI measurements using the GreenSeeker it is impossible to take in the entire canopy. Hence the choice of canopy area to be scanned is determinant to obtain an optimal characterization of vineyard vigor variability, when signal is neither insensitive nor saturated. The NDVI mapping provides relevant information for vine growers, however choice of lag for averaged NDVI calls for caution. For example, a medium NDVI value can indicate a medium vigor evenly distributed throughout the lag, or can encompass very high vigor areas mixed to canopy holes, linked to vine training. No matter how, resolution must be greater than intervine spacing. To conclude, ground-based NDVI measurements using the GreenSeeker are a new tool for precision viticulture research and better vineyard management, accessible to all grapevine growers.
|
NOTES
|
|---|
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.
|
REFERENCES
|
|---|
- Baldy, R., J. Benedicts, L. Johnson, E. Weber, M. Baldy, B. Osborn, and J. Burleigh. 1996. Leaf color and vine size are related to yield in a phylloxera infested vineyard. Vitis 35:201–205.[Web of Science]
- Bramley, R.G.V. 2005. Understanding variability in winegrape production systems. 2. Within vineyard variation in quality over several vintages. Aust. J. Grape Wine R. 11:33–42.
- Bramley, R.G.V., and R.P. Hamilton. 2004. Understanding variability in winegrape production systems. 1. Within vineyard variation in yield over several vintages. Aust. J. Grape Wine Res. 10:32–45.
- Carbonneau, A. 1976. Principes et méthodes de mesure de la surface foliaire. Essai de caractérisation des types de feuilles dans le genre Vitis. (In French.) Ann. Amélior. Plant. 26:327–343.
- Castelan-Estrada, M., P. Vivin, and J.P. Gaudillère. 2002. Allometric relationships to estimate seasonal above-ground vegetative and reproductive biomass of Vitis vinifera L. Ann. Bot. (London) 89:401–408.[Abstract/Free Full Text]
- Dobrowski, S.Z., S.L. Ustin, and J.A. Wolpert. 2002. Remote estimation of vine canopy density in vertically shoot-positioned vineyards: Determining optimal vegetation indices. Aust. J. Grape Wine Res. 8:117–125.
- Dokoozlian, N.K., and W.M. Kliewer. 1996. Influence of light on grape berry growth and composition varies during fruit development. J. Am. Soc. Hortic. Sci. 121:869–874.
- Grantz, D.A., and L.E. Williams. 1993. An empirical protocol for indirect measurement of leaf area index in grape (Vitis vinifera L.). HortScience 28:777–779.[Abstract/Free Full Text]
- Johnson, L., B. Lobitz, R. Armstrong, R. Baldy, E. Weber, J. De Benedictis, and D. Bosch. 1996. Airborne imaging for vineyard canopy evaluation. Calif. Agric. 50:1–12.
- Johnson, L.F., D.E. Roczen, S.K. Youkhana, R.R. Nemani, and D.F. Bosch. 2003. Mapping vineyard leaf area with multispectral satellite imagery. Comput. Electron. Agric. 38:37–48.
- Jones, C. 2004. GreenSeeker (Red NDVI) performance review testing summary. Available at http://nue.okstate.edu/Hand_Held/Greenseeker_Performance_Review_09_03_04external.doc (accessed 19 Oct. 2004, verified 23 Nov. 2008). Oklahoma State Univ., Stillwater.
- Mabrouk, H., and H. Sinoquet. 1998. Indices of light microclimate and canopy structure of grapevines determined by 3D digitising and image analysis, and their relationship to grape quality. Aust. J. Grape Wine Res. 4:2–13.
- Mullen, R.W., K.W. Freeman, W.R. Raun, G.V. Johnson, M.L. Stone, and J.B. Solie. 2003. Identifying an in-season response index and the potential to increase wheat yield with nitrogen. Agron. J. 95:347–351.[Abstract/Free Full Text]
- Murisier, F., and V. Zufferey. 1997. Rapport feuille-fruit de la vigne et qualité du raisin. (In French.) Rev. Suisse Vitic. Arboric. Hortic. 29:355–362.
- Ollat, N., M. Fermaud, J.P. Tandonnet, and M. Neveux. 1998. Evaluation of an indirect method for leaf area index determination in the vineyard: Combined effects of cultivar, year and training system. Vitis 2:73–78.
- Riou, C., P. Pieri, and B. Le Clech. 1994. Consommation d'eau de la vigne en conditions hydriques non limitantes. Formulation simplifiée de la transpiration. (In French, with English abstract.). Vitis 33:109–115.
- Rouse, J.W., R.H. Haas, J.A. Schell, D.W. Deering, and J.C. Harlan. 1974. Monitoring the vernal advancement and retrogradation (greenwave effect) of natural vegetation. Type III, Final Rep. NASA/Goddard Space Flight Ctr., Greenbelt, MD.
- Schultz, H.R. 1995. Grape canopy structure, light microclimate and photosynthesis. I. A two-dimensional model of the spatial distribution of surface area densities and leaf ages in two canopy systems. Vitis 34:211–215.[Web of Science]
- Smart, R.E. 1985. Principles of grapevine canopy microclimate manipulation with implications for yield and quality. A review. Am. J. Enol. Vitic. 36:230–239.
- Solie, J.B., M.L. Stone, W.R. Raun, G.V. Johnson, K. Freeman, R. Mullen, D.E. Needham, S. Reed, and C.N. Washmon. 2002. Real-time sensing and N fertilization with a field scale greenseekerTM applicator. In P.C. Robert (ed.) Proc. Int. Conf. Precision Agriculture and Other Precision Resources Management, 6th, Minneapolis, MN. 14–17 July 2002. Available at http://nue.okstate.edu/Papers/Minnesota_2002_Solie.htm (verified 23 Nov. 2008). ASA, Madison, WI.
- Sommer, K.J., and A.R.G. Lang. 1994. Comparative analysis of two indirect methods of measuring leaf area index as applied to minimal and spur pruned grape vines. Aust. J. Plant Physiol. 21:197–206.[Web of Science]
- Tregoat, O., N. Ollat, G. Grenier, and C.V. Leeuwen. 2001. Etude comparative de la précision et de la rapidité de mise en oeuvre de différentes méthodes d'estimation de la surface foliaire de la vigne. (In French.) J. Int. Sci. Vigne Vin. 35:31–39.
TOP
ABSTRACT
INTRODUCTION
