**5.1.2 LI measurement with indirect methods**

In indirect methods, different apparatus for estimating the different components of radiation, such as direct radiation, diffuse, land, atmospheric... are used which consider the net radiation balance in order to know how much available radiation reaching the surface.

The inherent difficulties in measuring PAR throughout a canopy and advances in radiometric techniques have led to the development of methods for remotely sensing radiation capture. Radiometric methods rely on differences in the spectral reflectance of vegetation and soil. Vegetative indices based on reflectance in broad wavebands have provided good estimates of radiation capture and yield in crop plants (Gallo et al., 1985; Hatfield et al., 1984). Vegetation indices have also provided good estimates of fractional groundcover (Boissard et al., 1992; White et al., 2000). More recently, spectroradiometers capable of measuring narrow band radiation have been used to monitor plant stress (Elvidge & Chen, 1995). Radiometric satellite data are now available for the evaluation of large areas, and small portable radiometers are becoming less expensive as the technology progresses. In this respect, good results have been obtained with measurements using digital photographic images to determine crop cover and radiation interception in soybean (Purcell, 2000) and lettuce (Klassen et al., 2003), crop cover in turfgrass (Richardson et al., 2001), and canopy and soil cover with straw mulch (Bennet et al., 2000; Beverly, 1996; Olmstead et al., 2004). Other important points are that the area of soil exposed to the sun can be differentiated from that covered by leaves while the angle of the camera is close to that of the sun (Purcell, 2000). With regards to differentiating between the green parts of the crop and the soil surface, results could vary in the case of soils of different colors as a result of their different behavior with respect to the reflection and absorption of radiation; this is particularly the case for different kinds of mulches. In this case, the validity of the method will largely depend on the capacity of the software to discriminate between parts of the crop's green canopy. In the presence of weeds or green cover, it may be necessary to prescreen images.

Digital images offer a series of additional advantages over other methods for estimating LI, assuming that the soil background can be distinguished from leaves, light transmission of leaves is small relative to light absorption, and that the angle of the camera to the horizon approximates the solar angle (Purcell, 2000) such as the direct treatment of images by computers. Moreover, a graphic record of the crop is generated in the case of studies of canopy evolution. This can be used for phonological monitoring (Shelton et al., 1988) to determine differences in color and fertility in maize (Ewing and Horton, 1999) and to study the incidence of pests and diseases.

Automated methods of digital image analysis are indirect methods of LI measurement. Initially they were not widely used because they generally require complex and expensive instrumentation, as well as making mistakes with the changing colors of soil and plant (Hayes & Han, 1993; Van Henten & Bontsema, 1995; Beverly 1996). However, no alteration of vegetation cover and the automation of image analysis has allowed the elimination of many subjective decisions of the observer.

Recent advances in high-resolution digital cameras and associated image manipulation software provide enhanced methods of visual discrimination and computer thresholding that are user-friendly and inexpensive. Three recent studies have demonstrated the accuracy of digital imaging analysis for monitoring plant growth. Paruelo et al. (2000) described a method for estimating aboveground biomass in semiarid grasslands using digitized photographs and a DOS-based program they developed. Purcell (2000) described a method for measuring canopy coverage and light interception in soybean fields using a digital camera and standard imaging software. Richardson et al. (2001) described a digital method for quantifying turfgrass cover following a modified version of Purcell (2000). Klassen 2002 used standard methods of measurement of radiation for comparison with the analysis of vegetation cover as with digital photography using the analysis software Adobe Photoshop 6.0 image. Olmstead et al. (2004) analyzed vegetation cover in grapevine crop through the analysis of digital images, using Sigma Scan Pro 5.0 compared with estimated visualization measures. Other authors used the measurement by digital photography analysis for other uses, Adamsen et al. (1999) to measure maize senescence.

A seemingly key advantage of using digital cameras is that they allow for continuous monitoring of vegetation (White et al., 2000), in the case of low-lying horticultural crops. These measures do not alter the disposition of the crop. Replacing standard procedures, such as the width of cultivation, direct quantification of the shadows or linear PAR sensors, are subjective and costly, and often inaccurate (Campillo et al., 2008).

Taking advantage of the latest developments in digital technology, it is now possible to measure the evolution of vegetation cover through digital photography and to determine the PGC using image interpretation techniques (Campillo et al., 2008; Rodríguez et al., 2000).

Fig. 9a. Digital images of processing tomato measure with a area method.

Campillo et al (2008), compared LI methodology (PAR) with various methods of PGC measurement. They used three methodologies to measure PGC in two low-lying crops, a winter crop (cauliflower) and a summer crop (processing tomato) in two consecutive years (2005 and 2006) and (2005) in cauliflower crop.

Area method (SA): In this method, crop row width was estimated by simulation based on measurements taken at three points within the marked area using a metric strip. The data were then used to estimate average row width and the PGC (Adams & Arkin, 1977; Giménez, 1985). Both row and frame width were determined in pixels using the measuring tool (IMAGE J 1.33). The sampling area was delimited by the width (X) and length (Y) of the reference frame (Fig. 9a) and the three measurements of row width were: x1, x2, x3. PGC was calculated using the expression:

of digital imaging analysis for monitoring plant growth. Paruelo et al. (2000) described a method for estimating aboveground biomass in semiarid grasslands using digitized photographs and a DOS-based program they developed. Purcell (2000) described a method for measuring canopy coverage and light interception in soybean fields using a digital camera and standard imaging software. Richardson et al. (2001) described a digital method for quantifying turfgrass cover following a modified version of Purcell (2000). Klassen 2002 used standard methods of measurement of radiation for comparison with the analysis of vegetation cover as with digital photography using the analysis software Adobe Photoshop 6.0 image. Olmstead et al. (2004) analyzed vegetation cover in grapevine crop through the analysis of digital images, using Sigma Scan Pro 5.0 compared with estimated visualization measures. Other authors used the measurement by digital photography analysis for other

A seemingly key advantage of using digital cameras is that they allow for continuous monitoring of vegetation (White et al., 2000), in the case of low-lying horticultural crops. These measures do not alter the disposition of the crop. Replacing standard procedures, such as the width of cultivation, direct quantification of the shadows or linear PAR sensors,

Taking advantage of the latest developments in digital technology, it is now possible to measure the evolution of vegetation cover through digital photography and to determine the PGC using image interpretation techniques (Campillo et al., 2008; Rodríguez et al., 2000).

Campillo et al (2008), compared LI methodology (PAR) with various methods of PGC measurement. They used three methodologies to measure PGC in two low-lying crops, a winter crop (cauliflower) and a summer crop (processing tomato) in two consecutive years

Area method (SA): In this method, crop row width was estimated by simulation based on measurements taken at three points within the marked area using a metric strip. The data were then used to estimate average row width and the PGC (Adams & Arkin, 1977; Giménez, 1985). Both row and frame width were determined in pixels using the measuring tool (IMAGE J 1.33). The sampling area was delimited by the width (X) and length (Y) of the reference frame (Fig. 9a) and the three measurements of row width were: x1, x2, x3. PGC

uses, Adamsen et al. (1999) to measure maize senescence.

are subjective and costly, and often inaccurate (Campillo et al., 2008).

Fig. 9a. Digital images of processing tomato measure with a area method.

(2005 and 2006) and (2005) in cauliflower crop.

was calculated using the expression:

$$PGC = \left(\frac{\left(\frac{\left(\mathbf{x}\_1 + \mathbf{x}\_2 + \mathbf{x}\_3\right)}{3}\right) \* Y}{\left(\mathbf{X} \* Y\right)}\right) \* 100\tag{2}$$

Fig. 9b. Digital images of processing tomato measure with a contour method.

Contour method (SC): In this method, the technique of drawing the crop's shade contour on paper and the subsequent measurement of the area in question is simulated (Kvet & Marshall, 1971). Figure 9b shows the processing of the digital image. To measure the area, the crop's contour was previously delimited using the IMAGE J 1.33 program. Areas with no vegetation cover that were within the canopy were measured and omitted from the surface area count. The crop surface area (S) was measured in pixels using the same program. This area was then related to the sampling area to estimate the PGC according to the following expression:

$$PGC = \left(\frac{S^2}{\left(X\*Y\right)} \* 100\right) \* 100\tag{3}$$

Fig. 9c. Digital images of processing tomato measure with a reclassification method.

Reclassification method (SR). With this method (Fig. 9c), the crop area (S) is determined by classifying the image according to the range of radiation levels shown on an RGB image of the crop (0 to 255 colors); this was done using a RGB max reclassification tool (GIMP 2.2). After the classification process, it is possible to measure the surface area occupied by green parts (crop) and to differentiate them from the soil or plastic. In contrast to the other two methods, here the crop must be subjected to homogeneous lighting conditions, because the presence of shadows may reduce a crop's color and impede subsequent color reclassification. PGC was calculated according to formula [3].

PGC measurements were compared with measurements made with a LI PAR bar.

Intercepted radiation: LI measurements were made using a 100-cm linear PAR sensor (LICOR Li-190; LI-COR, Lincoln, NE). They were made at solar noon, perpendicular to the crop row, in the same area in which the photographs had been taken. Samples taken from below the crop were compared with reference measurements taken above the crop row (ref). Percentages of LI were calculated by applying Eq. [4], in which it was necessary to know the percentage of radiation that was not intercepted by the crop (RP) as a quotient of the PAR measurements taken both above and below the canopy. According to the degree of plant development two situations for measurement of RP were proposed:


The sensor was covered with a material that blocks light and average measurements were taken in the center of the row (r7, r8), also using a half-length PAR bar (50 cm). This was done in a way that included the total width of culture (150 cm). In this situation, ref was measured using a half-length PAR bar (50 cm).

$$LI = \left(1 - RP\right) \* 100\tag{4}$$

$$RP = \left(\frac{\left(mean\left(r\_1; r\_2; r\_3; r\_4; r\_5\right) + 0.5\*mean(r\_{ef})\right)}{\left(mean\left(r\_{ef}\right)\*1.5\right)}\right) \tag{5}$$

$$RP = \left(\frac{\left(mean\left(r\_1; r\_2; r\_3\right) + \left(mean\left(r\_4; r\_5; r\_6\right) + mean\left(r\_7; r\_8\right)\right)\right)}{\left(mean\left(r\_{cf}\right)\*3\right)}\right) \tag{6}$$

Reclassification method (SR). With this method (Fig. 9c), the crop area (S) is determined by classifying the image according to the range of radiation levels shown on an RGB image of the crop (0 to 255 colors); this was done using a RGB max reclassification tool (GIMP 2.2). After the classification process, it is possible to measure the surface area occupied by green parts (crop) and to differentiate them from the soil or plastic. In contrast to the other two methods, here the crop must be subjected to homogeneous lighting conditions, because the presence of shadows may reduce a crop's color and impede subsequent color

Intercepted radiation: LI measurements were made using a 100-cm linear PAR sensor (LICOR Li-190; LI-COR, Lincoln, NE). They were made at solar noon, perpendicular to the crop row, in the same area in which the photographs had been taken. Samples taken from below the crop were compared with reference measurements taken above the crop row (ref). Percentages of LI were calculated by applying Eq. [4], in which it was necessary to know the percentage of radiation that was not intercepted by the crop (RP) as a quotient of the PAR measurements taken both above and below the canopy. According to the degree of plant

1. When the crop row width was less than 100 cm , RP was calculated by applying Eq. [5] as the average of five measurements taken under the crop (r1, r2, r3 , r4, r5). Measurements were taken every 20 cm using the total length of the PAR bar (100 cm) and adding 50 cm to the reference measurement to include the total width of crop (150

The sensor was covered with a material that blocks light and average measurements were taken in the center of the row (r7, r8), also using a half-length PAR bar (50 cm). This was done in a way that included the total width of culture (150 cm). In this situation, ref was

> 

*ef*

 

 *ef*

*mean r r r mean r r r mean r r*

<sup>123</sup> 456 78 ; ; ;; (;)

3

 

*mean r r r r r mean r*

<sup>12345</sup> ; ; ; ; 0.5 ( ) 1.5

*LI RP* 1 100 (4)

*ef*

*mean r* (5)

*mean r* (6)

cm). In this situation, ref was measured using the total length of the PAR bar. 2. When the crop row width was greater than 100 cm the maximum length of the PAR sensor, RP was calculated applying Eq. [6] as the average of three measurements taken beneath the crop on each side of the crop row (r1, r2, r3 left side and r4, r5, r6 right side). Measurements were taken at 20-cm intervals using a half-length PAR bar

PGC measurements were compared with measurements made with a LI PAR bar.

reclassification. PGC was calculated according to formula [3].

development two situations for measurement of RP were proposed:

(50 cm).

measured using a half-length PAR bar (50 cm).

*RP*

*RP*

Fig. 10. Relationship between canopy percent light interception (LI) and percentage of groundcover (PGC) determined by the different methods of analysis: area (A,D), contour (B,E), and reclassification (C,F) for the cauliflower in 2005 (A,B,C) and tomato crop in 2005 and 2006 (D,E,F). Values with different letters differ (P < 0.05) between years. From (Campillo et al., 2008)

Figure 10 show that there was a close relationship between the fraction of light intercepted by the canopy at solar noon and estimated PGC for all three methodologies in both years and crops. In all cases, there was a linear adjustment with a significant correlation coefficient (P < 0.01) and an r2 greater than 0.87. This indicates that any of the described methods would have been valid for estimating the amount of radiation intercepted by the crop.

However, the adjustment was different according to the method used. The adjustment with LI was narrower when using the SR method to estimate PGC (r2 = 0.92 and 0.96) followed by SC (r2 = 0.91 and 0.95) and SA (r2 = 0.87 and 0.94) for 2005 and 2006, respectively in processing tomato and (0.97, 0.96 and 0.89). The relationship between LI and PGC was somewhat stronger when the SR and SC methods were used, whereas the SA method produced greater errors in estimation.

Finally, the most accurate estimate of PGC for both crops was obtained with SR, because color discrimination made it easier to differentiate between vegetation and soil (Fig. 10c). Although the results obtained with SR and SC were similar, it should be borne in mind that the SR method was cheaper to apply because all image processing was performed by software without the need for human definition of the area to be measured. The use of processed digital images with the SR method supposed a considerable improvement with respect to the other two methods and also in the gathering of data directly from the crop. This method was economical and easy to apply. It also eliminated the subjectivity associated with operators having to define areas or points of measurement. Slight but significant differences were found between years applying the same methodology with the adjustment being better in 2006.

#### **5.2 Leaf area index measurement**

Determination of LAI is often the most expensive in a field study, because direct measurement (destructive methods) is time-consuming. We can classify in the same way as with LI, in direct methods (which can be destructive or non destructive) and indirect, based on properties of vegetation cover, being non-destructive.

#### **5.2.1 LAI measurement with direct methods**

Direct methods for determining leaf area have so far been restricted to the use of an automatic area-integrating meter. Tracing, shadow graphing, and the use of a planimeter to measure the total leaf area attached to shoots are all time-consuming and are tedious approaches; furthermore, in some experiments, there is not enough time to make such measurements (Manivel & Weaver, 1974). All direct methods are similar in that they are difficult, extremely labor-intensive, require many replicates to account for spatial variability in the canopy, and are therefore costly in terms of time and money and also destructive.

#### **5.2.2 LAI measurement with indirect methods**

Many indirect methods for measuring LAI have been developed.

Methods based on empirical relationships between leaf area and easily obtainable parameters such as the size of the leaves are available. In any case, the empirical relationship should always be check with direct action as they may vary during the crop cycle and some other varieties. Some used are S = A \* L \* I and S = A \* LB, where S is the area, L the length and I the maximum width plant element, A and B are empirical elements. Also can estimate the leaf area through relationships with the weight. A first group of methods is based on the S = M / where M is the leaf weight in grams and is the specific weight (g/m2) (Patón et al., 1998). Techniques based on gap-fraction analysis assume that leaf area can be calculated from the canopy transmittance (the fraction of direct solar radiation which penetrates the canopy) (Ford, 1997). Optical methods are indirect, non-contact, and are commonly implemented. They are based on the measurement of light transmission through canopies (Jonckheere et al., 2004). These methods apply the Beer-Lambert law, taking into account the fact that the total amount of radiation intercepted by a canopy layer depends on the incident irradiance, the canopy structure, and its optical properties (Breda, 2003). Monsi & Saeki (1953) expanded the Beer-Lambert extinction law to apply it to plant canopies. The Beer-Lambert law expresses the attenuation of radiation in a homogenous turbid medium. In such a medium, the flux is absorbed in proportion to the optical distance. The LAI is related to the incident solar radiation intercepted by the crop (LI) and extinction coefficient (K), which describes the angle of the blades in relation to the sun, through the formula proposed by Monsi and Saeki (1953):

$$LI = \mathbf{1} - e^{\left\{-\left(\mathbf{K} \star \mathbf{L} \, \mathbf{M}\right)\right\}} \tag{7}$$

This approach could also be used to estimate LAI using Eq. [7]; however, we would need to know the extinction coefficient for each crop and variety (Campbell, 1986). Several

Determination of LAI is often the most expensive in a field study, because direct measurement (destructive methods) is time-consuming. We can classify in the same way as with LI, in direct methods (which can be destructive or non destructive) and indirect, based

Direct methods for determining leaf area have so far been restricted to the use of an automatic area-integrating meter. Tracing, shadow graphing, and the use of a planimeter to measure the total leaf area attached to shoots are all time-consuming and are tedious approaches; furthermore, in some experiments, there is not enough time to make such measurements (Manivel & Weaver, 1974). All direct methods are similar in that they are difficult, extremely labor-intensive, require many replicates to account for spatial variability in the canopy, and are therefore costly in terms of time and money and also

Methods based on empirical relationships between leaf area and easily obtainable parameters such as the size of the leaves are available. In any case, the empirical relationship should always be check with direct action as they may vary during the crop cycle and some other varieties. Some used are S = A \* L \* I and S = A \* LB, where S is the area, L the length and I the maximum width plant element, A and B are empirical elements. Also can estimate the leaf area through relationships with the weight. A first group of methods is based on the S = M / where M is the leaf weight in grams and is the specific weight (g/m2) (Patón et al., 1998). Techniques based on gap-fraction analysis assume that leaf area can be calculated from the canopy transmittance (the fraction of direct solar radiation which penetrates the canopy) (Ford, 1997). Optical methods are indirect, non-contact, and are commonly implemented. They are based on the measurement of light transmission through canopies (Jonckheere et al., 2004). These methods apply the Beer-Lambert law, taking into account the fact that the total amount of radiation intercepted by a canopy layer depends on the incident irradiance, the canopy structure, and its optical properties (Breda, 2003). Monsi & Saeki (1953) expanded the Beer-Lambert extinction law to apply it to plant canopies. The Beer-Lambert law expresses the attenuation of radiation in a homogenous turbid medium. In such a medium, the flux is absorbed in proportion to the optical distance. The LAI is related to the incident solar radiation intercepted by the crop (LI) and extinction coefficient (K), which describes the angle of the blades in relation to the sun, through the formula proposed

<sup>1</sup> *K LAI LI e* (7)

This approach could also be used to estimate LAI using Eq. [7]; however, we would need to know the extinction coefficient for each crop and variety (Campbell, 1986). Several

**5.2 Leaf area index measurement** 

destructive.

by Monsi and Saeki (1953):

on properties of vegetation cover, being non-destructive.

**5.2.1 LAI measurement with direct methods** 

**5.2.2 LAI measurement with indirect methods** 

Many indirect methods for measuring LAI have been developed.

authors have discussed how to determine k (Hassika et al., 1997; Ledent, 1977; Smith, 1993; Vose et al., 1995) and the accuracy of methodology to be applied (Nel & Wessman, 1993). It is also important to consider that the extinction coefficient also depends on stand structure and canopy architecture (Smith et al., 1991; Turton, 1985) and that the canopy extinction coefficient is a function of wavelength (Jones, 1992), radiation type, and direction (Berbigier & Bonnefond, 1995). It is also important to maximize spatial integration by using large, linear and/or mobile sensors. Extinction coefficient, which varies with species, season and environmental conditions (Hay & Walter, 1989), take values in terms of leaf angles: spherical (0.5-0.7), conical (1), vertical or erectofila (0.3-0.7). The distributions of leaf angles have agronomic and ecological implications. Horizontal distribution implies a high k, allowing for increased intercepted radiation by small plants. the disadvantage is that when the LAI is high the light distribution is very unequal, the lower leaves receive little light, which tends to accelerate senescence. In the opposite, erectofila distribution can be advantageous to intercept radiation when the zenith angle is large (winter, high latitudes) and represents a more homogeneous distribution of radiation when the LAI is high.

This method involves ground-based measurements of total, direct, and/or diffuse radiation transmittance to the forest floor and it makes use of line quantum sensors or radiometers (Pierce and Running, 1988), laser point quadrats (Wilson, 1963), and capacitance sensors (Vickery et al., 1980). These instruments have already proven their value in estimations of LAI for coniferous (Marshall and Waring, 1986; Pierce and Running, 1988) as well as broad-leafed (Chason et al., 1991) stands. In comparison with allometric methods, the approach provides more accurate LAI estimates (Smith et al., 1991). However, the light measurements required to calculate LAI require cloudless skies, and there is generally a need to incorporate a light extinction coefficient that is both siteand species-specific as a result of leaf angle, leaf form, and leaf clumping, etc. (Vose et al., 1995). Measurements can be taken either by locating the sensors perpendicular to the crop rows (Egli, 1994) or by taking multiple measurements parallel to them (Board et al., 1992). This determination can, however, be costly; it depends on the number of measurements needed to characterize the study area, especially in low-lying crops, where vegetation must be moved to place sensors under it, which implies introducing alterations during data collection. There are several commercial systems available to measure indirectly the structure of vegetation and LAI, based on the Beer-Lambert law, including analyzer plant canopy (plant canopy analyzer LiCor LAI-2000) (Li-Cor, 1989); (Cintra et al., 2001; Malone, 2002). El LiCor LAI-2000 has an optical sensor and a control box easily manipulated by an operator. The LAI is estimated according to a model developed by Miller (1967), based on gap-fraction analysis (Barclay et al., 2000). Similar instruments is the CI-100 (Digital plant canopy imager). It consists of a digital camera with a lens of "fish eye" with a 180 degrees field of view.

The analysis of remote estimation methods, provides a temporal and spatial information. The new technologies, provide LAI data from digital cameras (Adamsen et al., 1999), video images (Beverly, 1996), multispectral digital sensors (Bellairs et al., 1996; Shanahan et al., 2001), aerial imagery (Blackmer et al., 1996; Flowers et al., 2001) and satellite images (Wiegand et al., 1979; Thenkabail et al., 1992; Green et al., 1997). One of the remote methods most used is Normalized Difference Vegetation Index (NDVI). These spectral reflectances are themselves ratios of the reflected to the incoming radiation in each spectral band individually, hence they take on values between 0.0 and 1.0. By design, the NDVI itself thus varies between -1.0 and +1.0. It should be noted that NDVI is functionally, but not linearly, equivalent to the simple infrared/red ratio (NIR/VIS). The advantage of NDVI over a simple infrared/red ratio is therefore generally limited to any possible linearity of its functional relationship with vegetation properties (e.g. biomass). This method is sensitive to background soil and weather conditions (Gilabert et al., 1997). There are different satellite sources where one can get the values of NDVI with different resolutions; AVHRR (Advanced Very High Resolution Radiometer), MODIS (Moderate Resolution Imaging Spectroradiometer), SPOT. Vegetation indices are widely used for the calculation of biomass and LAI (Blazquez et al., 1981; Serrano et al., 2000; Wanjura and Hatfield, 1986) and D'Urso, et al (2010) (figure 11)

Fig. 11. Correlation between measured LAI using LAI-2000 instrument and estimated LAI from NDVI; RMSE= 0.71, Italian case study (13% of ground measured data used for calibration). From D'Urso, et al 2010.

Campillo et al 2010, compared in two low-lying crops, a winter crop (cauliflower) and a summer crop (processing tomato) in two consecutive years (2005 and 2006) measurements of PGC (non-destructive method) and LAI (destructive method). The objective was relations of two parameters and the possibility of using a PGC methodology as a LAI measurement in vegetable crops.

are themselves ratios of the reflected to the incoming radiation in each spectral band individually, hence they take on values between 0.0 and 1.0. By design, the NDVI itself thus varies between -1.0 and +1.0. It should be noted that NDVI is functionally, but not linearly, equivalent to the simple infrared/red ratio (NIR/VIS). The advantage of NDVI over a simple infrared/red ratio is therefore generally limited to any possible linearity of its functional relationship with vegetation properties (e.g. biomass). This method is sensitive to background soil and weather conditions (Gilabert et al., 1997). There are different satellite sources where one can get the values of NDVI with different resolutions; AVHRR (Advanced Very High Resolution Radiometer), MODIS (Moderate Resolution Imaging Spectroradiometer), SPOT. Vegetation indices are widely used for the calculation of biomass and LAI (Blazquez et al., 1981; Serrano et al., 2000; Wanjura and Hatfield, 1986) and D'Urso,

Fig. 11. Correlation between measured LAI using LAI-2000 instrument and estimated LAI from NDVI; RMSE= 0.71, Italian case study (13% of ground measured data used for

Campillo et al 2010, compared in two low-lying crops, a winter crop (cauliflower) and a summer crop (processing tomato) in two consecutive years (2005 and 2006) measurements of PGC (non-destructive method) and LAI (destructive method). The objective was relations of two parameters and the possibility of using a PGC methodology as a LAI measurement in

et al (2010) (figure 11)

calibration). From D'Urso, et al 2010.

vegetable crops.

Fig. 12. Relationship between leaf area index (LAI) estimated through the destructive sampling of biomass and the percentage of shaded ground measured (PGC) by the reclassification method for cauliflower and processing tomato crops in 2005 and 2006.

A polynomial relationship (r2 > 0.88) was observed between the two variables in both crops. PGC increased with leaf area development in a curve-linear pattern composed of an initial linear phase followed by a saturating phase, which approached a maximum asymptotic value at full groundcover.

In cauliflower, significant differences were observed between the curves obtained for each year (r2 = 0.89 and 0.95 for the first and second years, respectively, Figure 12). This difference was the result of a significant change in the prevailing weather conditions during the crop cycle that affected the morphology of the leaves. Temperatures in the first year were lower than in the second and frequent frosts caused the curling of leaf margins, resulting in a lower PGC for the same LAI. The PGC–LAI curve adjustments for the tomato crop were significant (Figure 12). The curves coincided for both years, although with differences in the adjustment (0.89 and 0.93 for 2005 and 2006, respectively). In this case, a single curve would have enabled us to estimate the LAI by nondestructive methods using digital images. Although, in principle, the factors that can modify the arrangement of leaves could alter this relationship, this trial included treatments with different water statuses that could have induced changes in plant leaf angle, but this aspect did not affect the goodness of fit. It still remains to be seen how this equation would be influenced by morphological (plant height and leaf type) differences between varieties. The PGC of a crop depends on the leaf area development and on the distribution of the plant leaves on the space (plant architecture). PGC is therefore the dependent variable in the relationship between LAI and PGC. The equations obtained for these two crops are highly significant with a narrow adjustment; they therefore provide a method for estimating LAI based on known PGC values.

From the data obtained when comparing LAI values with those of PGC, and from that obtained by Campillo et al. (2008) relating to PGC as a good estimator of LI, from Eq. [7], we obtained extinction coefficients for growing tomatoes and cauliflower with values ranging between 0.75 and 0.85 and 0.60 and 0.70, respectively. These data are consistent with the value of 0.75 obtained by Heuvelink et al. (2005) for the cultivation of tomato and of 0.55 for growing cauliflower proposed by Olesen & Grevsen (1997). Tei et al. (1996) obtained similar extinction coefficients for other horticultural crops with morphological similarities to cauliflower such as beets and lettuce (0.68 and 0.60, respectively). Campbell (1986) made an overall estimate of extinction coefficients for various crops based on the angle distribution of their leaves; considering average values for crops with leaf angles that were mainly almost horizontal, the values obtained ranged between 0.50 and 0.70.

The method of estimation of LAI was applied on four crops: Tobacco, pepper, soybean and eggplant. The crops chosen for evaluation sought validation of the method in species with very different morphological architectures, both in distribution and area occupied by the plant, as well as the height of it. It turned to be an exponential relationship between the results derived from photography method and leaf area calculated with the planimeter. Linear correlation coefficients obtained for the various crops were 0.89 for eggplant, 0.91 for pepper, soya and 0.87 to 0.88 for tobacco (Figure 13). The correlations obtained in this evaluation were quite heterogeneous despite the morphological disparity and architecture of the different species tested, yielding correlations above 0.87 in all cases. In the case of crops of eggplant, pepper and tobacco, the growth of the canopy crop growth and leaf area occurred at the same time, the highest percentage agreeing with the largest canopy leaf area indices. This dynamic growth was not followed in the case of soybean, where the plant develops its canopy in great haste once covered. The development was apical exponential until reaching a constant height. Therefore, the feasibility of this approach is restricted to the early stages of soybean development, until it reaches the highest percentage of land shaded.

Fig. 13. Relationship between leaf area index (LAI) estimated through the destructive sampling of biomass and the percentage of shaded ground measured (PGC) by the reclassification method for Tobacco, Eggplant, Pepper and Soybean.
