Relationship Between Canopy and Leaf Spectral Response In Savanna

1. Introdução Analyses of various biophysical and biochemical factors affecting plant canopy Revista Brasileira de Geografia Física 05 (2012) 1203-1214 Galvincio. J. D.; Moura, M. S.; Silva, T. G. F.; Silva, B. B.; Naue, C. R. 342 reflectance have been carried out over the past few decades, yet the relative importance of these factors has not been adequately addressed. A combination of field and modeling techniques were used to quantify the relative contribution of leaf, stem, and litteroptical properties (incorporating known variation in foliar biochemical properties) and canopy structural attributes to nadir-viewed vegetation reflectance data (Asner 1998a). Many ecological processes (e.g., species growth, invasion, competition, facilitation, mortality) are dictated by existing spatial conditions (i.e., both biotic and abiotic) and lead to the generation of new spatial conditions. Hence, Pattern and process are perpetually intertwined. In order to understand how nature works, ecologists must be able to classify patterns and develop process-based spatial explicit models that can produce observed patterns at multiples scales (Drake and Weishampel 2000). Canopy architecture plays fundamental roles in the land-atmosphere interactions, yet quantification of canopy architecture using optical sensors in an open canopy remains a challenge. Savannas are spatially heterogeneous, open ecosystems, thus efforts to quantify canopy structure with methods developed for homogeneous, closed canopies are prone to failure. Savannas are globally important ecosystems of great significance to human economies (Sankaran et al. 2005). Savannas exist in water-limited regions, which force tree canopies open and heterogeneous (Eagleson and Segarra 1985), (Ryu and Science 2010). The open canopy structure allows grass to co-dominate in the savannas by occupying different niches in space and time. The co-dominance of trees and grass defines the functions and metabolisms in the savanna ecosystems (Higgins, Bond, and Trollope 2000);(House et al. 2003). However, how to quantify canopy architecture and how to monitor structure, function, and metabolism in savanna ecosystems remains


Introdução
Analyses of various biophysical and biochemical factors affecting plant canopy reflectance have been carried out over the past few decades, yet the relative importance of these factors has not been adequately addressed.A combination of field and modeling techniques were used to quantify the relative contribution of leaf, stem, and litteroptical properties (incorporating known variation in foliar biochemical properties) and canopy structural attributes to nadir-viewed vegetation reflectance data (Asner 1998a).
Many ecological processes (e.g., species growth, invasion, competition, facilitation, mortality) are dictated by existing spatial conditions (i.e., both biotic and abiotic) and lead to the generation of new spatial conditions.Hence, Pattern and process are perpetually intertwined.In order to understand how nature works, ecologists must be able to classify patterns and develop process-based spatial explicit models that can produce observed patterns at multiples scales (Drake and Weishampel 2000).vegetation that characterizes that landscape (Eagleson and Segarra 1985).Savanna is curious vegetation state characterized by the coexistence of grasses and trees.Although the exact ratio of grass to tree varies considerably with savanna type, the physiognomy of savanna remains clearly distinct from that of grassland and forest.Most authors, would agree that a complex web of factors, notably, water, herbivory, fire, soil texture and nutrients, influences the balance between grass and trees in savanna (Higgins, Bond, and Trollope 2000).Given this complexity, the question of how grasses and trees coexist over such a wide range of climate, edaphic, biogeographic and historical conditions is intriguing so intriguing that it has been referred to as the "savanna problem" (Higgins, Bond, and Trollope 2000).

Models based on physical processes
proved to be a promising alternative to describe the transfer and interactions of radiation inside the canopy based on physical laws and thus provide an explicit connection between the biophysical variables and canopy reflectance.
To date, there are few studies using high spatial resolution satellites to estimate the chlorophyll content of the plant.This difficulty occur due the estimates obtained to present date are unreliable, many underestimate the values of chlorophyll content and consequently all other estimates that use this data.It is believed that these underestimates occur due to differences in reflectance between the canopy and leaf.
Thus, this study aims to evaluate the similarity and the correlation between canopy reflectance and leaf and propose a model that improves the estimates in leaf scale.

Study area
The spatial location of the research sites under study is in the municipalities of Petrolina and Serra Talhada in the state of Pernambuco, Northeastern Brazil, Fig. 1.

Field data
Data were collected on a grid 500 x 500 meters, which corresponds to an area of 250,000 m2 per site.A total of two sites, one in Serra Talhada and another in Petrolina.
Each site was divided into a grid of 5 x 5 or 25 sampling points.

IKONOS satellite data
The IKONOS satellite data were acquired for the PELD, site 22,

Spectral reflectance data
Spectral reflectance was measured between 336 and 1045 nm with a spectral resolution of 1 nm, covering visible and nearinfrared portions of the electromagnetic spectrum.Fieldspec HandHeld (ASD, Boulder, USA) fitted with a fiber optic probe having a 25° field of view was used.The spectroradiometer was optimized using reference white plate.

Cluster analysis
Methods were used for cluster analysis was Ward, the method of partial correlation was Pearson and linear regression method.
The Ward method is also called "Minimum Variance".In this method the formation of groups is by maximizing homogeneity within groups.The sum of squares within groups is used as a measure of homogeneity.That is, the Ward method attempts to minimize the sum of squares within the group.The groups formed in each step are the result of group solution with the smallest sum of squares.

T-test
The two-sample t-test for testing whether differences exist between two population means was adopted in this study to determine difference between canopy and leaf reflectance.Numerous studies have shown that the two-sample t test is robust to considerable departures from its theoretical assumptions (that both samples come at random from normal populations with equal variances), especially if the sample sizes are equal or nearly equal (Boneau, 1960;Cochran, 1947;Posten et al., 1982;Zar, 1996).We tested the research hypothesis that the means of the leaf and canopy indices for each species were different, i.e., Ho: μ1 = μ2 versus the alternative hypothesis, H1: μ1 ≠ μ2, where μ1 and μ2 are the means of leaf and canopy reflectance, respectively.The t values were calculated using Eq. 1. (1) where, , sd 1 and sd 2 and n1 and n2 represent the means, standard deviations and sample sizes of the leaf and canopy data, respectively.

Correlation and regression
To learn how similar grouped data are was used to the partial correlation analysis of Pearson.
We used linear regression to develop a model to better estimate the reflectance in leaf using reflectance on the canopy.This model is suggestion for estimate in spatial scale smaller than 3 meters or area smaller than 9m 2 .

Radiometric calibration of the IKONOS
The radiometric calibration of satellite sensors IKONOS can be obtained by Equation 2: where DN λ is digital number in band λ and Coef λ e Bandawith λ are the coefficient of calibration in band λ , Table 1.d ) is 1 (one).This value r d can fluctuate between 0.97 and 1.03.
Note that the data were grouped into two groups sharply.Group 1 -refers to the measures in the canopy and Group 2 -the measurements taken in the leaves.Thus, it is suggested that the reflectance measurements of leaf and canopy are not similar.
Applying the t-test on the samples from the canopy and leaf was obtained significant differences between samples with significance level 0.01, ie a confidence interval of 99%.Note that this study show significant differences between canopy and leaf.Similar results was obtained in (Cho et al. 2008) when analyses the sensibility between hyperspectral index in different scales, canopy and leaf.

Correlation and regression analysis
Correlation was obtained between the samples of the canopy.The correlations between the samples were always greater than 0.80 with a significance level of 0.05.Ie, the data collected in the canopy are highly correlated.
Knowing that the data obtained in the   The results of this study revealed systematically higher near infrared (NIR) and visible (V) reflectance's at the leaf scale than at the top-of the canopy.The higher leaf V and NIR reflectance may be explained by the effect of multiple scattering caused by leaf stacking since the leaf reflectance were measured in situ, (Cho et al. 2008).Blackburn Scientist around the world have sought to improve the estimates of the different products estimated with remote sensing, (Liu et al. 2011), (Elmore and Craine 2011), (Cho et al. 2008), (Danson and Bowyer 2004), (Asner and Warner 2003), (Eamus, Hutley, and O'Grady 2001), (Drake and Weishampel 2000), (Asner 1998b) and (Knyazikhin et al. 1998), But improving the differences between the estimates of canopy reflectance and leaf virtually nonexistent.
The quantification of clumping effects at the ecosystem scale, which has been overlooked in most remote sensing products, reflectance, is crucial to obtain the correct information of ecosystems.

Conclusions
The model developed in this study may be used to improve the estimates of reflectance and indices spectral in leaf with high spatial resolution images, such as images IKONOS.
This study reinforces the importance of developing models to better to estimate the reflectance in scale of detail (leaf) since there are no similarities between the reflectances in the leaf and canopy but a high statistical correlation.

Conselho Nacional de
Desenvolvimento Científico e Tecnológico roles in the land-atmosphere interactions, yet quantification of canopy architecture using optical sensors in an open canopy remains a challenge.Savannas are spatially heterogeneous, open ecosystems, thus efforts to quantify canopy structure with methods developed for homogeneous, closed canopies are prone to failure.Savannas are globally important ecosystems of great significance to human economies (Sankaran et al. 2005).Savannas exist in water-limited regions, which force tree canopies open and heterogeneous (Eagleson and Segarra 1985), (Ryu and Science 2010).The open canopy structure allows grass to co-dominate in the savannas by occupying different niches in space and time.The co-dominance of trees and grass defines the functions and metabolisms in the savanna ecosystems (Higgins, Bond, and Trollope 2000);(House et al. 2003).However, referring to flat and open landscapes, and other times referring to the quantities derived from images with bands in the electromagnetic spectrum include vegetation indices, albedo and leaf area index, which may be related to the efficiency of light use by the plant community and, therefore, are valuable elements in the carbon assimilated by vegetation, water balance and energy balance between the surface and atmosphere, (Bonan 1995).In addition, satellite images in bands of green and red spectral provide estimates of leaf chlorophyll content (Cab), which is an important indicator of the physiological conditions of the plants.The advantages of using remote sensing for monitoring terrestrial ecosystems have been well documented in Brazil (B.B. Silva et al. 2013), (Galvíncio, Pimentel, and Mendonça 2012), (Pereira, França, and Galvincio 2012), (L.G. da Silva and

Fig. 1 .
Fig. 1.Spatial location of the municipalities Serra Talhada and Petrolina in Brazil.

( 1999 )
showed that the NIR and to a lesser degree, the visible reflectance increases with leaf stacking.He equally argues that the spectral reflectance properties of background materials and areas of shadow can have large influence upon that of the whole canopy even when there is complete canopy.It is known that the reflectance is the primary data obtained from satellite images and case presents error in data in your measurement / calibration, all other products derived this increase noise, as vegetation indices (NDVI -Normalized Difference Vegetation Index, SAVI-Soil Ajusted Vegetation Index.),Leaf Area Index (LAI),

Table 1 -
Radiometric calibration coefficient of by band.

Table 2 .
Reflectance in 10 points of dense forest in the savanna, site 22, Petrolina-PE, LTER-Long Term Ecological Program, using IKONOS imagery.