SPECTRAL REFLECTANCE OF IPÊ-AMARELO LEAVES UNDER DIFFERENT FORMS OF STORAGE AND COLLECTION TIME

The application of spectroradiometry techniques to the study of tree species allows the acquisition of information related to plant physiology and morphology, which can be used in conjunction with orbital images. However, it is known that when a leaf is extracted the senescence process is started, which is characterized by cell constituents degradation, loss of water and modifications of the foliar mesophyll structure. In this context, this study aimed to evaluate the influence of the collection time and storage form on the spectral response of ipê-amarelo leaves. Thus, 32 leaves were collected in distinct times (1h, 2h, 24h and 48h) and conditioned with(C) and without(S) thermal storage. The data were evaluated from the visual analysis of the spectral curves, derivative analysis and statistical analysis. The experiment was conducted in a Randomized block design and data were submitted to analysis of variance (ANOVA) and Tukey test (p = 0.05). Results indicated that thermal storage might delay the senescence process of the leaves. Both the collection time and the storage form affect the pattern of spectral behavior of the leaves of ipê-amarelo. The collection time has not shown significant differences between 1-2h. The largest differences were found between 2-24h after collection.


INTRODUCTION
Remote sensing is a technique that investigates the interaction between electromagnetic radiation and different types of targets.In vegetation studies, the appliance of this technique allows the acquisition of information on the distribution of different types of vegetation, phenological status, stress conditions, nutrient deficiency, among others.
In the case of Brazilian native species, such as ipê-amarelo, the need for refinement of the remote sensing techniques at laboratory or field level is increasing, justifying the choice of this species.These techniques play an important role in a wide variety of scientific studies and allow monitoring changes in natural areas at different scales, as well as differentiate and identify by remote sensing the different types of native vegetation.Therefore, quantifying the influence exerted by the environment in the acquisition of native species spectral behavior data becomes relevant and allows greater control of variables that can affect the final response.
The obtaining data stage in vegetation studies by remote sensors is essential.According to Novo (2010), it can be performed in three different levels: laboratory or field, aerial and orbital.These levels include the study of canopies and/or isolated leaves (extracted or not).In the isolated leaves approach, studies include acquisition of data at laboratory and field levels, aiming the spectral characterization of phenomena directly or indirectly related to plant physiology and morphology (FONSECA et al., 2002).
According to Ponzoni et al. (2012), the parameters that influence the spectral reflectance of leaves refer to their chemical, morphological, physiological and internal humidity composition and each of them exerts a predominant influence in at least three spectral regions of the optical spectrum (visible, near infrared and medium infrared).Thus, we can say that spectroradiometry is an efficient technique for more reliable spectral characterization among plant species, their compositions and humidity conditions.
Laboratory studies of spectral response of plant targets can also be used in conjunction with orbital images in order to obtain field parameters to assist in the interpretation of specific targets in a given period.For that, spectral curves are taken from different species using non-imaging spectroradiometers under laboratory conditions, which are compared with the curves obtained by the remote sensors.
Also called by hyperspectral remote sensing, the spectroradiometry provides advances in different scopes.Scientific efforts have been devoted in the spectral characterization of tree and agricultural species (FONSECA et al., 2002;BRANDELERO et al., 2012;REX et al., 2016;KÄFER et al., 2016;SCHUH et al., 2016) to evaluate the method of collecting spectroradiometric data and its influence on vegetation indices (SCHRODER et al., 2015) and to estimate biochemical and biophysical characteristics of the vegetation (GOERGEN et al., 2015;FOSTER et al., 2016).
In the scope of vegetation spectral characterization studies, few works discuss the influence of collection time and form of storing the leaves from the collection to the place where the radiometric measurements will be performed.Of particular note is the one developed by Souza et al. (1996), who evaluated the influence of time and type of storage on the spectral response of Eucalyptus grandis Hill ex Maiden leaves.It is known that when a leaf is extracted from the plant the senescence process begins, characterized by the degradation of cellular constituents, loss of water, and modifications on structure of foliar mesophyll and, consequently, of its spectral properties (SCHUH et al., 2016).
In this context, the aim of this study is to evaluate the influence of collection time and storage form on the spectral behavior of ipê-amarelo leaves.

Characterization and location of species
The species ipê-amarelo (Handroanthus chrysotrichus Mart.Ex A.DC. Mattos) belongs to the family Bignoniaceae and has important economic relevance due to attractive properties of wood, which is heavy and resistant, allowing its use for noble purposes.In addition, it has ornamental and ecological importance.Morphologically, the species is characterized for presenting a height of 4-5 m, 30-40 cm in diameter, compound leaves (5 leaflets) and yellow flowers (LORENZI, 2008).
The ipê-amarelo used in the collection of samples (leaves) is located near the State Center of Remote Sense and Meteorology (CEPSRM) at the Federal University of Rio Grande do Sul (UFRGS) in Porto Alegre -RS and it is found isolated, therefore, exposed to full sun.

Experimentation and Sampling
The experiment was conducted with a randomized statistic design in blocks (DBC).Two variation factors were evaluated, the time of collection and the form of storage, consisting of eight treatments, which will be described as follow.
Four samples (leaves) were collected for each treatment corresponding to the N, S, L, and O directions, located in the lower portion of the tree and in the middle portion of the branches, totaling 32 samples.The leaves were stored in properly identified punched paper envelopes.The treatments with thermal storage were packed in styrofoam thermal box, the others were stored in ambient conditions.The treatments were sent to the laboratory where the radiometric readings were performed.

Climatic Conditions
The three days of collection and the days prior to those defined were characterized by climatic variables, accumulated daily precipitation, temperature and relative humidity, in order to demonstrate the environment conditions of the leaves (Table 1 The treatments submitted to thermal storage condition had no external influence, since the styrofoam box represents a homogeneous environment.On the other hand, the treatments in ambient conditions, although they were placed in the same place, they may have been influenced by the environment; however, it is important to highlight that temperature and relative humidity of air on the days of measurements were uniform, as seen in Table 1.

Radiometric readings
The radiometric readings were performed at the Radiometry Laboratory of CEPSRM -UFRGS with the support of a spectroradiometer FieldSpec, which has the capacity to record reflectances between 300-2500 nm wavelengths.The readings were taken with the support of leaf clip, the equivalent of an integrating sphere.
The measurements of reflectance of the adaxial face of leaves, which were carefully positioned, were obtained so that only the foliar limb stayed in the sphere's orifice, avoiding the interference of their central vein on themselves.The procedures were the same for all treatments.A microcomputer coupled to the spectroradiometer stored the radiometric readings as they were obtained.
The original data from the measurements were converted to text format (.txt) through the software ASD ViewSpecPro (Version 6.0), in order to make the data format compatible with the other applications and softwares used.
For the visual analysis of the reflectance curves, graphs containing the spectral response of the treatments in each region were generated.In the derivative analysis, the first and the second derivative were automatically applied by the ASD ViewSpecPro software, and a GAP value of 5 was chosen for correction of imperfections.This technique consists of enhancing features that stand out in the spectral signature of a target, both in reflectance peaks and absorption bands (REX et al. 2016).
For statistical analysis of data, it was adopted the statistic software R (R CORE TEAM, 2017).The average reflectance factor for each treatment was determined, which corresponded to the arithmetic average of four repetitions per treatments.Ultimately, the data were submitted to analysis of variance (ANOVA) and Tukey's multiple average comparison (HSD-Honestly significant difference) with 95% of reliability.

Spectral Behavior
The spectral responses obtained for each treatment (Figure 1) show that the most pronounced differences occurred in the near infrared and medium infrared regions.Thus, the ST2 and ST3 treatments (both without thermal storage) in the near infrared region exhibited higher reflectances than the other treatments, in addition, they presented smoothness in the curve at the transition point between the visible and near infrared regions.
The aforementioned treatments also showed the lowest reflectances in the visible region and showed lower absorption peaks in the medium infrared region, a region frequently used for detecting hydric stress in vegetation (RODRIGUES et al. 2016).
Figura 1. Curvas de reflectância de folhas de ipê-amarelo submetidas a distintos tempos de coleta e formas de armazenamento.As fleshas indicam os pontos de absorção na região do infravermelho médio.Figure 1.Reflectance curves of ipê-amarelo leaves submitted to different forms of storage and collection time.The arrows indicate the absorption points in the medium infrared region.

Derivative Analysis
The derivative analysis (Figure 2) allowed to identify larger variations in reflectance and absorption spectra.In general, the first derivative (Figure 2a) shows the highest reflectance and absorption peaks at approximately the same wavelengths in the highest inclination of the original reflectance curve (500 nm, 700 nm, 1,400 nm, 1900 nm).
The second derivative (Figure 2b) presents similar behavior, but in a more detailed way.At the wavelength of 700 nm, an absorption peak occurs, which the first derivative was not evidenced.In addition, at approximately 1,400 nm and 1,900 nm it is verified the enhancement of reflection peaks, which were shown to be smoothed in the first derivative.
It should be noted as a result of this analysis that at the wavelength of approximately 1,000 nm (near infrared region), a reflectance peak, which could not be seen prior to the application of the first and the second derivative, was observed.In this perspective, the second derivative was able to show this peak, allowing to discriminate the treatments ST0 and CT0.Figura 2. Análise derivativa de curvas de reflectância de folhas de ipê-amarelo submetidas a distintos tempos de coleta e forma de armazenamento.(a)Refere-se à primeira derivada, enquanto (b) refere-se à segunda derivada.Figure 2. Derivative analysis of ipê-amarelo leaves reflectance curves submitted to different forms of storage and collection time.(a) Refers to the first derivative, while (b) refers to the second derivative.

Statistical analysis
A statistical analysis using ANOVA allowed to evaluate the null (H0), which verifies the existence of homogeneity of variance for the two variation factors considered in this study (time and storage).For the visible region (Table 2), it was possible to identify that there is at least one significant difference between treatments, i. e., at least one moment of acquisition of samples, thus rejecting the hypothesis H0.However, the same was not observed between the blocks, in which it was verified that there is no significant difference between the forms of storage for the visible region, accepting the hypothesis H0.In other words, in this spectral region time significantly affected the spectral response of the ipê-amarelo leaves, whereas the form of storage was not determinant to find significant differences in the spectral curves.
For the near infrared regions (Table 3) and medium infrared (Table 4), two sources of variation analyzed showed significant differences.Therefore, the hypothesis H0 was rejected accepting the alternative hypothesis that there was heterogeneity of variances.Therefore, both time and storage influenced the spectral response of the ipêamarelo leaves.
Tabela 2. Análise de variância dos diferentes tempos de coleta e forma de armazenamento para a região do visível.The Tukey test at 5% probability allowed to identify which treatments differed from each other in relation to the two sources of variation analyzed.As the blocks did not have statistical differences in the visible region, i. e., the use of thermal storage did not influence the spectral response of the curves in this region, only the collection times were analyzed (Table 5).The treatments T0 and T1 did not differ among themselves, the same was observed for treatments T2 and T3.However, treatments with the greatest time difference, such as T0 in comparison to T2 and T3 and also T1 in comparison to T2 and T3, differed statistically, demonstrating that between the period of 2h-24h, significant changes occur in the spectral curves in the ipê-amarelo leaves due to water loss by the leaf.
In the near infrared, in addition to collection time, there was influence of thermal storage on spectral response of curves.Thus, the results of the Tukey test were evaluated considering the interactions of the two variation sources (Table 6).The treatments ST0 and ST1, as well as CT0 and CT1, showed to be equal in the two blocks, indicating that thermal storage does not influence between 1-2h, since the leaf does not present significant water loss in the period of only 1h in this spectral region.Tabela 5. Teste de comparação de médias (Tukey) dos diferentes tempos de coleta para região do visível.Table 5. Averages comparison test (Tukey) of different collection times for the visible region.Finally, in the medium infrared region the result was similar to that obtained for the near infrared, i. e., there was also no difference between 1-2h.Between blocks, the difference again occurred.The exception remained for treatments ST0 and CT0 and for ST1 and CT1.Additionally, it was possible to observe that the treatments ST2, CT2 and ST3, CT3 did not differ significantly, showing that the water loss that occurred between 24-48h after the collection presented no significant influence on the spectral responses as well.In this region, the greatest differences were found between 2-24h after collection, as well as in the near infrared.Tabela 6. Teste de comparação de médias (Tukey) dos diferentes tempos de coleta e armazenamento para região do infravermelho próximo (IVP) e infravermelho médio (IVM).Table 6.Averages comparison test (Tukey) of different collection times and storage form for the near infrared region (NIR) and medium infrared (MIR).

Interactions
).The daily data were obtained from the National Institute of Meteorology (INMET) of the Meteorological Station of Porto Alegre located at latitude -30.053536º, longitude of -51.174766º and altitude of 41 meters.Tabela 1. Dados das variáveis climáticas precipitação, temperatura e umidade relativa diária.Table1.Climatic variables data precipitation, temperature and daily relative humidity.
* Indicates the collection dates.

Table 2 .
Variance analysis of different collection times and storage form for the visible region

Table 3 .
Variance analysis of different collection times and storage form for the near infrared region.

Table 4 .
Variance analysis of different collection times and storage form for the medium infrared region.