Samuel Martins da Costa Coura
1
Gabriel Pereira
1
Francielle da Silva Cardozo
1
Andre Lima 1
Yosio Edemir Shimabukuro
1
1
Instituto Nacional de Pesquisas Espaciais - INPE
Caixa Postal 515 - 12201-970 - São José dos Campos - SP, Brasil
{samuel,gabriel, cardozo,andre,yosio, }@dsr.inpe.br
Abstract
Remote sensing is an essential tool to obtain data of land-use and land-cover as well biophysical parameters of the vegetation that can be assimilated in numerical models. The spectral radiance is defined as the radiometric component that describes the behavior of radiation in space and can be characterized by the relationship between spectral reflectance and incident irradiance. Additionally, solar angle variations affect directly the scattering of incidence irradiation shifting the reflectance of surface objects. Relating to leaves reflectance, the pigments such as chlorophyll, xanthophyll and carotenoids promotes the absorption of electromagnetic radiation in the region of electromagnetic spectrum denominated as visible photosynthetically active radiation, providing absorption features.
Although the vegetation spectral characteristics are very well known in the remote sensing literature, few ground studies have been tested in tropical areas, especially in the Amazon region. The main objectives of this study are to analyze the variations in spectral reflectance and albedos to a sample of
Amazon rainforest located in Tapajos National Forest (TNF) due to solar angle and analyze the spectral signature responses of common species of several forest functional types. The reflectance measurements in different hours along the day were taken in the 45 meter tower, aimed to preserve same settings of orbital circumstances. The reflectance factor of different tropical leaves were also measured with a FieldSpec Pro FR Spectroradiometer from National Institute for Space Research
(INPE) to estimate the reflectance factor of selected species of trees in TNF and also litter fall samples, we used a box covered with a high density polyester material. Considering the diversity of trees species in the Amazon and satellites imagery, we found expressive differences in the spectral reflectance factor due to the variations in shadow, anisotropy behavior and transmittance properties. Also, variations of
40% could be found in dossel reflectance. These characteristics could affect the parameters of biophysics and radiation numerical models changing the exchange between surface and atmosphere.
We found expressive differences in the spectral signatures of tropical trees. Several studies are planned to be conducted in the tropics using hyperspectral technology such as orbital data of Hyperion or
AVIRIS aircraft data. Our study can provide complementary information to understand the spectral propriety of tropical vegetation types.
Introduction
Remote sensing is an essential tool to obtain data of land-use and land-cover as well biophysical parameters of the vegetation that can be assimilated in numerical models. Surface objects interact with electromagnetic radiation performing spectral signatures responses in satellite imagery. These

interactions fluctuate according with physical-chemical and biological characteristics of the surface objects. Moreover, surface objects present anisotropic and isotropic behavior with the incident energy.
The spectral radiance is defined as the radiometric component that describes the behavior of radiation in space and can be characterized by the relationship between spectral reflectance and incident irradiance (Milton, 1987). Additionally, solar angle variations affects directly the scattering of incidence irradiation shifting the reflectance of surface objects.
Relating to leaves reflectance, the pigments such as chlorophyll, xanthophyll and carotenoids promotes the absorption of electromagnetic radiation in the region of electromagnetic spectrum denominated as visible (photosynthetically active radiation), providing absorption features in 425 nm, 450 nm, 480 nm,
650 nm, 675 nm and 700 nm. The reflectance peak is found in green electromagnetic region due the chlorophyll. Moreover, an important component of leaf that interacts with radiation is the cell structure.
Some of these components are almost transparent to electromagnetic radiation in the infrared region between 700 nm to 1400 nm, such as epidermis and cuticle. Moreover, when electromagnetic energy interacts with the spongy mesophyll and with air cavities inside the leaf, the spectral reflectance in this region increases due to multiple scattering and refraction. In the middle-infrared the spectral response of leaves is mainly dominated by electromagnetic absorption due water content, being evident absorptions features in 1400 nm, 1900 nm and 2500 nm (Teillet, 1997). Although the vegetation spectral characteristics are very well known in the remote sensing literature, few ground studies have been tested in tropical areas, especially in the Amazon region. The main objectives of this study are to analyze the variations in spectral reflectance and albedos to a sample of Amazon rainforest located in
Tapajos National Forest due to solar angle and analyze the spectral signature responses of common species of several forest functional types.
Study Area
The spectral reflectance measurements were collected in the 67 km tower of Tapajos National Forest
(TNF), State of Para, Brazil. TNF is covered by two major types, as a function of geomorphologic characteristics: (1) Tropical Lowland Forests and (2) Open Rainforests. The first forest type (1) are found in areas with altitude less than 100 m, with short variation in slope and with large volume of high commercial timber values; and in Precambrian dissected relief, between an altitude of 100 m and 600 m. The second forest group (2) is frequently located in heavily dissected plateaus with slope erosion areas, narrow valleys and degraded medium texture soils (RADAMBRASIL, 1976).
In TNF the main dominant species are: Manilkara huberi, Clarisia racemosa and Copaifera multijuga , large trees that reach an upper canopy or emergent position in primary forests. Couratari guianensis, a medium to large-sized tree that usually reaches an upper canopy or emergent position in primary forests; G enipa americana ,
Coccoloba latifólia
, Eschweilera sp., E. coriácea , Miconia guianensis and
Cecropia palmate, trees with medium size that attains a middle or upper canopy position, common in abandoned shifting cultivation sites or in secondary forests; Duguetia echinophora, a small to mediumsized tree of primary and secondary forests that occurs in middle canopy position; and Astrocaryum mumbaca, a small apiny palm (Parrotta, 1995). Figure 1 shows the study area.

Figure -1- Tapajos National Forest in Pará State, Brazil.
In the Tapajós National Forest has been carried out many field campaigns of data collection (Table 1) and several studies developed (Table 2). Despite this great effort, few researches have been made with use of the integration and exploitation of different sensors on a multiscale and multitemporal approach with tropical vegetation spectral proprieties, which may provide knowledge regarding the intrinsic characteristics. The Table 1 depicts the researches campaigns carried out in the Tapajos National
Forest. Many studies in northern hemisphere aim to form, keep and update spectral libraries storing spectral reflectance data of many kinds of vegetation types. Such data is frequently applied to provide map of vegetation families and even species. It is necessary to highlight that in a tropical environment the canopy complexity and the number of ecosystems and species is too larger than those located on the temperate zones.

Campaign
SAREX “South
American Radar
Experiment”
Objectives
To assess the contribution C band
SIR-C SAR “Space
Shuttle Imaging
Radar, L and C - band”
Mission with band-P
(INPE-Germany
Company)
To evaluate the interaction of C and L band with vegetation
To evaluate the interaction of Xband and P to the vegetation
Year
1992
1994
2000
Reference
Hernandez-Filho et al. (1994);
Yanasse et al.
(1994)
Yanesse et al.
(1997)
Santos et al.
(2001); Freitas et al. (2001)
Instrument
Canadian airborne radar
Airborne radar
German airborne radar
Videography Collection of images to be in support ratings
Spectroradiometry Collection of radiometric data and positioning
2000 Steffen (2002)
2000 Steffen (2002)
Videography camera –
Arizona- USA
Radiometer (Spectralon
Plate) and GPS
Table 1 - Campaigns Remote Sensing held at the Tapajos National Forest
As noticed on the Table 1, several researches were executed on the Tapajos National Forest (TNF). The area is well preserved within its borders, although on it edges many human and natural impacts depicts an antagonist situation, where fire, deforestation, logging and agriculture expansion are current facts on the landscape surrounding the TNF. Due to this complex reality around the TNF, the Brazilian government has motivated many researches on this area aiming to develop environmental and social policies regarding the maintenance of the ecological services and as well the environmental stability.
In order to achieve a broader understanding about the several ecological and biogeochemical processes encompassing the tropical environments, a wide range of sensors were used to better provide tools to get the expected results as the LBA ( Large Scale Biosphere-Atmosphere Experiment in Amazonia ). In the Table 2 some of the main surveys on the TNF are listed, can be realized the diversity of topics related to ecology, forestry, remote sensing (mapping, inventory), and geology among others.
The integration of many kind of remote sensing sensors/satellites and field data together improve, the comprehension and amplitude of uncountable biophysical process, spectroradiometric measurements has a considerable potential to work as one more source of data to contribute on the dawning awareness concerning such knowledge area to tropical forests/species.

System
Radar
Forest regrowth
Biomass
Theme
Statistical classifiers
Satellite or
Sensor
Reference
JERS-1 Luckman (1997, 1998)
Kuplich (2002), Kuplich et al.
(2005)
JERS-1 Luckman et al. (1998)
Kuplich et al. (2005)
Radarsat
Sarex
Sant’Anna et al. (1996);
Yanasse et al. (1994)
Backscattering radar
Multitemporal studies
JERS-1 Amaral et al. (1996)
JERS-1 Kuplich e Curran (1999)
Hernandez Filho et al. (2000)
Optical
Integration
(Optical –
Radar)
Airborne missions
Vegetation spectral behaviour
Vegetal phenology
Forest inventory
Land Use
Selecting logging
Multitemporal studies
Geology
Vegetation mapping
Forest regrowth
Interferometry
Several sensors
TM
TM
Hernandez-Filho et al. (1994,
1996); Yanasse et al. (1994;
1997); Santos et al. (2001); Freitas et al. (2001)
Bernardes (1998)
Verona (2002)
TM
TM
TM
TM
Hernandez Filho et al. (1993)
Hernandez Filho et al. (1995)
Araújo et al. (1999)
Sant’Anna et al. (1996)
Landsat TM Torres (2000)
TM and
Radarsat
Shimabukuro et al. (1998)
Yanesse et al. (1997) TM and SIR
C
JERS and
ERS
Luckman (2000)
Table 2 - Remote Sensing Work developed in the Tapajos National Forest.
Material and methods
The reflectance measurements in different hours along the day were taken in the 45 meter tower, aimed to preserve same settings of orbital circumstances. Considering the technical difficulties to obtain spectral reflectance of a ground object, we used a spectral ratio denominated as reflectance factor. The reflectance factor of an object is given by the ratio between the measurement of the spectral radiance of the object and the spectral radiance collected from an ideal Lambertian surface acquired in the same conditions of illumination and observation (Parrotta, 1995).
The reflectance factor of a dossel sample in the 45 m tower in TNF were measured with a FieldSpec
Pro FR spectroradiometer from National Institute for Space Research (INPE), with a IFOV of 25°, spectral range varying from 350nm to 2500nm, spectral resolution ranging from 3 to 10 nm and

acquisition time of 0.1 second per spectrum. We used a lambertian Spectralon 11 plate with a spectral reflectance of approximately 100% in all solar electromagnetic spectrum.
The reflectance factor of different tropical leaves were also measured with a FieldSpec Pro FR
Spectroradiometer from National Institute for Space Research (INPE) to estimate the reflectance factor of selected species of trees in TNF and also litter fall samples, we used a box covered with a high density polyester material. This material presented a low value of transmittance and reflectance, with a high absorption, mainly due to material composition and black color. In cited box, the samples of leaves are placed and the spectral reflectance factor was measured. It is noteworthy that polyester material allowed the measurements of reflected flux, since all transmitted electromagnetic radiation that cross leaf were absorbed, eliminating background contamination.The FieldSpec spectroradiometer radiance measurements (L) can be represented by ratio of radiant flux ( ) per solid angle ( ), cosine of zenith angle ( ) and the area (A):
[W.m
-2
.sr
-1 .µm -1
] (1)
Reflectance factor ( RF ) can be expressed as a relationship between the sample radiance ( ) and field lambertian plate radiance ( ) in the same observation ( ) and acquisition characteristics ( ).
However, the lambertian plate used in the field does not have a reflectance of 100% in all solar electromagnetic spectrum and the spectroradiometer used in field must be calibrated. Therefore, these factors must be corrected using an ideal lambertian plate ( ) with laboratorial use only and a calibrated unit of INPE:
(2)
Results and discussions
Figure 2 shows an average reflectance factor of 30 measurements collected for each local time in the 67 km TNF site with fixed geometry. In the graphs, we can observe that fluctuations in reflectance factor in photosynthetic active radiation (PAR) region remains almost constant for all samples, varying from absolute values of 5% at 8:30 local time (LT) to 7.5% at 10:30-12:30 LT. Moreover, in contrast with
PAR (400-700 nm) region, in shortwave infrared spectral range (700–2500 nm), where the electromagnetic radiation reflectance are controlled mainly by leaf cell structures and tissues such as epidermis and mesophyll, the reflectance factor presented a substantial difference varying according to distinct hour angles. Therefore, the spectral values between 1350-1450 nm and 1850-1950 nm intervals are not showed because this feature represents the absorption band of the electromagnetic energy by water vapor in the atmosphere.
To describe the spectral propriety of the vegetation, we selected 12 common species of trees based the frequency distribution of trees in TNF and also litter fall samples. The tree species selected were:
Manilkara huberi, Clarisia racemosa and Copaifera multijuga are large trees that reach an upper canopy or emergent position in primary forests. Couratari guianensis is a medium to large-sized tree that usually reaches an upper canopy or emergent position in primary forests. G enipa americana ,
Coccoloba latifólia
, Eschweilera sp., E. coriácea , Miconia guianensis and Cecropia palmata are trees with medium size that attains a middle or upper canopy position, common in abandoned shifting cultivation sites or in secondary forests. The Duguetia echinophora is a small to medium-sized tree of

primary and secondary forests that occurs in middle canopy position. The Astrocaryum mumbaca is a small apiny palm (Parrotta, 1995)..
Figure 2 – Spectral reflectance of canopy in different local time.
In Figure 2, variations of 20% in spectral reflectance in near infrared (700-1300 nm) can be identified.
At 8:30 LT, the canopy reflectance in this region presented values near 25%, however, from 11:00 to
13:00 LT the values increased to 45%. In Amazon rainforest, the variations in spectral reflectance values are associated with shadows and with anisotropy behavior of dossel. According to [5], the amount of shadow in a scene is an important factor that influences the reflectance of surface objects.
Shading objects in instantaneous field of view (IFOV) can vary in quantity and in intensity. Also, we can observe that the decrease in shadow area results in increase of canopy spectral reflectance factor, mainly for spectral region of electromagnetic spectrum in green/red and near infrared. Furthermore, another canopy property that influences the reflectance factor is the leaf transmittance, for example, lower transmittance values of leaves can represent in intense shadows, decreasing, consequently, the reflectance of determinate sample
Figure 3 shows an average reflectance factor of 30 measurements collected for each tree species in TNF site. We noticed that reflectance factor in the visible region (400-700nm) of electromagnetic spectrum remains almost constant for all samples, varying in function of different leaf pigments concentration such as carotenoids, chlorophyll and xanthophylls.

Figure 3 – Spectral reflectance of tree leaves.
In contrast with visible electromagnetic region, in shortwave infrared spectral range (700 nm –
1350nm), where the radiation interaction process are controlled mainly by leaf cell structures and tissues such as epidermis and mesophyll, the reflectance factor undergo for changes in the spectral signature due to physical-chemical and biological properties of each leaf. Species like Cecropia palmata, Duguetia echinophora, Miconia guianensis, Astrocaryum mumbaca, Clarisia racemosa,
Copaifera multijuga and G enipa americana have similar spectral behavior along all electromagnetic solar spectrum, this fact could be associated with similar cellular structure, chemical composition and water content.
Furthermore, the following species
Manilkara huberi, Couratari guianensis, Coccoloba latifólia,
Eschweilera sp., and E. coriacea yields a distinct spectral signature compared with species mentioned above. One of the possible reasons to explain this difference is related to tissue structure such as spongy mesophyll which has a significant role in the infrared radiation reflectance. Leaves that are thinner than others usually have a higher transmittance because there is less of spongy tissue and water content in its structures. Moreover, the spectral values between 1350nm-1450nm and 1850nm-1950nm intervals are not showed because this feature represents the absorption band of the electromagnetic energy by water vapor in the atmosphere.
Figure 4 shows the percentage changes that occurs in albedo values as a function of spectral measurement made at 10:00 LT. In the graph, we can notice that the albedo values increase from the

sunrise to 11:00 LT, decreasing with progression of hours. In this case, the albedo variation could reach
40% from the peak of spectral reflectance when compared with initial or afternoon hours of day.
Figure 4 Albedo variation due to hour angle.
Conclusions
Considering the diversity of trees species in the Amazon and satellites imagery, we found expressive differences in the spectral reflectance factor due to the variations in shadow, anisotropy behavior and transmittance properties. Also, variations of 40% could be found in dossel reflectance. These characteristics could affect the parameters of biophysics and radiation numerical models changing the exchange between surface and atmosphere.
Due to the diversity of trees species in the Amazon, we found expressive differences in the spectral signatures of tropical trees. Several studies are planned to be conducted in the tropics using hyperspectral technology such as orbital data of Hyperion or AVIRIS aircraft data. Our study could provide complementary information to understand the spectral propriety of tropical vegetation types
References
Amaral, S.; Shimabukuro, Y.E. (1996). Sensoriamento remoto por radar (SAR): Pré-processamento de imagens Radarsat (“Fine Mode”) na Região da Floresta Nacional do Tapajós. Primeiras jornadas latinoamericanas de percepción remota por radar -técnicas de procesamiento de imágenes, Buenos
Aires, 37-42.
Araújo, L.S. (1999). Análise da cobertura vegetal e de biomassa em área de contato Floresta/Savana a partir de dados TM/Landsat e JERS-1. São José dos Campos. 129 p. (INPE-7253-TDI/696).
Dissertação (Mestrado em Sensoriamento Remoto) - Instituto Nacional de Pesquisas Espaciais.
Bernardes, S. (1998). Índices de vegetação e valores de proporção na caracterização de floresta tropical primária e estádios sucessionais na área de influência da Floresta Nacional do Tapajós - Estado do Pará.
São José dos Campos. 97 p. (INPE-6890-TDI/651). Dissertação (Mestrado em Sensoriamento Remoto)
- Instituto Nacional de Pesquisas Espaciais.
Colwell, J. E. (1974). Vegetation canopy reflectance. Remote Sensing of Environment, 3, 175-183.

Freitas, C.C.; Sant’Anna, S.J.S.; Soler, L.S.; Santos, J.R.; Dutra, L.V.; Araújo. L.S.; Mura, J.C.;
Hernandez Filho, P. (2001). The Use of Airborne P-Band Radar Data for Land Use and Land Cover
Mapping in Brazilian Amazonia. [CD ROM]. International Geoscience and Remote Sensing
Symposium, 9-13 July.
Hernandez Filho, P.; Lee, D.C.L.; Shimabukuro, Y.E. (1996). Classificação da cobertura vegetal utilizando dados do SAR aerotransportado e do TM/Landsat: Região da Floresta Nacional do Tapajós.
Investigationes Geográficas: Boletin del Instituto de Geografia, México, 5, 19-25.
Hernandez Filho, P.; Shimabukuro, Y.E.; Lee, D.C.L.; Santos Filho, C.P.; Almeida, R.R. (1993).
Relatório final do projeto de inventário florestal na Floresta Nacional do Tapajós. INPE - São José dos
Campos, (INPE–5422-PRP/170), 122p.
Hernandez Filho, P.; Dutra, L.V.; Amaral, S. (1994). Resultados preliminares da análise visual de dados SAREX e descrição das atividades de campo na Floresta Nacional do Tapajós. INPE - São José dos Campos, (INPE-5567-NTC/314), 42p.
Hernandez Filho, P.; Yanasse, C.C.F.; Sant’Anna, S.J.S.; Kuplich, T.M.; Santo, P.P.; Dutra, L.V.;
Orgambide, A.C.F.; Oliveira, P.M.; Cruz, R.J.D.; Azevedo, V.R.G. (1995). Análise dos dados TM:
Floresta Nacional do Tapajós. INPE - São José dos Campos, (INPE-5614-RPQ/670), 58p.
Kuplich, T.M.; Curran, P.J.; Atkinson, P.M. (2005). Relating SAR image texture to the biomass of regenerating tropical forests. International Journal of Remote Sensing, 26: 4829-4854.
Kuplich, T.M.; Curran, P.J. (1999). Temporal analysis of JERS-1 SAR images over regenerating forests in Brazilian Amazonia. In: International Geoscience and Remote Sensing Symposium - IGARSS, 1999,
Hamburg. Proceedings. Piscataway : IEEE, IV,1895-1897.
Luckman, A.; Baker, J.; Honzák, M.; Lucas, R. (1998). Tropical forest biomass density estimation using JERS-1 SAR: Seasonal variation, confidence limits, and application to image mosaics. Remote
Sensing of Environment, 63, 126-139.
Luckman, A.; Baker, J.; Wegmüller, U. (2000). Repeat-pass interferometric coherence measurements of disturbed tropical forest from JERS and ERS satellites. Remote Sensing of Environment, 73, 350-360.
Luckman, A.; Baker, J.; Kuplich, T.M.; Yanasse, C.C.F.; Frery, A.C. (1997). A study of the relationship between radar backscatter and regenerating tropical forest biomass for spaceborne SAR instruments, Remote Sensing of Environment, 60, 1-13.
Milton, E.J. (1987). Principles of field spectroscopy. International journal of Remote Sensing, 8, 12,
1807-1827.
Parrotta, J.A.; Francis, J.K.; Almeida, R.R. (1995). Trees of the Tapajós: a photographic field guide .
International Institute of Tropical Forestry and United States Department of Agriculture and Forest
Service. Río Piedras, Puerto Rico: USDA, 370 p.

RADAMBRASIL (1976). Departamento Nacional de Produção Mineral. Folha AS.21- Santarém.
Geologia, geomorfologia, pedologia, vegetação e uso potencial da terra. Rio de Janeiro: DNPM, 510 p.
Sant’Anna, S.J.S.; Yanasse, C.C.F.; Frery, A.C. (1996). Estudo comparativo de alguns classificadores utilizando-se imagens Radarsat da região do Tapajós. Primeiras jornadas latinoamericanas de percepción remota por radar - técnicas de procesamiento de imágenes, Buenos Aires, 187-194.
Santos, J.R.; Araújo. L.S.; Freitas, C.C.; Sant’Anna, S.J.S.; Dutra, L.V.; Mura, J.C.; Gama, F.F.;
Hernandez Filho, P. (2001). Inventory of forest biomass in Brazilian Amazon a local approach using airborne P-band SAR data. [CD ROM]. International Geoscience and Remote Sensing Symposium,
IGARSS-2001.
Teillet, M.; Staenz, K.; Williams, D.J. (1997). Effects of spectral, spatial, and radiometric characteristics on remote sensing vegetation indices of forested regions. Remote Sensing of
Environment, 61,139-149.
Torres, R.B. (2000). Contribuição ao estudo do arcabouço estrutural da área do rio Tapajós, bacia do amazonas, através da ánalise integrada de dados de topografia, geologia, magnetometria, gravimetria e sensoriamento remoto. São José dos Campos. 153 p. (INPE-7996-TDI/749). Dissertação (Mestrado em
Sensoriamento Remoto) - Instituto Nacional de Pesquisas Espaciais.
Verona, J.D. (2002). Monitoramento da cobertura vegetal na região da Floresta Nacional do Tapajós -
Pará, utilizando dados multitemporais do sensor Thematic Mapper (TM) do Landsat. Dissertação
(Mestrado em Sensoriamento Remoto) - Instituto Nacional de Pesquisas Espaciais.
Yanasse, C.C.F.; Sant’Anna, S.J.S.; Frery, A.C.; Rennó, C.D.; Soares, J.V.; Luckman, A.J. (1997).
Exploratory study of the relationship between tropical florest regeneration stage and SIR-C L e C data.
Remote Sensing of Environment, 59, 180-190.
Yanasse, F.C.C.; Frery, A.C.; Sant’Anna, S.J.S.; Hernandez Filho, P.; Dutra, L.V. (1994). Statistical analysis of SAREX data over Tapajós - Brazil. INPE - São José dos Campos, (INPE-5559-PRE/1798),
16p..