International Journal of Applied Earth Observation and Geoinformation
Possibilities of discriminating tropical secondary succession in Amazônia using hyperspectral and multiangular CHRIS/PROBA data
Introduction
New satellite sensors acquiring quasi-simultaneous multiangular images facilitate reflectance anisotropy characterization of the land cover types (Mahtab et al., 2008). The anisotropic effects of solar illumination and sensor viewing geometry detected by multiangular instruments can provide improved and unique diagnostic information on the sensed surface (Diner et al., 2005, Schaepman et al., 2005). Changes produced in reflectance due to off-nadir viewing are dependent on the geometry of data acquisition, the wavelength and on the vegetation type under study (Strub et al., 2003, Goodin et al., 2004). Furthermore, modifications in vegetation structure produce different patterns of reflectance anisotropy. As a result, multiangular data can be used also to predict accurately the vertical structure of forest canopies measured by lidar systems (Kimes et al., 2006). When coupled to nadir data, the anisotropic information may be useful to improve classification accuracy of vegetation, as demonstrated by several researchers (Sandmeier and Deering, 1999, Liesenberg et al., 2007, Su et al., 2007, Mahtab et al., 2008; references therein).
Among the available multiangular satellite sensors, the Compact High Resolution Imaging Spectrometer (CHRIS), on board the European Space Agency's Project for On Board Autonomy (PROBA) satellite launched in October 2001, provides a new and unique opportunity to evaluate orbital hyperspectral and multiangular data simultaneously for different applications (Barnsley et al., 2004). This small-satellite mission is a technology-proving experiment that provides pointing in both along-track and across-track directions to produce multiangular data. The hyperspectral instrument is capable of acquiring data in five operational modes with different combinations of spatial resolution and spectral band sets. In Mode 1, the images are acquired with nominal spatial resolution of 36 m, in 62 bands (410–1000 nm) and in 5 view angles due to the pointability of the platform (Barnsley et al., 2004). However, in spite of being considered the first full spectrodirectional spaceborne instrument, only a small number of investigations have tested the potential of CHRIS/PROBA data for different applications. First results over land and water bodies were presented by Guanter et al. (2005).
In this study, the potential of hyperspectral and multiangular CHRIS/PROBA data (Mode 1) collected over the Amazonian forest was evaluated for the discrimination between primary forest and three stages of secondary succession (initial, intermediate and advanced) after deforestation. Several deforested areas in Amazônia have reverted to secondary succession, which increases the interest to map this class because of its importance to carbon cycle and ecosystem changes (Foody et al., 1996, Lucas et al., 2002, Lu et al., 2003, Vieira et al., 2003). Discrimination between successional stages is usually difficult because of the smooth transition between them, the heterogeneity of their vegetation structure, the land use history before the abandonment of productive activity and the frequency of cutting or burning events (Lu et al., 2003, Vieira et al., 2003). Despite the limited geographic coverage for mapping purposes (swath width of 14 km), the experimental CHRIS/PROBA provides an opportunity to evaluate the impact of the anisotropy on the discrimination of secondary succession.
Section snippets
Study area
The study area (14 km × 14 km) is located in the Tapajós National Forest (Pará state, Amazônia) and its surroundings between 03°05′S/54°54′W and 03°11′S/55°02′W. In the Köeppen classification, the climate of the study area is AmW. The rainy season occurs between December and May and the average values of annual precipitation and temperature are 2272 mm and 26 °C, respectively (Santos et al., 2003, Aragão et al., 2005). The area is relatively flat with minimum and maximum altitude values of 65 m and 200
Image acquisition and pre-processing
CHRIS/PROBA images were acquired on 23 June 2006 (dry season and clear sky conditions) close to the principal plane with a solar zenith angle of 33°. Data were collected in the visible and near infrared wavelengths (410–1000 nm) in 62 bands (Mode 1) with a variable nominal full-width-half-maximum between 5 nm and 10 nm. The ground instantaneous field of view (GIFOV) was 40 m at nadir and the swath width 14 km. By pointing the platform within a short time period (less than 2.5 min), data were acquired
Spectral-angular variations of primary and secondary succession
Variations in CHRIS-derived surface reflectance anisotropy between primary forest and the initial (SS1), intermediate (SS2) and advanced (SS3) secondary succession after deforestation are shown in Fig. 1 for two extreme view angles (−36° and +36°). Reflectance increased in the 410–1000 nm range from the forward (Fig. 1a) to the backward (Fig. 1b) scattering direction due to the predominance of sunlit canopy components at +36° view angle. For a given view angle, the near infrared reflectance
Conclusions
In comparison with the single view angle approach (nadir), the CHRIS/PROBA-derived multiangular approach produced an overall discrimination improvement of secondary succession in the study area, especially between primary forest and old stages of vegetation regrowth (SS3 and SS2). These classes have been reported in the literature as the most difficult to be discriminated. The largest anisotropic behavior was displayed by primary forest and SS3 in the backward scattering direction in which
Acknowledgements
The authors are grateful to European Space Agency (ESA) for providing CHRIS/PROBA data through selection of Category 1 PROBA Proposal 3774, and to Luiz Eduardo Aragão and Fernando Espírito Santo for providing useful field inventory data. Thanks are also due to the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), to Waldir Paradella and Cleber de Oliveira for helpful assistance in geometric correction, and to the anonymous reviewers for very useful suggestions.
References (20)
- et al.
Landscape pattern and spatial variability of leaf area index in Eastern Amazonia
Forest Ecol. Manage.
(2005) - et al.
The value of multiangle measurements for retrieving structurally and radiatively consistent properties of clouds, aerosols, and surfaces
Remote Sens. Environ.
(2005) - et al.
Identifying terrestrial carbon sinks: classification of successional stages in regenerating tropical forest from Landsat TM data
Remote Sens. Environ.
(1996) - et al.
Predicting lidar measured forest vertical structure from multi-angle spectral data
Remote Sens. Environ.
(2006) - et al.
Variations in reflectance with seasonality and viewing geometry: implications for classification of Brazilian savanna physiognomies with MISR/Terra data
Remote Sens. Environ.
(2007) - et al.
Classification of successional forest stages in the Brazilian Amazon basin
Forest Ecol. Manage.
(2003) - et al.
Structure analysis and classification of Boreal forests using airborne hyperspectral BRDF data from ASAS
Remote Sens. Environ.
(1999) - et al.
Airborne P-band SAR applied to the aboveground biomass studies in the Brazilian tropical rainforest
Remote Sens. Environ.
(2003) - et al.
Spectrodirectional remote sensing for the improved estimation of biophysical and chemical variables: two case studies
Int. J. Appl. Earth Observ. Geoinform.
(2005) - et al.
Classifying successional forests using Landsat spectral properties and ecological characteristics in eastern Amazônia
Remote Sens. Environ.
(2003)
Cited by (34)
Monitoring restored tropical forest diversity and structure through UAV-borne hyperspectral and lidar fusion
2021, Remote Sensing of EnvironmentMapping forest successional stages in the Brazilian Amazon using forest heights derived from TanDEM-X SAR interferometry
2019, Remote Sensing of EnvironmentCitation Excerpt :The Tapajós National Forest region has been studied for decades as a supersite representative of the dense rainforest of the eastern Brazilian Amazon (Santos et al., 2003; Espírito-Santo et al., 2005; Espírito-Santo et al., 2014; Bispo et al., 2016; Silva et al., 2016; Cassol et al., 2019; Pôssa et al., 2018; Wiederkehr et al., 2019). Many studies have explored the successional forest stages of this area considering field measurements (Lu, 2005; Silva et al., 2016; Cassol et al., 2019) and also remote sensing datasets (Santos et al., 2003; Espírito-Santo et al., 2005; Galvão et al., 2009; Cassol et al., 2019; Pôssa et al., 2018; Wiederkehr et al., 2019). Table 4 with the local field data description used in this study is compatible with Table 1, which is based on a review by Lu (2005) using field inventory measurements also from other areas.
Mapping land cover change in northern Brazil with limited training data
2019, International Journal of Applied Earth Observation and GeoinformationCitation Excerpt :This allows a distinction to be made between deforestation as a removal of primary forest and agricultural management practices that cyclically remove vegetation cover with less ecological value. Secondary vegetation is dominated by intermediate to advanced stages of succession (Galvão et al., 2009). We did not include a class for annual crops in our analysis because it only covers a very small proportion of our study area according to the TerraClass datasets (Almeida et al., 2016).
Quantifying tropical dry forest succession in the Americas using CHRIS/PROBA
2014, Remote Sensing of EnvironmentCitation Excerpt :As such, there is a need to move from this binary classification, so proper carbon pools can be quantified (Sanchez-Azofeifa, Castro, Rivard, Kalascka, & Harriss, 2003). One key element on efforts to map tropical secondary forests using remote sensing is the use of “age since its last disturbance” as a key temporal variable (Galvao, Ponzoni, Liesenberg, & Santos, 2009; Helmer, Brown, & Cohen, 2000; Lu, Mause, Brondızio, & Moran, 2003; Saldarriaga, West, Tharp, & Uhl, 1988; Uhl et al., 1988; Vieira et al., 2003). This concept has been criticized in the literature since it fails to consider local environmental factors, such as soil, climate or previous land use (Castro, Sanchez-Azofeifa, & Rivard, 2003; Lucas et al., 2000).
Vegetational succession assessment in a fragment of the Brazilian Atlantic Forest
2023, Environmental Monitoring and AssessmentCharacterizing Biophysical Attributes in Tropical Secondary Forest Environments with Multitemporal Hyperspectral Images
2022, International Geoscience and Remote Sensing Symposium (IGARSS)