J.R. dos Santos

Land Use and Land Cover Mapping in the Brazilian Amazon Using Polarimetric Airborne P-Band SAR Data

In September 2000, an airborne synthetic aperture radar (SAR) mission acquired unprecedented full polarimetric P-band data over the Tapajos National Forest (Para State), which is an area in the Brazilian Amazon which has been continuously monitored in the last three decades. Eight land use/cover classes were identified, namely, primary forest, regeneration older than 25 years, regeneration between 12 and 25 years, regeneration between 6 and 12 years, regeneration younger than six years, crops/pasture, bare soil, and floodplain (FP). The objective of this paper is to analyze the potential of full polarimetric P-band data in distinguishing different land use/cover classes with a minimum established Kappa value of 75%, using the latest development on SAR statistical characterization. The iterated conditional mode (ICM) contextual classifier was applied to amplitude, intensity images, biomass index, and some polarimetric parameters (entropy, alpha angle, and anisotropy) extracted from the polarimetric P-band data. As the accuracy obtained for eight classes was not acceptable, another two sets, with five and four classes, were formed by the combination of the previous ones. They were defined by confusion matrix analysis and by the graphical analysis of average backscatter values, entropy, [alpha] angle, and anisotropy images and by the H/alpha plans of the land use samples. The classification accuracy with four classes (three levels of biomass plus FP) was then considered acceptable with a Kappa value of 76.81%, using the ICM classification with the adequate bivariate distribution for the HV and VV channels.

Power spectrum analysis of SAR data for spatial forest characterization in Amazonia

Power spectrum analysis was used for the analysis of spatial forest features from airborne X‐band synthetic aperture radar (SAR) data in the Brazilian Amazon. Spectral estimates were arrived at empirically by periodograms and correlograms, and from autoregressive moving‐average (ARMA) models. The spectral estimates derived from SAR data were validated by those derived from ground data with locational match. The results obtained by ARMA modelling revealed particularly good correspondence between remote sensing and reference data: repeating patterns at pixel level could be detected in the images. These patterns were shown to arise from canopy structure and distances between major tree individuals; and thus allowed the extraction of parameters of spatial forest structure, particularly of forest density. The method was applied to an example area of primary tropical forest, and its spatial patterns were modelled.

Mapping and monitoring deforestation areas in Amazon region using semi-automatic classification of Landsat Thematic Mapper images

The INPEs operational project (PRODES) to estimate annual gross deforestation in Amazon region based on manual analysis of 229 TM images faces several problems during the interpretation process (variable scales of different scenes, closing polygons in the interpretation maps due to complexity of deforestation pattern). Thus, the availability of results in a digital format has been restricted. This authors propose an approach to map and monitor deforested areas in the Amazon using digital analysis of TM/Landsat. This methodology will automate the PRODES manual interpretation tasks and will build a GIS database. This approach was developed and validated using TM image Path 231/067 (1997, 1998, and 1999) over Rondonia. The original TM bands were converted to vegetation, soil, and shade fraction images applying a linear mixing model. The selected fraction image was segmented using a region growing algorithm, classified using a per region clustering algorithm and the results were manually edited to generate the final map. Results showed 10,252 km/sup 2/ of deforestation up to 1997; increments in the deforested area for 1998 and 1999 were 695 and 388 km/sup 2/, respectively. A total of 1,149 km/sup 2/ was burned in 1998 (only 16% in recent clear cut areas). The proposed methodology is feasible and very useful for global studies using fine resolution satellite data such as Landsat TM.

Analysis of airborne SAR data (L-band) for discrimination land use/land cover types in the Brazilian Amazon region

The objective of this paper is to show the potential of multi-polarized mosaic-images from a MAPSAR (L-band) simulation campaign performed in the Amazon region (test site Tapajos) in order to discriminate land use/land cover classes. An Enhanced-Frost filter was used and the thematic discrimination was done by an algorithm for attribute extraction based on Bhattacharya distance. A comparison was made among the radiometric aspects of the SAR mosaic for 10 thematic classes obtained, converting these B-distances to JM distance values. This allowed to define which individual or multiple polarizations are more appropriate for the identification of thematic classes.

Identification of the tropical forest in Brazilian Amazon based on the DEM difference from P and X bands interferometric data

In this paper the difference between digital elevation models, derived from P and X bands interferometric data, is used as a main information to identify land cover classes. The radar data used in this work were collected on September of 2000 over Tapajos National Forest, which is a region of Brazilian Amazon, Para State. The SAR images were acquired from an airborne polarimetric system, AeS-1, that could provide P and X bands interferometric data. During the radar mission ground survey was carried out, and the georeferenced information about the forest typology were acquired, and used as a support for the thematic identification and calibration of the remoted sensing data. The X-band DEM was generated using one-pass interferometric data and the P-band DEM was generated using two-pass interferometric data. The grid of the DEMs has a spatial resolution of 2.5 meters. Images from P and X bands and coherence maps were also used in order to improve the classification. Supervised and unsupervised classifications techniques are used and their results are shown.

Inventory of forest biomass in Brazilian Amazon: a local approach using airborne P-band SAR data

The objective of this study is to explore the use of airborne P-band SAR polarimetric data, to stratify biomass by primary and secondary vegetation typology. To ensure that different landscapes of Amazon upland forest are represented, a test-site located in the lower Rio Tapajos region, Pari State, was selected. The backscatter signals derived from the complex image of the P-band SAR were correlated with field data obtained from a forest inventory, for different physiognomic-structural aspects of the tropical rainforest The estimation of above-ground biomass for these forest types was modeled by DBH and total height measurements, including the use of general allometric equations. Statistical regression models were applied to establish the relationship between biomass and radar data at HH, HV and VV polarization. The overall objective of this P-band experiment is to improve the regional monitoring process of biomass dynamics as well as landscape changes, due to human action.

Potential use of JERS-1 data for biomass estimation of tropical forest environments in Brazilian Amazonia

The objective of this study is to show the importance and limitations of JERS-1 data as a tool for inventory and monitoring vegetation types and the respective biomass of tropical environments. Several study areas (Acre, Amazonas, Mato Grosso, Para and Roraima States) were chosen to collect information about physiognomic and structural characteristics of the vegetation types and their topological aspects at JERS-1 image. The behavior of backscatter signals derived from JERS-1, which were correlated with the variability of biomass of the different formations, is presented. Statistical tests with regression models are used to understand the relation between field survey and microwave data and to verify the possibility to generate the cartography of local biomass for intervals of classes. The general variations of intra-class structure must be considered as a limitation to obtain a highly significant performance on the use of L-band/HH polarization.

SAR Interferometric Approaches for the Analysis of Structural Forest Parameters: State of the Art and Perspectives for Brazilian Studies

This paper presents three practical examples of airborne InSAR data application to improve the knowledge of forest structures. Two experiments were done in the Amazon tropical forest to study the spatial distribution of VLTs in the primary forest using LM filtering and a series of Markov processes and others, to map and model the estimation of biomass variations in primary and secondary forests. The third experiment refers to the relation of SAR data and the volumetric configuration of Eucalyptus sp. stands. The advances on the analysis of PolInSAR data are very helpful to increase, in the near future, the regional inventorying of land cover changes in the Brazilian territory.

Regrowth biomass estimation in the amazon using JERS-1/RADARSAT SAR composites

Synthetic Aperture Radar (SAR) is known to have a response that is directly related to the amount of living material that it interacts with. It is this property that our research seeks to exploit in order to better understand carbon dynamics in the Amazon. The vegetation density causes the radar response to saturate such that vegetation that is more dense than some threshold is indistinguishable from each other. However, the areas of regrowth are likely to have a low enough biomass during the first 10 years of regrowth to be accurately assessed using radar. Our efforts involve obtaining appropriate pairs of radar images at L and C bands from different sites and for both seasons. These data are then orthorectified to allow accurate calibration and incidence angle correction. The seasonality of the data is used to deal with the moisture sensitivity of the data, and the different frequency data is used to help classify the data into several classes for use in class-specific biomass estimates. We have chosen 2 sites in Brazil for our study. we use the JERS-1 (L-band) and RADARSAT (C-band) data to create a 2-channel composite. These data are then classified into the following classes: flat area (water, bare soil), short vegetation, regrowth, and trees. We report on the accuracy of both our classification and biomass estimation efforts.

Biomass change in disturbed, secondary, and primary tropical forests from TanDEM-X

The time variation of phase height from interferometric SAR (InSAR) from TanDEM-X is shown for 3 years, in Tapajos National Forest, Brazil. Its RMS, for one secondary stand, about a model linear in time is 0.5 m. This RMS is compared to that for 30 stands at one epoch. The single-epoch RMS for a model linear in mean field height is 2.2 m. It is suggested that the improved performance of the temporal variation may be due to errors in finding the phase height of the ground, which is necessary for single-epoch estimation, but not needed for the “change” measurement. Pending further fieldwork, a tentative conversion of 20 Mg/ha/yr corresponding to 1 m/yr, is proposed. Abrupt discontinuities, as well as phase height rates as a function of stand age/aboveground biomass, are discussed.