A influência dos parâmetros biofísicos da vegetação de mangue em regeneração no retroespelhamento de imagens radarsat-2 multipolarizada na Amazônia, Brasil

Abstract

The main goal of this work is to understand the relationship between backscattering (σ°, β° e γ) of a multipolarized Radarsat-2 Fine Beam band C image with biophysical parameters of mangrove vegetation in regeneration. The study site was the Bragança peninsula (northeast of Pará State) far from the Amazon River mouth about 380 km (southeast). 30 years ago it was built a highway causing disturbances in the hydrological regimes and vegetation death, then it was deforested and currently the natural regeneration is incipient. Field data was taken in 17 stands of 10 square meters and it was delimited and positioned with a DGPS (Differential Global Positioning System) and a total station. Structural characterization of the sampling units was made with the following field data: CBH (circumference at breast height), height and specie in 3090 specimen and then it were estimated values of DBH (diameter at breast height) and BA (basal area). A cluster analysis of the sample units has defined four groups of vegetation in different regeneration stages: pioneer stage, initial regeneration, intermediate regeneration and advanced regeneration. Fromard et al. (1998) equation was used to biomass estimation. It was used a SLC (Single Look Complex) multipolarized image (Radarsat-2) from June 11th, 2010. VIMAGE/Focus/PCI software was used to retrieve backscattering values of SAR (Synthetic Aperture Radar) image, using as reference the limits of the parcels. Statistical models of simple and multiple regressions were used in the analysis of the relationship between vegetation structure and backscattering values of the SAR image. Results shown that linear sigma backscattering with cross-polarization (VH) had the strongest relationship with all vegetation structure analyzed. Although the function of multiple linear regression with HH, VH and VV polarization got the best adjustments with the biophysical parameters mean height, DBH and Biomass (R2 values were respectively 0,81, 0,79 and 0,79). Backscattering values of the image with adjusted functions were used to generate maps of these three biophysical variables and it was coherence with field data, mainly mean height and DBH maps. Biomass map got a low amplitude variation suggesting saturation of the sign, in mangroves in regeneration, less than indicated in the literature to band C in homogeneous mangrove areas.