BIOMASS ESTIMATION OF FOREST AND SAVANNA TRANSITION VEGETATION ZONE BY JERS-1 AND SIR-C BACKSCATTER DATA

摘要

The inventory and monitoring of transition zones between tropical rain forest and savanna formations in Brazilian Amazonia are an essential step for an accurate analysis of global change and biodiversity studies. The objective of this study is to analyse the empirical behaviour of the biomass from forest/savanna transition zones referring to backscatter signals of JERS-1 and SIR-C images. The complementary objective is to discriminate among vegetation types and to map the distribution of its' biomass using both sensors. The area under study is located in Mato Grosso State (Brazil), at the border with Rondonia State, representing a contact zone where typical botanical species of both formations are intermingled. The SIR-C and JERS-1 images were georeferenced, based on a bilinear method and the pixel resampling was made to get a spatial resolution of 25 m. Comments were made related to the physiognomic-structural details of vegetation types and spatial distribution of backscatter at primary forest, tall woodland, savanna woodland, tree and/or shrub savanna. The relationship between backscatter and biomass values is based on the analysis of their adequacy into a regression model where these variables were adjusted. Using the logarithm regression model, the results show that SIR-C data present the highest determination coefficient, specially the LHV band (r~(2) velence 0.8275) when compared to the L_(HH) band of JERS-1 data (r~(2) velence 0.6791). Considering the same polarization (L_(HH)), the SIR-C data (r~(2) velence 0.7406) is also better than JERS-1 data. The methodological approach used in this study can be very useful to determine the dynamics of the biomass, taking into account the settlement of humans that occurred in the contact zones of forest and savanna in Amazonia.