Biomass estimated through IKONOS II satellite of the upland Dense Ombrophilous Forest in Central Amazon

Abstract

The implications of the reduction of tropical forest in the carbon cycle has created a demand for development of non destructive methods to biomass estimates a key component in modeling biogeochemical cycles. The research had as objective to estimate indirectly the above ground biomass (AGB) of a pristine forest, correlating field data to IKONOS II satellite data. Biomass was obtained by indirect method and the vegetation analysis by means of horizontal structure. The original bands, vegetation index like ARVI, NDVI, MSAVI and R43, PCA and Tasseled Cap transforms and Spectral Mixing Linear Model were used in modeling. The integration of ground truth and satellite data was performed by statistical procedure, by the application of linear multiple regression using STEPWISE method. For the spatialization, the minimum and the maximum values of the confidence interval for biomass estimates provided by the model were adopted as classes. The confidence interval was 343,05 ± 14,90 Mg. ha^-1 for 95% probability. The highest correlations between biomass and digital variables ARVI, SLMM_VEG and NDVI are negative. Modeling demonstrated that the variables R43, NDVI, TCap3 and TCap4 were the best correlation with biomass data.