Geostatistical techniques for estimating aboveground biomass in eastern Amazonia

The Amazon plays a crucial role in global environmental debates due to its vital ecosystem services, including the carbon stock in its vegetation. Given the challenges of collecting field data in remote areas, remote sensing products such as those provided by the Global Ecosystem Dynamics Investigation (GEDI) mission are a valuable alternative for estimating aboveground biomass and carbon stocks, particularly in regions of high conservation interest. This study aimed to map aboveground biomass (AGB) and estimate carbon stocks in the Carajás Mosaic, a set of protected areas in eastern Amazonia, using geostatistical interpolation methods on GEDI-derived AGB data. We tested four methods: inverse distance weighting, ordinary kriging, regression kriging, and cokriging. Vegetation and terrain indices were evaluated as auxiliary variables. Results revealed high spatial variability in AGB, with significant correlations between AGB and spectral and terrain variables. Among the methods, regression kriging and cokriging showed a good spatial dependence structure, with cokriging providing the most accurate estimates. Overall, the results enabled a precise analysis of AGB estimates in these protected areas, providing insights into carbon distribution and emphasizing the importance of combining geostatistics and remote sensing for effective forest management and conservation planning.