Carbon budget estimation in Central Amazonia: Successional forest modeling from remote sensing data

Abstract

The carbon budget resulting from the dynamics of forest vegetation was estimated spatially for a study region with intensive land use change in the Central Amazonia forest. Vegetation height was recovered from airborne SAR interferometry, and was used along with an established relationship between forest height and age for mapping the successional stages of vegetation. A map of forest ages could be generated and validated (age RMSE was 3.5 years). Biomass stocks and annual rates of increment in biomass could be attributed to the forest ages by a comprehensive growth model for forests in the study area. A conceptual model of land use change was developed for the study area that accounts for four different types of land use: primary forest, secondary forest, degraded forest and nonforest. The transition probabilities between those land use types were recovered from internal modeling of available data, from literature sources, and from large-scale remote sensing results. The land use change matrix, area-age densities of secondary forests, and a growth model, yield a spatialized estimate of the carbon budget. The committed emissions from annual land use change were computed. For the year 2000-2001 the carbon balance was negative, on an area of ca. 5700 ha, land use dynamics resulted in a release of approximately 16,000 t of carbon, mainly arising from the cutting of primary forest for agricultural purposes. The secondary forest carbon budget was almost balanced, and forest degradation was revealed less important. © 2004 Elsevier Inc. All rights reserved.