In quantifying the carbon budget of the Amazon region, temporal estimates of the extent and age of regenerating tropical forests are fundamental. However, retrieving such information from re mote-sensing data is difficult, largely because of spectral similarities between different successional stages and variations in the reflectance of forests following different pathways of regeneration. In this study, secondary-forest dynamics in Brazilian Amazonia were modeled for the 1978-2002 period to determine area and age on a grid basis. We modeled the area, age, and age class distribution of secondary forests using empirical relationships with the percentage of remaining primary forest, as determined from large-area remote-sensing campaigns (the Pathfinder and Prodes projects). The statistical models were calibrated using de tailed maps of secondary-forest age generated for seven sites in the Brazilian Legal Amazon. The area-age distribution was then specified from mean age by a distribution assumption. Over the period 1978-2002, secondary-forest area was shown to have increased from 29,000 to 161,000 km2 (that is, by a factor of 5). The mean age in creased from 4.4 to 4.8 years. We generated a time series of secondary-forest area fractions and suc cessional stages that provides wall-to-wall cover age of the Brazilian Amazon at a spatial resolution of 0.1 decimal degrees (approximately 11 km). Validation against reference data yielded root mean squared errors of 8% of the total area for estimate of secondary-forest area and 2.4 years for mean secondary-forest age. Using this approach, we provide the first published update on the state of secondary forests in Amazonia since the early 1990s and a time series of secondary-forest area over the 2 5-year period.
|Number of pages||15|
|Early online date||31 May 2006|
|Publication status||Published - 01 Jun 2006|
- area-age distribution
- forest age
- land-use model
- secondary forest