B51I-08
A new pan-tropical estimate of carbon loss in natural and managed forests in 2000-2012
Friday, 18 December 2015: 09:45
2004 (Moscone West)
Alexandra Tyukavina1, Alessandro Baccini2, Matthew Hansen3, Peter Potapov3, Stephen V. Stehman4, Richard A Houghton2, Alexander Krylov3, Svetlana Turubanova3 and Scott J Goetz5, (1)University of Maryland College Park, College Park, MD, United States, (2)Woods Hole Research Center, Falmouth, MA, United States, (3)University of Maryland, College Park, MD, United States, (4)SUNY College of Environmental Science and Forestry, Syracuse, NY, United States, (5)Woods Hole Research Ctr, Falmouth, MA, United States
Abstract:
Clearing of tropical forests, which includes semi-permanent conversion of forests to other land uses (deforestation) and more temporary forest disturbances, is a significant source of carbon emissions. The previous estimates of tropical forest carbon loss vary among studies due to the differences in definitions, methodologies and data inputs. The best currently available satellite-derived datasets, such as a 30-m forest cover loss map by Hansen et al. (2013), may be used to produce methodologically consistent carbon loss estimates for the entire tropical region, but forest cover loss area derived from maps is biased due to classification errors. In this study we produced an unbiased estimate of forest cover loss area from a validation sample, as suggested by good practice recommendations. Stratified random sampling was implemented with forest carbon stock strata defined based on Landsat-derived tree canopy cover, height, intactness (Potapov et al., 2008) and forest cover loss (Hansen et al., 2013). The largest difference between the sample-based and Hansen et al. (2013) forest loss area estimates occurred in humid tropical Africa. This result supports the earlier finding (Tyukavina et al., 2013) that Landsat-based forest cover loss maps may significantly underestimate loss area in regions with small-scale forest dynamics while performing well in regions with large industrial forest clearing, such as Brazil and Indonesia (where differences between sample-based and map estimates were within 10%). To produce final carbon loss estimates, sample-based forest loss area estimates for each stratum were related to GLAS-lidar derived forest biomass (Baccini et al., 2012). Our sample-based results distinguish gross losses of aboveground carbon from natural forests (0.59 PgC/yr), which include primary, mature secondary forests and natural woodlands, and from managed forests (0.43 PgC/yr), which include plantations, agroforestry systems and areas of subsistence agriculture. Latin America has the highest proportion of aboveground carbon loss coming from natural forest clearing (72%) among the continents. The total (above- and belowground) annual carbon loss from forest disturbances in the tropics is estimated to be 1.3 PgC/yr, which accounts for 13% of global year 2012 CO2 emissions (IPCC 2014).
