@InProceedings{KellerCFPLDSO:2022:FoDeRa,
author = "Keller, Michael Maier and Csillik, Ovidiu and Ferraz, Antonio and
Pinag{\'e}, Ekena Rangel and Longo, Marcos and Duffy, Paul and
Saatchi, Sassan S. and Ometto, Jean Pierre Henry Balbaud",
affiliation = "{US Forest Service San Juan} and {NASA Jet Propulsion Laboratory}
and {NASA Jet Propulsion Laboratory} and {Oregon State University}
and {Lawrence Berkeley National Laboratory} and {Neptune and
Company} and JPL and {Instituto Nacional de Pesquisas Espaciais
(INPE)}",
title = "Forest degradation rates and carbon changes in the Brazilian Arc
of Deforestation using repeated airborne lidar",
year = "2022",
organization = "AGU Fall Meeting",
publisher = "AGU",
abstract = "The Brazilian Amazon is a hotspot of deforestation and forest
degradation caused by logging, fire, and deforestation-associated
fragmentation and edge effects. While the extent of deforestation
and associated carbon losses are relatively well known, the
quantification of the carbon losses caused by degradation and the
carbon gained by recovery of degraded forest ranges widely. We
present a detailed analysis of forest changes and associated
carbon gains and losses by using repeated randomized airborne
lidar surveys for 2016 and 2017 over 102 different transects
covering more than 50,000 ha throughout the Brazilian Arc of
Deforestation . We directly measured changes in canopy height and
used previously calibrated allometric equations to estimate
aboveground carbon changes. After gridding the surveyed area to 50
m x 50 m, we found that 21.6% of the area analyzed suffered losses
in canopy height that exceeded 0.5 m, while only 16.3% of the area
had canopy height recovery higher than 0.5 m. This translates to
an annual carbon loss of 102.8 GgC, while carbon gained through
forest regrowth was 33.4 GgC. Canopy height losses that exceeded 5
m accounted for 6.1% of the loss area identified but were
responsible for 28.3% of the total aboveground carbon loss. When
separated according to legally protected status, carbon changes on
loss areas averaged -7.1 ± 7.6 (standard deviation) MgC/ha-y
inside indigenous territories, -9.8 ± 13.0 MgC/ha-y within
conservation units and -10.1 ± 12.1 MgC/ha-y outside the two
protected categories. Carbon changes in gain areas averaged 4.0 ±
1.8 MgC/ha-y with no discernible differences among the three
categories. To attribute carbon losses to different degradation
drivers, we trained a machine learning model based on lidar point
cloud metrics and visual interpretation of high resolution
satellite imagery to differentiate between multiple types of
deforestation and forest degradation (e.g. logging, fire).
Extrapolating the results to the extent of the Arc of
Deforestation represented by our randomized airborne campaigns, we
find that forest degradation would account for a substantial
portion of Brazilian carbon emissions were it considered in
national budgets. Our study presents one of the first large-scale
quantifications of carbon losses due to forest degradation from
logging and fire.",
conference-location = "Chicago, IL",
conference-year = "12-16 Dec. 2022",
urlaccessdate = "28 abr. 2024"
}