LiDAR

The High-resolution Carbon Geography of Perú
for Forest Conservation and Management
Gregory P. Asner
Carnegie Institution for Science
With special thanks to MINAM-DGOT, Wake Forest University,
and the Peruvian Air Force
COP-20
Lima, Perú 2014
2014
State-of-the-Science
The High-resolution
Carbon Geography of
Perú
128,546,068 hectares
(321 million acres)
at 1-hectare resolution with
uncertainty reported for every
hectare throughout
the country
The Value of High Spatial Resolution Mapping
• Resolves more activity at finer scales (i.e.
degradation, regeneration) than does low
resolution mapping
• Creates a more direct bridge to field
inventory, which leads to better plot
network development and which feeds
back to make better maps
• Engages stakeholders at all scales (single
hectare owners to multi-million hectare
government agencies)
The Critical Hectare
• 10,000 m2 or about 2.5 acres
• The most common unit of land
management, ownership and tenure
in the world.
• People can intuitively understand
something about as large as a hectare,
and not often more
• When aggregated to the hectarescale, studies show that careful fieldbased estimates of biomass converge
on true forest biomass values with 1015% uncertainty.
• But a tropical forest can easily contain
> 600 trees per hectare (with
diameters > 10 cm). Very hard to
(re)measure on the ground!
Limits of Field Inventory
• Only valid at large scale if properly
arranged (random, stratified
random, systematic-aligned, etc)
• Must constantly deal with
inaccessibility, remoteness,
standardization = expensive
• Hard to maintain over time
• Hard to engage stakeholders on-theground with plot inventory since it
isn’t spatially explicit
A Different Approach
Three components:
•
Automated satellite analysis of
forest cover, deforestation,
degradation
•
New science and technology for
high-resolution carbon mapping
from the air
•
Long-term emissions monitoring at
high resolution
Asner 2009 (Environmental Research Letters)
A way to massively sample forest structure
LiDAR: Light Detection and Ranging
Light Detection and Ranging
(LiDAR)
Carnegie Airborne Observatory
LiDAR Easily
Sees Structural Change
LiDAR Easily
Sees Structural Change
LiDAR Easily
Sees Structural Change
Available Measurements of Forests from LiDAR
Field-estimated (Plot) Carbon Stock
LiDAR predictably calibrates to
field-estimated carbon stock
Airborne LiDAR Canopy Profile
Asner et al. (2012): A Universal Airborne LiDAR Approach for Tropical Forest Carbon Mapping
Tactical Use of Field Plots for LiDAR Calibration
LiDAR-based forest biomass estimation is statistically
indistinguishable from field-based biomass estimation
Asner et al. (2014): Mapping tropical forest carbon: calibrating plot estimates to a simple LiDAR metric
To save time & money, we build geophysical
relationships between satellite and LiDAR data
Geostatistical Modeling
MINAM-DGOT Deforestation Mapping
Training Satellite Data with Airborne LiDAR: Panama example
Geostatistical Model Updated
as Aircraft Collects Data
CAO LiDAR
(real-time carbon stocks in red)
Deploy Aircraft & Sample Country Until Statistical Needs Met
> 10.0%
5.0-10.0%
2.0-5.0%
1.0-2.0%
0.5-1.0%
5,000 hectares
0r n = 5,000!
The High-resolution
Carbon Geography
of Perú
128,546,068 hectares
(321 million acres)
at 1-hectare resolution with
uncertainty reported for
every hectare throughout
the country
Years of work
More Validation
One day of flight
Carbon Stocks in Every Hectare of Perú
Uncertainty in Every Hectare of Perú
Distribution of Carbon Stocks
throughout Peru’s forests
Natural Effects of Elevation on Aboveground
Carbon Stocks
Carbon Stocks by Jurisdiction
Peru’s Balance Sheet
Peru’s Balance Sheet
Carbon Benefits of Expanding Protected Areas
Gross Carbon Emissions from
Deforestation and Degradation
Madre de Dios
2000
2013
 1 Tg = 1 million metric tonnes
 Carbon emissions from forest disturbance average 47% of deforestation.
Gross Carbon Emissions Balance Sheet from
Deforestation and Degradation
Madre de Dios
Land Use
Total Area
(ha)
Regional Carbon
Flux (Tg C)
Mean Flux Density
(Mg C ha-1)
S.D. Flux
Density
(Mg C ha-1)
Gold mining
3,203
(0.238)
74.3
15.9
Forest disturbances
17,740
(1.360)
76.7
20.3
Deforestation
43,933
(2.930)
66.7
19.1
Secondary regrowth
63,173
1.932
30.6
16.7
Net flux
(2.598)
Many Sub-Projects
Along the Way
High-resolution Carbon
Mapping Costs
Rapid Increase in Coverage,
and Decline in Costs
Envir Res Letters 2009
PNAS 2010
Biogeosciences 2011
Frontiers in Ecology and Envir 2011
Remote Sensing of Envir 2011
Carbon Balance and Management 2012
Biogeosciences 2012
Carbon Balance and Management 2013
PLOS One 2014
Ecological Applications 2014
Carbon Balance 2014
PNAS 2014
Reaching the Large-Area, High-resolution Scale
at Low Cost
Additional
Carnegie Resources
Capacity Building
 Satellite: 841 organizations in 127 countries
 LiDAR: 33 organizations in 9 countries
 Field inventory: 12 organizations in 5 countries
CAO Maps Online
http://cao.carnegiescience.edu/maps
CAO LiDAR Training and Data Provision in California
Colombia: IDEAM, IGAC, FAC
Perú: MINAM, WWF-Perú, FAP
Ecuador: IGM
CLASlite Forest Monitoring Software
• Software to automatically analyze
imagery from 8 different satellites
• Maps forest cover, deforestation, and
forest degradation
• Heritage of high accuracy mapping
results from hundreds of national and
sub-national studies worldwide
• Today, 841 organizations in 127
countries with active CLASlite licenses.
Top users are government:
IDEAM (Colombia): 27
IPAM (Brazil): 27
ANAM (Panama): 26
MINAM (Peru): 25
INPA (Brazil): 18
CLASlite Iterative
Science-to-Capacity Building Process
CAPACITY BUILDING
Broad capacity building through in-country workshops
TECHNICAL DEVELOPMENT
Version 1.0 –
Version 2.3
Early Workshops in Multiple Countries
Joint project implementation
through institutional partnership
and collaboration
MINAM-DGOT
Version 3.2
http://claslite.carnegiescience.edu
https://class.claslite.carnegiescience.edu
John D. and Catherine T. MacArthur Foundation
Gordon and Betty Moore Foundation
Avatar Alliance Foundation
Grantham Fndn for the Protection of the Environment
Mary Anne Nyburg Baker and G. Leonard Baker Jr.
William Hearst III
http://cao.carnegiescience.edu
http://claslite.carnegiescience.edu
Extra slides
Geostatistical Modeling = Statistical + Geophysical
Modeling
Plot-scale LiDAR Validation
(using local wood density)
Regional-scale Validation
(using regional basal-area-weighted wood density)