Optimization of Forest Big Data
Uploaded by
Victor HugoOptimization of Forest Big Data
Uploaded by
Victor HugoOptimization of Forest Big Data
Estimating tropical forest carbon stocks using 3D high-resolution
remote sensing for large scale mapping
Optimization of Forest Big Data
Estimating tropical forest carbon stocks using LiDAR remote sensing
This work is part of a larger project conducted by:
• Niro Higuchi, Ph.D.
INPA (the Brazilian National Institute of Amazonian Research), Brazil.
CADAF Project (Carbon Dynamics of Amazonian Forest)
It is being undertaken in partnership among several research groups, institutions,
and countries.
CADAF Project (Carbon Dynamics of Amazonian Forest)
This work is part of a larger project conducted by:
• Niro Higuchi, Ph.D.
INPA (the Brazilian National Institute of Amazonian Research), Brazil.
It is being undertaken in partnership among several research groups:
• INPE (The Brazilian National Institute for Space Research)
• JICA (Japan International Cooperation Agency)
• JST (the Japanese Scientific and Tecnology)
• University of California, Berkeley – USA.
• Stanford University, USA.
• UFAM (Federal University of Amazonas), Brazil.
•…
Title
OUR MAJOR RESEARCH CHALLENGE
• We are interested in estimating carbon stocks by means of extrapolation
and spatialization based on forest inventory using remote sensing LiDAR
(Light Detection and Ranging) technology…
… through the application of combinatorial optimization, graph theory,
image retrieval, pattern recognition, and machine learning techniques.
Motivation
IPCC (The Intergovernmental Panel on Climate Change)
“There is an urgent need to provide additional guidance on the design of forest
monitoring systems including issues such as forest inventory design, stratification,
sampling, pools, accuracy/uncertainty assessment and the combination of ground-based
inventories with remote sensing and modelling approaches.”
Amazon RainForest
Forests sequester and store more carbon than any other terrestrial
ecosystem and are an important natural brake on climate change.
The Amazon Rainforest is the world’s largest tropical rainforest:
5,500,000 square kilometers.
Title
The Amazon Rainforest is the world’s largest tropical rainforest:
5,500,000 square kilometres.
09 (nine) countries:
Brazil, Peru, Colombia, Venezuela,
Ecuador, Bolivia, Guyana, Suriname
and French Guiana.
The majority of the forest is
contained within Brazil, with 60% of the
forest.
The largest and most biodiverse tract
of tropical rainforest in the world, with
an estimated 390 billion individual
trees divided into 16,000 species.
Amazon RainForest
Huge and Complex
Area 5,500,000 km2 A twisted entanglement
of tree branches.
Process of Carbon Stocks Estimation
Where carbon is stored?
Carbon pools are all living biomass of trees and all
non-living biomass.
Biomass: leaves, twigs,
branches/trunk, and roots
Forest Inventory for Carbon Stock Estimation
What is Forest Inventory?
• Systematic collection of data and forest information
for analysis.
• Measurements: species, diameter at breast height
(DBH), height, site quality, age, etc.
The forest inventory can be accomplished in two
ways:
• Ground based inventory.
• Inventory with remote sensing and modelling.
Ground based Inventory
• Permanent plots (e.g. 20x125 meters).
• DBH (1.30 m), tree height and weight (fresh and Permanent Plots
dry mass).
• Representative trees: dominant (largest) and
emergent (tallest) trees.
• Allometric equations.
Biomass and Allometric Equations1
Allometric equations are
widely used as an indirect
means of estimating forest
biomass of an area.
Allometric equations are
mathematical models derived
of regression analysis.
Carbon
H2O
Biomass: leaves, twigs,
branches/trunk, and roots
[1] Higuchi et al., 1994; Overman et al., 1994; Santos, 1996; Higuchi et al., 1998.
Ground based Inventory
INPA/LMF's Project Experience
Temporary and Permanent Plots
Inventory Plots
Biomass and Carbon: Harvest Process
Drying Process
After releasing 40% of water,
out of the 60% of remaining dry
mass, 48% is carbon.
Carbon
H2 O
Biomass: leaves, twigs, branches/trunk, and roots
Allometric Models
1,30 m
DBH ≥ 20 cm
(Diameter at Breast Height)
Dominant trees
(larger diameter trees)
fresh weight = – 2,694 + 2,038(ln DBH) + 0,902(ln H)
Emergent trees
60% dry weight (48% carbon!) (tallest trees)
40% water
Inventory by Remote Sensing
• Satellite Remote Sensing
• Aerial Remote Sensing
LiDAR
Light Detection
And Ranging
Aerial
Remote
Sensing
LiDAR (Light Detection And Ranging)
AIRBONE LiDAR SYSTEMS
1.Laser Scanner.
2.Inertial Measurement Unit (IMU).
3.Global Positioning System (GPS).
Aerial LIDAR Octocopter UAV
UAV
Unmanned Aerial Vehicle
(low altitude)
LiDAR Point Clouds
Light Detection And Ranging
Georeferencing (3D) Laser Images
Digital Surface Model (DSM)
Digital Terrain Model (DTM)
Preliminary and Ongoing Results
• Empirical experiments applying LiDAR UAV technology, and
generating of models in the remote sensing lab using data
collected.
• Development of an algorithm to determine the average height of
trees in a permanent plot.
• Graph methods for three-dimensional shape reconstruction of the
forest from LiDar cloud points.
• Heuristic strategies for minimizing sample size for use in airborne
LiDAR-based forest inventory.
• Feature extraction from LiDAR images to identify an individual
tree / a tree canopy.
• Solving graph covering problem to get representative trees.
Issues to be addressed
• High cost (equipment/devices, flight hours, etc).
• Governmental barriers/prohibitions (aircraft overflights).
• Technological and scientific challenges (multidisciplinary problem,
large-scale, dynamic, unconventional spatial data).
• Few works with with advanced focus in Computer Science
approach (a good thing!).
IComp/UFAM Research Group
Eulanda Miranda, Ph.D.
Rosiane de Freitas, Ph.D. João Marcos Cavalcanti, Ph.D.
Pattern Recognition and
Combinatorial Optimization and Image and Information
Machine Learning
Graph Theory Retrieval
Denys Silveira
Rosana Gemaque Sócrates Lever Undergraduate Student
Ph.D. Student Master Student Computer Science
Main Research Partners
Niro Higuchi, Ph.D. Jeffrey Chambers, Ph.D.
Forest and carbon dynamics, Tropical forests and climate change
Tropical forest ecology, vegetation dynamics
Forest management
Carlos Henrique Celes
Ph.D. Student
Thank you!
Gracias!
Rosiane de Freitas
[email protected]
You might also like
- No ratings yetCombined effect of pulse density and grid cell size on predicting and mapping aboveground carbon in fast-growing Eucalyptus forest plantation using airborne LiDAR data Polymorphisms in PPARG and APOE: relationships with lipid profile of adolescents with cardiovascular risk factors 17 pages
