Lasers, Satellites, and Drones: An overview of remote ...€¦ · 10/06/2014 · estimates from satellite imagery, with those from LiDAR, and with estimates produced from combining
Post on 28-May-2020
3 Views
Preview:
Transcript
Lasers, Satellites, and Drones:
An overview of remote sensing research
Growing Confidence in Future Forestry
Building on the achievements of FFR programme.
Further developing expertise in LiDAR and image
analysis.
Tools to shift the resolution of forest assessment
from the stand to the sub-stand level - Precision
forestry
Development of an integrated remote
sensing platform
Overview of Remote Sensing Research Phenotyping Platform
• Concepts
• Individual tree analysis
• Terrestrial LiDAR
• UAVs
LiDAR and Forest Inventory
• k-Nearest Neighbour
• Multiple datasets
• Terrestrial LiDAR
LiDAR Economic Analysis
Leaf Area Index Research
Satellite Imagery Research
What is phenotyping?
Phenotype
The interaction between an organism’s
genes and its environment.
Composite of an organism’s observable
traits.
Phenotyping is the comprehensive
assessment of complex plant traits such
as growth, architecture, and wood
properties.
What is the phenotyping platform?
A suite of technologies
and techniques that
will allow us to
phenotype trees
remotely.
What is the phenotyping platform?
Sensors
Aerial LiDAR
Terrestrial LiDAR
Panchromatic imagery
Multi/hyper spectral
imagery
Phenotyping platform development
Individual tree genetics trials
Aerial LiDAR coverage;
Known genetic composition;
Conventional phenotyping for
validation.
Tree level analysis of aerial and
terrestrial LiDAR
Develop methods to augment or
replace conventional measures in
trials
Phenotyping platform development
Model and extract key metrics
• Height;
• Tree position;
• DBH;
• Tree architecture;
• Stiffness;
• Foliar health.
*Objective measures
Aerial LiDAR – Tree Level Analysis
2D and 3D crown metrics
Direct measurement of tree
metrics in the upper canopy
Models linking crown metrics to
variables of interest to
phenotyping
Further Objectives
Improve inventory estimates of
tree size and quality
Investigate effect of irregular
spacing on tree quality
What can terrestrial LiDAR offer?
Detailed description of the stems and
lower canopy of subject trees.
A means of co-locating trees.
But…
Significant research effort required to
develop tools to extract tree metrics
from the terrestrial point cloud.
Terrestrial LiDAR project work stream
underway to address this.
FARO Focus 3D
Tripod Mounted/
Hemispheric Scanner
5kg
$100,000
Collects RGB
imagery
Accuracy (mm)
Range 150 m+
ZEB1 (Zebedee)
Portable, handheld scanner
Lightweight ~1kg
$25,000
cm accuracy
Range = 30m
Continuous Scanning
Terrestrial LiDAR for Phenotyping
Development Steps
• Integrate terrestrial with aerial
LiDAR.
• Develop procedures for phenotyping.
Outputs
• Tools that feed into the phenotyping
platform delivering stem metrics.
• Science publication outputs.
Why Terrestrial LiDAR Now?
3D point clouds are the future of
measurement and visualisation
Google scale investment
announced
Cheaper and better technology
Build expertise and understanding
be ready to deliver this technology
to our industry
What can UAVs offer?
Low cost data collection and
regular returns to small areas.
Cameras or light weight LiDAR
units
Numerous additional forestry
applications.
Use currently restricted by
battery life and CAA
regulations.
RIEGL VUX-1
Unmanned aerial vehicles (UAVs)
NOT a toy Up to 62 points per m2
Horizontal accuracy
0.34m
Tree height SD 0.15m
“Development of a UAV-LiDAR System with Application to Forest Inventory”
LiDAR and Forest Inventory - Background
Aerial LiDAR provides auxiliary information
that can be useful for forest inventory.
Better Precision
Fewer plots - $ saving.
Productivity surfaces – Better resolution
information.
Yield estimates for arbitrary AOIs defined after
sampling:
• Felling coupes;
• Riparian areas;
• New stands.
LiDAR and Forest Inventory - Background
A LiDAR based inventory system must
provide:
• Yield tables including log product
estimates;
• Sampling error for AOIs;
• Use current software and models.
Through FFR k-NN estimation has been
delivered as an approach that meets
these demands.
LiDAR and Forest Inventory Where to now?
Is k-NN suitable for PHI?
Use guidelines
Neighbour selection
techniques
Alternative patch level
LiDAR metrics.
Sample size and sampling design
Optimisation of k (model property)
Alternative approaches to sampling
error
LiDAR and Forest Inventory Integrating additional datasets?
Satellite imagery.
Multiple LiDAR campaigns:
Spatially and temporally separated.
Different scanner settings and hardware.
Partial coverage.
Incorporating pre-existing inventory
information.
Individual tree metrics.
Is terrestrial LiDAR useful for forest inventory?
Objectives
Validate the technology and the best currently
available solutions as a forest measurement tool
in NZ conditions.
• Scanners mentioned previously;
• Treemetrics Autostem algorithm.
Is terrestrial LiDAR useful for forest inventory?
Methodology
Scan recently measured conventional forest
inventory plots.
Experimental design - range of conditions
(undergrowth/ silviculture/ terrain / age);
Understory vegetation removal timed.
FARO – Multi and Single scan plots
ZEB1 – Operator will walk around the plot
Is terrestrial LiDAR useful for forest inventory?
How will we judge success?
1. Replace some component of the current
measurement system in a cost effective manner.
2. Outputs that can be integrated into current forest
information systems.
What is LiDAR worth to forest management?
Identify and survey several NZ forest managers using LiDAR. Detail
the financial returns and break-even points.
Forest inventory Forest engineering Erosion risk Harvest configuration
H & S
Hydrology
Operational planning Fire risk Habitat management
LiDAR Cost-benefit – Project plan
Identify and survey forest management case studies about the costs – benefits of LiDAR.
Produce and document a valuation model that can incorporate these values.
Publish results and make valuation model available to forest managers.
A useful parameter for assessing
forest productivity.
Interventions (fertiliser application)
can induce large increases in LAI.
Rapid response (1-2 years).
LAI is difficult and expensive to
measure.
LiDAR offers a new approach.
Leaf Area Index (LAI) and Forest Productivity
Photo: Grant Pearse
What can LiDAR-derived LAI offer?
High resolution site quality
assessment
Targeted site interventions
(fertiliser)
Monitor and quantify
value of site interventions
Novel method for
disease monitoring
Transition to precision forestry
High resolution satellite imagery
Datasets
• RapidEye - Full coverage of Kaingaroa - 5m resolution
• Worldview2 - Imagery for three 100km2 sites covering the genetics trials
- Multispectral 2m resolution Panchromatic – 0.5m resolution)
Research Plan
1. Generate spectral and textural information from satellite imagery. Including
vegetation ratios, texture measures, and estimates of LAI (tested alongside LiDAR-
derived estimates).
2. Use these to calculate stand estimates of forest variables and compare the stand
estimates from satellite imagery, with those from LiDAR, and with estimates
produced from combining both data sources using a kNN imputation model.
3. Evaluate the estimates produced at stand level and potentially at the tree level
(canopy extent derived from LiDAR).
Objective
• Further develop this technology and deliver to the forestry sector.
Concluding comments
Full and varied programme of
remote sensing research.
Blend of scientific research and
case studies aimed at delivering
near term benefits to the sector.
Guided by the development of a
industry led remote sensing/
phenotyping cluster group.
Acknowledgements
Research Providers Contributors
Scion Research Jonathan Dash
University of Canterbury Dave Pont
Silmetra Limited Mike Watt
Indufor Grant Pearse
Interpine Pete Watt
top related