Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
LiDAR/EFI Cross-Country CheckupWebinar hosted by theCanadian Wood Fibre Centre (CWFC)
February 6, 2020
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
CWFC Research Program2
Understand the characteristics of
desirable wood fibre
Locate trees with desirable
characteristics
Produce trees with desirable characteristics
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
3
Research Program 2020-2023
Petawawa Research Forest: Remote Sensing Supersite
2012 DTM 2013 Vexcel Ultracam
2012 CHM
2012 EFI Merchantable Volume 2012 EFI Mean Height2007 Forest Inventory
https://opendata.nfis.org/mapserver/PRF.html
https://pubs.cif-ifc.org/doi/pdf/10.5558/tfc2019-024
Joanne White, Hao Chen, Murray Woods, Brian Low, Sasha Nasonova, Andy Yang
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
• Canadian Centre for Earth Observation and Mapping (CCMEO)
• Collaboration with provincial, territorial and municipal partners
• 350,000 km2 acquired since 2015• High Resolution Digital Elevation
Model (HRDEM) available on the Open Government website
• Federal Airborne LiDAR Data Acquisition Guideline
• LiDAR data quality control system• New geospatial layers derived from
high-resolution elevation
5
NRCan National Elevation Data Strategy
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
6
© Her Majesty the Queen in Right of Canada, as represented by the Minister of Natural Resources, 2020
Join the mailing list:https://forms.gle/QNRQeHWgK5udK3AU6
Link will be pinned to the top of my Twitter:@adamdick
7
1Cross-Country Checkup
Forest Inventory Section
Newfoundland and Labrador
2 February 6, 2020
“Re-focused Forest Inventory”Doing more with less
• Field Work• Photo Interpretation• Conclusion
3 February 6, 2020
“Re-focused Forest Inventory”Field Program
• Reviewed PSP Network
• Dropped Labrador plots• Dropped some poor sites• Dropped long travel plots• Most were helicopter plots• Dropped 300+ plots
4 February 6, 2020
“Re-focused Forest Inventory”Field Program
• Reviewed PSP Network
• Added plots in managedstands
• Approximately 200 plots will be added
• Does not add our helicopter use
• PSP network more focused
5 February 6, 2020
“Re-focused Forest Inventory”Field Program
• Timber Cruising
• Site data for certain areas• Re-developed TC
procedures• Cooperative effort
between HQ and Districts• 10 blocks cruised this
year to date• Results overall good
6 February 6, 2020
“Re-focused Forest Inventory”Doing more with less
• Field Work• Photo Interpretation• Conclusion
7 February 6, 2020
“Re-focused Forest Inventory”Photo Interpretation
• In-house - 1 district, contract out - 1 district
• Funds cut for contract work
• Divide into core and non-core
• Core get regular interpretation – non-core updated for disturbances with full interpretation every 2nd cycle. February 6, 2020
8 February 6, 2020
“Re-focused Forest Inventory”Photo Interpretation
• Disturbance Update
• 2018 started to use Sentinel imagery for updating
• No real issue with clouds.
• Updates are more frequent, near full coverage, and free
• Thanks Dale Wilson
“Re-focused Forest Inventory”Photo Interpretation
• New typing initiatives
• Capture species in10% classes
9February 6, 2020
“Re-focused Forest Inventory”Photo Interpretation
• New typing initiatives
• Creation of stand origin age layer• To grow forest between inventories• Relying on 2nd growth and managed stands• Using plots, harvest, silviculture and
disturbance data to capture stand age• Sounds good – but devil in the details• Cleaned up archived data layers will
probably be as valuable as improved age and volumes estimates
10February 6, 2020
“Re-focused Forest Inventory”Doing more with less
• Field Work• Photo Interpretation• Conclusion
12F
ebru
ary
6, 2
020
“Re-focused Forest Inventory”Conclusion
• The need for forest inventory data is timeless
• NL forest inventory program has been re-focused
• Adversity helps to re-focus• Today - hoping to hear some
ways in which our program might be improved upon.
12February 6, 2020
Questions ?
13
Thank You!
14
6 Feb. 2020
AERIAL PHOTOGRAPHY
LAND ELEVATION
LAND COVER / FOREST INVENTORY
HYDRONETWORK
PUBLIC AND PROTECTED LAND
SOIL CONSERVATION / CLASSIFICATION
COASTLINE VULNERABILITY TO EROSION & FLOODING
RESPONSIBLE FOR MAPPING, MONITORING
AND MAINTAINING OUR NATURAL RESOURCES FOR PROGRAMS,
POLICY & LEGISLATION
AERIAL IMAGE ACQUISITION10 YEAR CYCLE
1894 SAMPLE SITES
804 IN FOREST MONITORED ON A 10 YEAR CYCLE FOR FOREST BIOMETRICS
1090 IN AGRICULTURE MONITORED YEARLY FOR CHEMICAL BIOLOGICAL & LANDUSE
CONTINUOUS FOREST INVENTORY GRID
Enhanced Forest Inventory Metrics
AREA-BASED FOREST INVENTORY METRIC
PREDICTION USING PRINCE EDWARD
ISLAND CONTINUOUS FOREST INVENTORY
PLOT MEASUREMENTS AND AIRBORNE
LiDAR POINT-CLOUD STATISTICS
Paper by Thomas W.R. Baglole
Intended Use: New forest metrics (volume, height, basal area,
diameter, Biomass, + Predictive model)as aid to State of the Forest Report
Increased efficiency for public and private sectors;
New & updated products such as surface water flow, habitat classification, carbon budget modelling.
Improve the ability of PEI Forest, Fish & Wildlife to achieve on its forest management related mandates.
Aid forest managers in further closing gaps on information needs for practicing sustainable forest management
Improved decision making and landscape management across the Island.
Questions? / DiscussionForest Industry on PEI in state of
flux since announcement of
closure of closest pulp mill in Nova
Scotia.
Lack of Private Industry push for EFI
Nova Scotia Forest Inventory Update
FEB 6, 2020
Inventory Program Snapshot • Four main programs:
1. Ground Measurement (PSP) Program
2. Photo Interpretation Program
3. Geographic Information System (GIS) Program
4. Enhanced Forest Inventory (EFI) LiDAR
22
Permanent Sample PlotsRepresentative of NS Forests
• 3,228 active plots
• Randomly placed across the province
• Assessed for bias each time they are visited
• Treat as surrounding
forest
• Dropped or moved if biased
• 5 year cycle
• Monitor, track changes in forest over time
3
Photo-Interpretation Program• Delineate and capture homogeneous forest stands in a digital\spatial database
format
• Internal and contract
• Interpret five (5) main attributes
1. Height
2. Crown Closure
3. Species
4. Land Capability
5. Volume Estimates
4
5
6Spot Interpretation
Departmental license purchased
from Planet Labs of provincial mosaic
Daily revisit
Pansharpened 1.5m
R,G,B and R,G,NIR
2017 2018
7
7
808 Commercial Thinning, SW
Began with pilot area in 2016
Process of QC is ongoing for 2018 data
Largest collection year was 2019
• First large delivery last week
• QC process ongoing
Note: Map is indicative of area flown, not the
area that has been accepted through QC.
8LiDAR Acquisition
LiDAR Data Derived surfaces and LAS tiles
available online once QC is finished
• https://nsgi.novascotia.ca/datalocator/elevation/
Larger blocks of data can be accessed by contacting [email protected]
9
Updates & Current Focus Expanded personnel
Processing deliveries from 2019
EFI research collaborations
Utilizing LiDAR products to inform the photo-interpretation program
Continue developing library of LiDAR-statistics
10
Deliveries ~ 30,000 tiles expected by end of March
First large delivery last week
Calculation of statistics
Looking at potential field data gaps
New PSP locations to process and import into system
11
EFI Research Collaborating with Dr. John A Kershaw and
PhD Candidate Ting-Ru Yang at UNB
Systems of equations for EFI
• Leverages all available field data
Applied to the 7 eastern counties of NS
Additional details once research is published by Ting-Ru
12
LiDAR and Photo-Interpretation Potential changes to what is interpreted and how it is interpreted
Ways to automatically define vertical structure visible by photo-interpreters
• What layers/distinct canopies exist that can be interpreted?
• How can we characterize?
13
53.8% of stand
A002-02299
is comprised of trees ranging in height from 11.0 to 13.9 metres
30.1% of stand
A002-02296
is comprised of trees ranging in height from 1.0 to 3.9 metres
and 30.5% is comprised of trees ranging in heightfrom 15.0 to 17.9 metres
Distribution of Modeled Peak Heights in Forest Stands
* classes with less than 0% by count excluded
use of the gaussian mixture model
when the peaks aren’t obvious
LiDAR Statistics Statistics calculated using the R package
lidR
Not all Divisions have the ability to process raw point clouds
• Storage
• Computing power
Raster outputs are manageable
16
LiDAR Statistics More options and versatility than simply
providing EFI predictions
Anticipate statistic library useful for predicting potential suitability
• Old growth
• Habitat
Dynamic and growing
17
Questions?
18
NB Renewable Resource Inventory
Cross Canada Checkup
Thursday, February 6, 2020
Renewable Resource Inventory
• Forest Update using DAP and Satellite Imagery• LiDAR and Forest Metrics• Continuous Landscape Inventory• Wetland/Habitat Developments• Species Project
DAP Acquisition2013 -2017 (ERD contract)
• 30 cm GSD
• 4-band imagery
• 1m horizontal accuracy
• 2 kmx 2 km tiles for orthos
• Digital Stereo pairs
2018 SNB Contract:
• 10 cm GSD
• 4-band imagery
• 8TB
Year Size (Ha) Size (GB)
2013 700,000 385
2014 736,000 435
2015 648,000 375
2017 1,280,000 750
2019 DAP Acquisition
• 20 cm GSD• 11,500 km²• 4-bands of color (RGBNIR)• 97% Completed
Landbase Update• Landbase built annually
• Annual Crown harvest/ silviculture updates
• Annual private woodlot silviculture updates
• 1/10 of province interp update• Forest stands
• Water (NBHN)
• Wetlands (WESP 2015)
• Non-forest
• Agriculture
• Mines
• Residential
• Industrial
• Species not from LiDAR …yet!
• Improvements• Improved spatial accuracy between
categories (FO,NF,WL,WA)
• ID wind susceptible stands
• Site more consistently depicted
• Less species grouping: more single species calls
• LiDAR verified
• Made in NB incorporating forest dynamics
• Grow Sapling stands to Young stands
Change Detection
SPOT Imagery• Natural color (RGB) & false-
colour infrared layers
• Seamless, <5% cloud, consistent data coverage
• 1.5 m resolution (pan-sharpened)
• 2019 vintage, fully refreshed yearly
• Projection: Planet supports any standard EPSG definition
• 1 user (FPS) $70,000
• 3 or more users $250,000
Change detection examples:
• Harvest update on Private land
• Natural Disturbance update
• NDVI products and other feature extraction
• Audit tool for Crown harvest/ silviculture
Change Detection
• Private Woodlots• Sentinel and SPOT
LiDAR Acquisition
Status
20132014
Year pt/m² Sensor Area (ha)
2013 1 Reigl 680 718,000
2014 1 Reigl 680 718,000
2015 6 Reigl 680/780 964,500
2016 6 Reigl 680/780 1,876,000
2017 6 Reigl 680/780 2,568,000
2018 6 Reigl 1560 2,391,000
Year Acquired Processed Available
2013
2014
2015
2016
2017
2018
LiDAR Point Cloud: http://geonb.snb.ca/li/
LiDAR Derived
Products• Digital Elevation Model (DEM) 1 and 10m
• Digital Surface Model (DSM) 1m
• Canopy Height Model (CHM) 1m and 10m
• Slope 1m
• Hillshade 1m
DEM DSM
CHM Slope
Hillshade 45ºHillshade 315º
LiDAR Forest
Metrics (EFI)• 20 m X 20m grid across the
entire forest of NB
• 13 metrics are purchased from the vendor
• 39 metrics are prepared in house, most derived from the base 13 metrics
Continuous Landscape
Inventory (CLI)Continuous Landscape Inventory
• 2Km x2km grid across NB
• 15,000 Forested Plots
• 1/10th of plots measured annually
• 3000 plots will be PSPs
• Replaced previous PSP and FDS programs
• Primary uses are:• LiDAR Calibration/validation
• Wood Supply Model Calibration
• Growth and Yield Calibration
• 1000 plots used for calibration of LiDAR in 2018
• Forest Ranger Staff from all 18 districts offices collect the data
• Post corrected GPS locations for all plots are done in house
CLI• Vertex Hypsometer• Topcon Survey Grade
GPS • Panasonic Tough book• Angle Gauge
CLI completed to date
5200 CLI complete since 2016
Forest Inventory/Species Objectives• Increase the frequency of the Forest Update
– Automate species determination
– Continue doing EFI with some future RS solution
• Maintain or improve the accuracy of species determination– Differentiating between the common SW species have forest
management implications
– Greater demand on locating less common tree species
• Identify forest attributes for:– Live trees ; at least 5 species /cell
– Dead standing trees
– % live tree crown cover/cell
– Tree crown count
Proposed Approach
• Species Prediction Pilot Project– RFP to select 1-4 Vendors to provide species predictions on an
AOI
– Vendors to describe approach and types of RS products to be used
– Vendors deliver a species prediction product to DNRED
– DNRED will provide Training data; CLI/PSP, Coop PSPs, FDS, Landbase, LiDAR products, etc.
– DNRED will evaluate on predetermined test sites in the AOI
– Successful vendors get a chance to further improve processes
– Vendor with best solution that meets species and cost standards could provide Province wide Species prediction
Proposed Deliverables and Performance Metrics
EFI Cell
Content
Attribute
Cell Content Description Preferred Performance Metric
Species
prediction and
composition
For each EFI cell (see example in Figure 1), a proportional
breakdown by crown coverage will be estimated and
include:
• % of live individual or grouped tree
species – minimum of 5 tree species.
Individual species is preferred.
Appendix A provides a preferred list
of species and alternate species
groups.
• % of dead trees. SW vs HW is
preferred.
• % of no trees. Where no tree species
is present, the proportion of
“NoTree” found is stated.
Other notes:
• Proportions will total 100% (%species + %dead +
%NoTree = 100%)
• Where insufficient data is available, individual tree
species may be grouped further - as defined in
Appendix A.
• Minimum of 60%
accuracy for an individual
species
• Minimum of 70%
accuracy for a species
group
Percent crown
closure
Subtraction of 100 – NoTree. Represents the % crown
closure within the EFI cell.
-
Tree crown
count
Estimate of the number of live tree crowns present within
the EFI cell.
• Minimum of 80%
accuracy
Examples of Deliverables Coinciding with EFI Grid
Proposed Species Prediction Pilot Project Area
Wetlands/Habitat Inventory
• Wetlands• Monitoring
o CLI/PSP Protocols for Wetlands
o Plots measured in cooperation with ELG
• Regulatory
o Development of a Wetlands reference map based on DNRED wetland layer
• Wet Areas Mapping a provincial approach
• Habitat• Deer Wintering Habitat
Project
Enhanced forest inventory developments in Québec
Cross-country checkup
February 6th 2020
Antoine Leboeuf ing.f., Ph.D.
Four important projects
1) LiDAR acquisition and developments
2) Hydrography
3) Other developments
4) LiDAR continuity project
1. LiDAR - Acquisition report
80 % of the managed area to be covered by March 31st
2012-2019 2020
401 000 km² 66 500 km²
1. Base products (in a map sheet tile, with suggested symbology .lyr)
(i) DTM
(ii) DTM Hillshades
(iii) Canopy Height Models (CHM)
(iv) Slope
2. Operability products
(i) Forest operation constraints (steep slope, landlocked areas)
(ii) Slopes at 5 m
(iii) Contour lines (2 m, 5 m)
(iv) CHM focal
(v) etc.
1. LiDAR - Production of derivative products (In house)
2. Hydrography
1. Linear hydrography
Software for linear hydrography optimized.
More than 40 000 km² in covered now.
2. Surfacic hydrography
Lakes from ecoforest maps are superimposed to the linear hydrography
We work with a group (3 ministry) that aimed to use these maps to build a new version of the official hydrographic map.
2. Hydrography
3. Riparian area “Écotone”
2. Hydrography
3. Riparian area “Écotone”
3. Other current developments 3.A. Point cloud classification – water bodies (lidR)
3.B. Point cloud classification – power lines (lidR)
3.C. Surficial deposits (organics and marines) – coop student and research partner
4. LiDAR continuity project
o LiDAR acquisition one year in advance compared to previsions
o New technologies (small camera in the LiDAR sensors)
o 2020 – 4 500 km² with this acquisition (to define aerial photos parameters as wide as possible)
o 2021 – 9 000 km² and final decisions (go, no go).
Thank you!
Questions ?
Ontario Forest Resources Inventory : Single Photon LiDAR
2020 Cross Country Checkup
February 6 2020
2018 Single Photon LiDAR (SPL)Ontario Forest Resources Inventory
Presentation Overview
2018 Single Photon LiDAR (SPL)Ontario Forest Resources Inventory
1. Overview of Single Photon LiDAR
2. Scheduling 3. Progress4. Vertical Accuracy
Assessment5. Fixed Area Field Plot Sample
Design6. Questions
Beam Splitter
Simplified Single Photon LiDAR
3
Single Photon LiDAR vs Line Scanner LiDAR
SPL : Single photon LiDAR
- Technology for large mapping areas.
- High flying height allows for overlap and maintaining high point densities
- High efficiency LiDAR system for supporting change detection
- Fewer flight lines, reduced data processing
4
SPL100 (30deg Field of View) ALS80 (30deg Field of View)
Flying Height (AGL) 3,800m (2000m swath width) 1,200m (640m swath width)
Aircraft Speed 180kts 110kts
Capture rate (single swath) 670sqkm hr 90sqkm/hr
Processing time 80x flight time 4x flight time
Specification for 25pt/m Data Capture
Single Photon LiDAR - Technology
Products & Derivatives
• Classified point cloud
• Bare earth Digital Elevation Model, Canopy Height Model, Digital Surface Model, signal width (intensity for Single Photon LiDAR)
7
Scheduling:
0
20000
40000
60000
80000
100000
120000
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027
Annual FRI Production VS FMP Scheduling
Annual FRI Production km2 FMP 2029-2042
7
2018 Progress
Exploring the Innovation Potential of Single Photo LiDAR for enhancing Ontario's Forest Inventories
1. Characterizing terrain under varying forest types and canopy densities;
2. Quantifying the comparative performance of SPL in an area-based approach to forest inventory attributes & incremental advantages to supporting Individual Tree Approach inventories.
Co-Leads:
Dr. Joanne White – CFS
Murray Woods – MNRF (retired)
Melissa Vekeman – CWFC
Jordan MacMillan – CIF
Project Partners:
Annie Morin – CNL
David Belanger – CCMEO
Dr. Jili Li - FPinovations
KTTD2 Project
9
Canada Centre for Mapping and Earth Observation (CCMEO) Analysis – David Belanger
• assessed this 2018 SPL LiDAR dataset on behalf of the
Canadian Forest Service and the Canadian Wood Fibre
Centre.
• study included a vertical accuracy assessment based
upon only 9 RTK survey points, none of which were in
vegetated areas.
• concerns noted about the relative low density of
ground returns produced by SPL in some vegetated
areas, in comparison to linear mode LiDAR
• recommended that further assessments be
conducted in vegetated areas using a sufficient
number of checkpoints.
LiDAR DatasetGround Return Density%
(> 2 pts/m2)
2012 LML Leaf-on 34%
2018 SPL Leaf-on 31%
*Based on a 20 m raster where more than 2 ground returns/m2 recorded
LiDAR Dataset
Ground Return Density%
(> 2 pts/m2)
CCMEO MNRF
2012 LML Leaf-on 34% 31.9%
2018 SPL Leaf-on 31% 36.6%
2019 SPL Leaf-off (3.8km) - 87.2%
2019 SPL Leaf-off (2km) - 95.6%
LiDAR DatasetGround Return Density%
(> 2 pts/m2)CCMEO MNRF
2012 LML Leaf-on 34% 31.9% 18.1%**
2018 SPL Leaf-on 31% 36.6% 35.6%
2019 SPL Leaf-off (3.8km) - 87.2% 81.1%
2019 SPL Leaf-off (2km) - 95.6% 93.8%
**Intersection of all dataset extents, minus water
10
Ground Point Density / m2
Landcover2019 SPL
2km
2019 SPL
3.8km
2018
SPL
2012
Linear
Black Spruce 4.3 3.0 2.9 1.3
Jack Pine 5.9 3.2 4.4 3.0
ConPlant 4.0 4.0 4.5 1.4
Red/White Pine 4.6 4.0 2.4 1.2
Intolerant Hardwood 5.4 3.5 1.2 0.9
Tolerant Hardwood 6.1 4.8 2.1 0.7
Mixedwood 5.4 3.1 1.8 1.2
Low Vegetation 4.3 4.3 2.5 1.5
Average: 5.2 3.7 2.3 1.2
Ground Point Density of Survey Plots by Landcover*
* 2007 Inventory Polygons
Ministry of Natural Resources and Forestry
2012 LML Leaf-on
2018 SPL Leaf-on
2019 SPL Leaf-off
Site 1: Integrated wetland and stream areas
12
Natural White & Red Pine Stand
2018 SPL Leaf-on
Ground Returns
Jack Pine Stand
Ground Returns
Category Measure (cm) QL 1
2019 SPL 2km
Leaf-off (N)
2019 SPL 3.8km
Leaf-off (N)
2018 SPL
Leaf-on (N)
2012 Linear
Leaf-on (N)
Non-Veg. Mean Vertical Error 2.5 4.7 4.3 6.1 11.2
RMSEz 10 9.1 9.5 7.4 12.1
NVA (RMSEz) 19.6 17.9 (79) 18.6 (79) 14.4 (85) 23.8 (85)
Non-Veg NVA (95th Percentile) 14.1 16.3 13.8 17.3
Vegetated VVA (95th Percentile) 30 14.5 (221) 16.9 (221) 23.4 (236) 18.7 (236)
Classified (95th Percentile)
Road Gravel Road 14.8 (47) 18.6 (47) 10.8 (53) 18.3 (53)
Asphalt Road 13.7 (32) 12.4 (32) 15.1 (32) 16.2 (32)
Conifer Black Spruce 13.8 (37) 15.0 (37) 29.5 (37) 18.8 (37)
Jack Pine 7.0 (15) 15.1 (15) 7.4 (15) 7.7 (15)
ConPlant 11.9 (21) 14.5 (21) 15.8 (36) 20.1 (36)
RedWhite Pine 13.6 (27) 20.8 (27) 16.2(27) 17.4 (27)
Hardwood Intolerant Hardwood 14.9 (37) 15.6 (37) 17.6 (37) 19.7 (37)
Tolerant Hardwood 13.6 (35) 14.0 (35) 15.6 (35) 18.1 (35)
Other Mixedwood 17.6 (34) 19.3 (34) 26.5 (34) 16.3 (34)
Low Vegetation 8.7 (15) 5.0 (15) 24.7 (15) 23.4 (15)
Testing against accuracy standards
for a 10-cm Vertical Accuracy Class
equating to:
Non-vegetated Vertical Accuracy
(NVA)
of +/- 19.6-cm at 95% CI
Vegetated Vertical Accuracy
(VVA)
of +/- 29.4-cm at the 95%
Percentile
Preliminary Results
Note: All values in cm
*RMSE method of testing NVA invalid if Mean Vertical Error exceeds 1/4 RMSE limit.
Development Of A Forest Inventory Using Single Photon LiDAR & Assessing Decadal Forest Change
Grant McCartney – RYAM Forest ManagementDr. Nicholas Coops – UBC Forestry IRSSMartin Queinnec – UBC Forestry IRSS
Structurally Guided Sampling for Locating Plots
Wall to Wall metrics (20mx20m) calculated for the FMU
Principle Components Analysis performed on Forested Polygons only using 20 SPL metrics
• Height percentile metrics: p05, p10, p20, …, p90, p95, p99
• Average height of first returns / Average square height
• Cover: % first returns above 1.3 m, 5 m, 10 m and 15 m
• Structural Variability: standard deviation
• PCA 1 - 76% of the variance
• PCA 2 - 11% of variance
• PCA 3 – 7 % of the variance
16
Determine Candidate Cells For Sampling
Between 30m and 200 m of roads (90% of existing plots are within 200 m of roads)
Within 200 m of any existing plot accessible via 2x4 truck or 4x4 truck but > 200 m from roads
Remove 100 m wide band around power lines
Keep only cells located within FOR polygon (productive forest stands)
3,383,903 candidate cells for samplingExample where the road to access the IMF plot (square shaped) is not
in the database.
Comparison: Existing plots vs. New plots
18
257 plots: 97 existing G&Y, 20 IMF, 160 new
Plot by Species Group
20
Working Group
(WG)
#
plots
%
plots
%
RMF
area
PO (Poplar) 110 42.5 15.2
Sb (Black spruce) 62 15.2 49.9
Pj (Jack pine) 41 11.9 11.2
BW (White birch) 36 11.8 14
SW (White
spruce)
4 1.5 2.2
CE (Cedar) 2 0.8 3.7
LA (Eastern
Larch)
2 0.8 1.9
BF (Balsam Fir) 1 0.4 1
Sx (Spruce mix) 1 0.4 0.5
21
Strata 45
Strata 52
Strata 44Strata 54
Strata 85
Strata 63
Questions ?
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Using UAV’s to Measure
Renewal Success
“If a picture is worth a thousand words”…
Are pictures worth a thousand numbers too?
(256 at a time)
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
Background: Renewal Assessment Mandate – Prompt renewal of harvested forest lands, a requirement of Forest Act and Forest Management Agreements
Forest Act - Forest renewal34(1.1) The holder of a timber cutting right must do one of the following:(a) pay to the Crown the forest renewal charge established in the regulations on Crown timber harvested by the holder;(b) pay the forest renewal charge established in the regulations on Crown timber harvested by the holder to a third party who has entered into an agreement with the minister to perform forest renewal on Crown lands that the holder has harvested;(c) if the minister approves, carry out forest renewal on Crown lands that the holder has harvested.
Conditions on approval34(1.2) As a condition of granting approval under clause (1.1)(c), the minister may impose any term or condition on the holder of a timber cutting right that he or she considers appropriate.
Forest renewal standards34(1.3) A third party who enters into an agreement with the minister under clause (1.1)(b), or the holder of a timber cutting right who performs forest renewal under clause (1.1)(c), must ensure that(a) the renewal is performed in accordance with the terms and conditions set out in the timber cutting right under which the timber was harvested and meets the standards established in the regulations; or(b) the renewal meets the standards established in the regulations, if the timber cutting right does not address forest renewal.
FML agreements:22 (D)The Company acknowledges its primary forest management and renewal responsibilities by ensuring that all harvested areas in FML X are regenerated to approved Provincial Standards. The Company's renewal responsibilities only apply to stands harvested after the date of the signing of the Agreement.
Traditionally forest renewal assessment meant…
Circular fixed area plots of 10 metres2 in size with a radius of 1.78 metres are established in a systematic grid pattern. Plots are checked to see if they contain at least one acceptable tree and one performing tree. An acceptable tree is a healthy tree, of certain height and of appropriate age. If an acceptable and/or performing tree is present then the plot is considered performing and/or stocked. A performing tree has increased height requirements for softwood species. A healthy tree cannot have any damage associated with it…
Why
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
UAV Transformation Capital (Ideas) Fund
• Forestry and Peatlands acquired two UAV’s for Measuring Reforestation success
– We anticipate using the UAV’s to acquire around 2,040 hectares this season
• 500 ha will be ground sampled for validation (~ 20%)• 210 ha of hardwood• 43 ha of softwood leading mixedwoods• 1,793 ha of softwood (2008 Woodridge fire area)
• “Additions”:– Regional requests– Tree improvement sites– Assisted Migration trial– Forest Health– Other…
• A Mission Request (AGOL) App has been built– Tracks mission information and status
What we are using:
Ground School Training
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
Change
Manual Evaluation of Renewal Areas: Visual Based assessment Manual, ocular and quantitative analysis required to “pass” renewal blocks
We are currently looking at alternatives to manual evaluation of these areas– Computer recognition of hardwood/softwood species – Determination of Density numbers
– This could be:• Traditional remote sensing techniques• Deep Cycle Machine Learning• Other
– We are considering a Deep Cycle Machine Learning approach that the University of Winnipeg demonstrated in a GeoManitoba pilot project
– We are considering an online software as a service offering from PicTerra– Other players like Amazon (Amazon Web Services) also have offerings that
are being considered
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
Current Traditional RS: Maximum Likelihood Classification, ISO Unsupervised…
• Unsupervised Classification, just to see what can be pulled, play with the number of classes, natural breaks, etc.
• Supervised Classifications may help too but we can interpret trees too
The scale is 1: 80. This maximum likihoodsupervised classification has done a reasonable job in identifying young spruce trees
15
UAV Focus
Basic data … plus
Additional data
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
What does NSR look like, using 10m cells
Green cells have trees meeting standards, white cells do not
How do we determine this
Simplified process for success
Acquire the UAV Imagery and classify Determine hardwood softwood splits, convert them to polygons
Generate a 10 by 10 metre grid
Get average heights using the canopy modelDetermine success
The process up close…
Ground sampling verification
Results, so far: (2 + 6 months)
FMU BLK_ID DATEUAV
PERSON DAYS
VALIDATIONSURVEY
YEARAREA (ha)
GROUND PLOTS
GROUND SURVEY
DAYS
24ER2010-055 May-07-19 2.0 Complete 2019 67.21 134 3.4
24ER2008-011 May-14-19 1.0 2019 20.75 62 1.6
24ER2009-023 May-14-19 1.0 Complete 2019 85.33 175 4.4
24ER2008-017 May-15-19 3.0 Complete 2019 146.71 295 7.4
24ER2008-013 May-21-19 1.0 2019 6.75 20 0.5
24ER2009-024 May-21-19 2.0 2019 17.74 53 1.3
24ER2009-030 May-21-19 2.0 2019 135.24 270 6.8
24ER2008-061 June-13-19 0.7 2019 50.22 100 2.5
24ER2008-063 June-13-19 0.7 2019 39.97 80 2.0
24ER2010-009 June-13-19 0.7 2019 12.64 40 1.0
24ER2009-012 June-18-19 1.5 2019 45.35 91 2.3
24ER2009-110 June-18-19 1.5 Complete 2019 129.76 265 6.6
24ER2009-014 June-26-19 3.0 2019 166.71 333 8.3
24ER2006-022 June-27-19 1.5 Complete 2019 16.86 51 1.3
24ER2006-023 June-27-19 1.5 2019 19.12 57 1.4
24ER2009-029 July-12-19 2.0 2019 255.49 511 12.8
24ER2008-020 July-19-19 2.0 2019 84.87 170 4.2
24ER2008-011 July-23-19 1.0 2019 36.19 105 2.6
24ER2009-019 July-23-19 1.0 2019 77.00 154 3.9
24ER2010-031 July-23-19 1.0 2019 83.57 167 4.2
24ER2009-020 July-24-19 1.0 2019 73.32 147 3.7
24ER2009-026 July-30-19 1.5 2019 60.71 121 3.0
24ER2009-027 July-30-19 1.5 2019 57.82 116 2.9
24ER2005-072 August-15-19 1.0 2019 94.01 188 4.7
24ER2006-008 August-15-19 1.0 2019 15.57 47 1.2
46NW03-013 September-17-19 1 2019 87.12 174 4.4
46NW04-002 September-17-19 1 2019 53.66 107 2.7
46NW09-006 September-17-19 1 2019 68.87 138 3.4
46NW05-007 September-19-19 3 Complete 2019 135.05 270 6.8
Totals 42.0 2143.6 111.0
The UAV was 40% more efficient in field time over the ground assessment
An office analytics component is still required before finalization of renewal assessment status – moving to automate
The UAV data provides a permanent record
Continuous improvement techniques can be used on the data as technology changes
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
Software as a Service…?
Other Options:
We have learned and understand more now. So now we could explore: • Pilot the Deep Cycle Machine Learning
(DCML)• Develop / automate the processes internally• Farming it all out• Make use of Software as a Service (SaaS)• Develop approaches to incorporate more
elements (forest health, inventory)
Additional Elements
• Forest Health data– NDVI
• Species composition determination (Manual) to update the base inventory
Overview:
• Background (The Why)• Platform (The What)• Change• Data, processing and time• Outcomes• Other options• Next steps• Questions
Next Steps
• Increase annual data acquisition to around 5,000 ha
• Standardize processes with stakeholders• Explore SaaS/DCML capabilities
• Potentially acquire additional sensors and/or rotary platform(s)
• Regional initiative/support
Questions?
Orthophotos and NDVI
Forest Inventory Update 2020
The development of an efficient inventory for Saskatchewan
Lane Gelhorn, RPF
Need for a New Paradigm: challenge of products
HW
D
SWD
1
Remote Sensing of Elevation: DTM or DSM
Need for a New Paradigm: opportunity of new data
2
Need for a New Paradigm: opportunity of new data
3
Richer Content: 3 bands to 4 plus surface
Need for a New Paradigm: opportunity of new data
4
Inventory Approach for Saskatchewan
5
Inventory Approach for Saskatchewan
Fed/Prov
6
Inventory Approach for Saskatchewan
Province
6
Prov/Industry
Digital Terrain Model 5m Percent Softwood 10mLand Classification 10m
Crown Coverage 10m Tree Height 5m and 20m Ecosite PEM 30m
7
Basal Area 20m Quadratic Mean DBH 20mStems > 10cm 20m
Gross Volume 30/08 20m Gross Volume 0/0 20m Stand Polygons (1ha)
8
New Map Concept Needed
✓ 3D representation to show slope (hillshade of DTM)
✓ Two scales (1ha poly, 20m raster)
✓ Within-polygon variability
9
Familiarity is the Thin Edge
✓ Retro Label (eg HS24C)
✓ Retro labels
10
New Ways to Display Info Needed
✓ Label Pie Charts
11
New? Landbase Application
✓ First time since 1950 that we have mapped area outside of the provincial forest
12
But our FRI can’t do it all.
13
Species, Age, SI
Inventory Approach for Saskatchewan
Industry
14
Description of the Area, Available Data
Purpose of the Inventory
Deviations from the Standard Attributes
Scale of Application
Methodology Employed
Ownership and Use
Funding
Audit
Timelines
FMI: Forest Inventory Plan Provides Flexibility
15
Forest Inventory Standard Development: Two Saskatchewan Trials
We tested 7 inventories on two landbases in order to understand the accuracy and cost profile of each. This was used to inform standard development: not how to do the inventory, but how accurate we could expect the results to be.
The result is the forest inventory code chapter and standard, now available for public review.
16
We build and assess inventories at 400m2 (20m pixel or 11.28m radius field plot) but are used to looking at stand level accuracy.
Need to develop specificcurves to adjust for interdependencies between adjacent cells in order report equivalent accuracy statistics.
RMSE at 400m2 plots
The “Known Unknown”: Expansion of ABA metrics to 1ha or stand level
Currently we are establishing dense cruise plots to examine how RMSE scales up from 400m2 to 1/2ha to 1ha. We anticipate a different result for each inventory method.
Coloured lines indicate various inventory methods: ABA LiDAR, SkyForest, ITC LiDAR, RADAR, etc.
RM
SE p
er h
a
17
The “Shoulda” Knowns : Lessons Learned the Hard Way
Because we planned one field season per map, sometimes our field samples were not as statistically efficient as they could be (we got less bang for our buck) because they relied on coarse initial estimates for stratification.
We are now trying a staggered, iterative delivery
+ + = Cluster (Strata)Sample then Reproduce Calibrated Version
PctSW Radar Height
Sample the centroid of the k-means cluster
18
Some photogrammetric blocks are just not a good idea – sometimes you have an ‘albatross’
Ground control is important, but opportunities might be limited
The “Shoulda” Knowns: Lessons Learned the Hard Way
19
Persistent Unknowns: Things we just can’t figure out how to do
Crown cover is an intuitive metric from the air, but for a ‘ground up’ inventory ?
We are now using LiDAR samples to power our crown coverage estimates, but are searching for a compatible VCC solution for every field plot
20
Your suggestions, and questions, are welcome!
Enhanced Forest Inventory National Overview - Alberta Update
Chris Bater (presenting), Bev Wilson, Jinkai Zhang, Cosmin Tansanu, Hilary Cameron
Forestry Division, Alberta Agriculture and Forestry
6 February 2020
1
Partners
2
Overview of the operational remote sensing needs and current status in Canada
Enhanced techniques for forest inventory, and growth and yield assessments and modelling
Linking remote sensing to the wildfire triangle and fire behaviour
Linking research results to the improvement of forest planning and management
https://www.youtube.com/playlist?list=PL9zT662LR6d89s6uK4rGY_YSLNVOX0JQn
Remote sensing for forest practitioners, 2018
3
Current status of lidar coverage
36,784,073 hectares acquired with acquisition years ranging from 2003 to 2017.
Most actively managed forests and provincial parks have been scanned.
~$25 million spent on acquisition
Mountain pine beetle infestation provided impetus for purchase. Terrain data were needed for harvest planning. Data were not collected for vegetation inventory.
4
NetmapBuilding on wet areas mapping
5
GOA hydro layer WAM predicted stream channels
Current statusOver 400 requests for the data
Energy industry
Forest industry
Environmental consultants
NGOs
Academia
32,170,548 ha mapped (87% of our lidarholdings)
7
Credit: Lee Benda, TerrainWorks
Building Virtual Watersheds
9
Credit: Lee Benda, TerrainWorks
Improved cross-tile flow accumulation Wet areas mapping Netmap
10
Riparian zone mapping with Netmap – a pilot project
http://www.netmaptools.org/Pages/NetMapHelp/8_2_delineate_riparian_zones.htmLast accessed 5 May 2015
Estimating channel width for stream lines
Source: Anderson, R.J., Bledsoe, B.P., and Hession, W.C. (2004). Width of Streams and Rivers in Response to Vegetation, Bank Material, and Other Factors. Journal of the American Water Resources Association 40, 1159–1172.
12
Estimating channel width for stream lines
13
y = 0.8014x0.3726
R² = 0.696
0.10
1.00
10.00
0.10 1.00 10.00 100.00
Stre
am w
idth
(m
)
Drainage area (km^2)
Whitemud River watershed stream width measurements
Only 23 of 41 predicted stream reaches visited with water present had measureable channels
Field stream width measurementsAquatic habitat surveys from the Fisheries & Wildlife Management Information System (FWMIS)
> 100,000 locations within FMU boundaries
~80% of those include width measurements
14
Wet areas mappingGOA hydro Netmap
15
Sediment delivery to fish habitat
Photo credit: Jared Fath, University of Alberta
16
Sediment production versus sediment delivery to streams
Only a fraction ofroad segments (10-20%)deliver sedimentto streams
Almost 100% ofroads producesediment
17
Current statusCurrent focus is around parameterizing sediment delivery numbers (e.g. tons/year) and improving stream width models for Foothills Natural Region
18
Derived ecosite phaseEcological modelling
19
Landcover in Alberta
20
Ecosystem classification in Alberta Hierarchical system
Ecological Unit Example
Natural Region Foothills
Natural Subregion Lower Foothills
Ecodistrict
Ecosection
Ecosite Low-bush cranberry (e)
Ecosite Phase Low-bush cranberry –aspen phase (e2)
21
(Mapcode 5C) e ecosite low-bush cranberry (Lower Foothills)
e4 low bush cranberry - Swe3 low bush cranberry –Aw-Pl-Sw
22
ECOSYS (Ecosite guides and raw data)• ECOSYS Demo
• https://dotnetprod.env.gov.ab.ca/EcoSys/ (production database internal)
• https://securexnet.env.gov.ab.ca/EcoSysExternal/ (raw data)
• 26,000+ plots (soils, veg, site)
• Subregion ecosite guides (open data)
• https://open.alberta.ca/publications/9781460131701
23
Derived Ecosite Phase
AVIE – Alberta Vegetation Inventory – enhanced
Slope position
(from lidar-derived 5 m digital elevation model)
Knowledge-based rules
Derived Ecosite Phase
24
Attribute listAttributes here…….
25
Next Release of DEP
• Incorporates some digital Phase 3, new AVIE and lidar
• Attributes for Alberta Wetlands Classification System
• Updated Ecological Site Guides
Structure retention
28
Photo credit: Jim Witiw –DMI
Structure Retention Planning Tool (Scott Nielsen, Francois Robinne, U of A)
29
Mountain pine beetleWith freely available satellite imagery
30
Individual tree crown classification of mountain pint beetle mortality
31
Mapping mountain pine beetle mortality with Sentinel-2 and/or Landsat
• Intent is to map mountain pine beetle mortality in Crown-managed forest management units in west-central Alberta.
• Simple classification (e.g. presence/absence) is the basic objective.
32
Reforestation
33
Imagery CollectedProgress to Date
Current Focus
Next Steps
Forest Type Count Hectares
HwPl 4 49
Hw 4 30
Sw 6 58
HwSx 4 57
SwHw 4 37
PlHw 3 29
Pl 14 96
Total 39 356
Collected 356 ha of imagery in the Drayton Valley and Rocky Mountain House regions:
• 39 Openings captured in 3cm RGB
• 7 Openings captured in 3cm NIR
Credit: Andrew Chadwick, University of British Columbia
Assessing reforestation success using high spatial resolution remote sensing
35
Extracting Inventory Metrics
• Individual conifer detection outputs individual crown footprints
• Footprints enable extraction of inventory metrics at two scales:
Block Level1. Stem Count 2. Density3. Spacing 4. Composition
Individual Level 1. Height 2. Crown Area3. Species
Progress to Date
Current Focus
Next Steps
Credit: Andrew Chadwick, University of British Columbia
Wildfire Fuels and perimeters
37
Automatic Burned Area Mapping• Large wildfires (Class E, >200 Ha)• Data used:
• Precompiled referenced scenes for past two years• Landsat data (Landsat 7 and 8)--From USGS
• Normally available on the same day (late evening or mid night)• Sentinel 2 --from ESA or Peps
• Available one day later
• Automatically downloaded whenever the satellite passed over the fire• Automatic 2-class classification: burned vs.non-burned(e.g. green islands)
• Change detection: dNBR or Spectral Angle based• Single image classification using Random Forest/SVM
• Outputs• Pre- and current multispectral images for the fire• dNBR image—related to burn severity when field data is available• Shape file for burned areas
Burn Severity Map
Firesmart and multispectral lidarTesting lidar-derived models of….
- crown base height,
- canopy bulk
- canopy fuel load
- density
- species(?)
…..to feed the next generation of wildfire behaviour models
Credit: Hilary Cameron, Wildfire Management Branch & University of Alberta41
Industry activity
42
Industry lidar acquisitions for enhanced forest inventories
43
Company FMU EFI provider
Lidar return density
(pts/m^2)
Inventory typeFRIP amount
requested
Foothills Forest
Products Inc.E8, E10 Foresite 16
Individual tree crown
$1,094,519
Millar Western Forest
ProductsW11, W13, W15
Lim Geomatics
12 Area based $1,985,108.65*
Canadian Forest
Products Ltd.G15** Foresite 16
Individual tree crown
Unknown
Blue Ridge Lumber Inc.
W14 Foresite 16Individual tree
crownUnknown
*Includes cost of reprocessing GOA lidar in W6, W11, W13, W14, W15, S20, W01 and E01 **EFI analysis is currently underway for three areas of interest within G15, totaling ~25% - 30% of Canfor's FMA area
44
3 pts/m2 16 pts/m2
Examples of variables provided by EFI vendors to forest industry
Individual tree crown variables (trees > 10 m) Area-based analysis variables
Species Average height
Height Top height
Diameter Gross merchantable volume
Basal area Quadratic mean diameter
Local density Basal area
Crown size Density
Height to live crown Merchantble stems
Slope Piece size (m^3/tree and trees/m^3)
Aspect Log size (m^3/m)
Elevation Log size (m^3/log)
Gross & net merchantable volume
DWB factor
Log product volumes (utillization specifications provided by forest company)
Biometrics
46
47
Huang, S., Zaichkowsky, M., Weeks, D., Li, C., Brown, C., Parlow, M., Buckmaster, G., Tansanu, C., Yang, Y., 2019. Method comparison and method calibration applicable to forest measurements and model predictions. Technical Report Pub. No.: T/2019–RA01. Forest Stewardship and Trade Branch, Forestry Division, Alberta Agriculture and Forestry, Edmonton, Alberta. https://open.alberta.ca/publications/9781460143759
Remote sensing-related publications and reports • Barber, Q.E., Bater, C.W., Braid, A.C.R., Coops, N.C., Tompalski, P., Nielsen, S.E., 2016. Airborne laser scanning for modell ing understory shrub abundance and productivity. Forest Ecology and Management 377, 46–54.
https://doi.org/10.1016/j.foreco.2016.06.037• Barber, Q.E., Bater, C.W., Dabros, A., Pinzon, J., Nielsen, S.E., Parisien, M.-A., Submitted. Persistent impact of conventional seismic lines on boreal vegetation structure following wildfire. Canadian Journal of Forest Research.• Bater, C.W., White, J.C., Wulder, M.A., Coops, N.C., Niemann, K.O., In prep. Estimation of total aboveground biomass with LiDAR: A comparison of sample designs. Remote Sensing.• Bater, C.W., Wagner, M., Anderson, A.E., Diiwu, J., White, B., Benda, L., Miller, D., In Prep. Synthetic streams and virtual watersheds: linking lidar, channel geometries, and erosion data to improve forest stewardship. Forestry Chronicle.
• Benda, L., Andras, K., Miller, D., 2016. WIN-System: A Decision Tool for Cumulative Watershed Effects Assessment in Alberta (Unpublished report). Terrain Works, Mt. Shasta, California.• Benda, L., Miller, D., 2015. Integrating Wet Areas Mapping with NetMap’s Virtual Watershed to Support Spatially Explicit Riparian Zone Delineation and Management in Alberta (Unpublished report). Terrain Works, Mt. Shasta, California.• Bjelanovic, I., Comeau, P.G., 2019. Site index determination using remote sensing - Geocentric and phytocentric Site Index determination in boreal forests using remote sensing (No. FRIAA Project FFI-16-013). Univ. of Alberta, Dept. of Renewable
Resources, Edmonton, Alberta.• Bjelanovic, I., Comeau, P., White, B., 2018. High Resolution Site Index Prediction in Boreal Forests Using Topographic and Wet Areas Mapping Attributes. Forests 9, 113. https://doi.org/10.3390/f9030113• Chicoine, D., Mihajlovich, M., 2011. Field Audit of Wet Areas Mapping in the Central Mixedwood, Lower and Upper Foothills Eco-regions of Alberta. Alberta Sustainable Resource Development, Edmonton, Alberta.• Chicoine, D., Mihajlovich, M., 2010. Wet areas mapping for silvicultural prescriptions. Incremental Forest Technologies Ltd.,, and Alberta Sustainable Resource Development, Forest Management Branch, Edmonton, Alberta.• Coops, N.C., Tompalski, P., Nijland, W., Rickbeil, G.J.M., Nielsen, S.E., Bater, C.W., Stadt, J.J., 2016. A forest structure habitat index based on airborne laser scanning data. Ecological Indicators 67, 346–357.
https://doi.org/10.1016/j.ecolind.2016.02.057• Guo, X., Coops, N.C., Gergel, S.E., Bater, C.W., Nielsen, S.E., Stadt, J.J., Drever, M., 2018. Integrating airborne lidar and satellite imagery to model habitat connectivity dynamics for spatial conservation prioritization. Landscape Ecology.
https://doi.org/10.1007/s10980-018-0609-0• Guo, X., Coops, N.C., Tompalski, P., Nielsen, S.E., Bater, C.W., Stadt, J.J., 2017. Regional mapping of vegetation structure for biodiversity monitoring using airborne lidar data. Ecological Informatics 38, 50–61.
https://doi.org/10.1016/j.ecoinf.2017.01.005• Mao, L., Bater, C.W., Stadt, J.J., Dennet, J., Nielsen, S.E., Chen, Y., Submitted. Plant phylogenetic structure is associated with canopy height in boreal forest community assembly. Forests.• Mao, L., Bater, C.W., Stadt, J.J., White, B., Tompalski, P., Coops, N.C., Nielsen, S.E., 2017. Environmental landscape determinants of maximum forest canopy height of boreal forests. Journal of Plant Ecology 12. https://doi.org/10.1093/jpe/rtx071• Mao, L., Dennett, J., Bater, C.W., Tompalski, P., Coops, N.C., Farr, D., Kohler, M., White, B., Stadt, J.J., Nielsen, S.E., 2018. Using airborne laser scanning to predict plant species richness and assess conservation threats in the oil sands region of
Alberta’s boreal forest. Forest Ecology and Management 409, 29–37. https://doi.org/10.1016/j.foreco.2017.11.017• Matasci, G., Hermosilla, T., Wulder, M.A., White, J.C., Coops, N.C., Hobart, G.W., Bolton, D.K., Tompalski, P., Bater, C.W., 2018. Three decades of forest structural dynamics over Canada’s forested ecosystems using Landsat time-series and lidar
plots. Remote Sensing of Environment 216, 697–714. https://doi.org/10.1016/j.rse.2018.07.024• Mora, B., Wulder, M.A., Hobart, G.W., White, J.C., Bater, C.W., Gougeon, F.A., Varhola, A., Coops, N.C., 2013. Forest inventory stand height estimates from very high spatial resolution satellite imagery calibrated with lidar plots. International
Journal of Remote Sensing 34, 4406–4424. https://doi.org/10.1080/01431161.2013.779041• Mulverhill, C., Bater, C.W., Dick, A., Coops, N.C., 2019. The utility of terrestrial photogrammetry for assessment of tree volume and taper in boreal mixedwood forests. Annals of Forest Science.• Mulverhill, C., Coops, N.C., Tompalski, P., Bater, C.W., Rosychuck, K., In Prep. Digital terrestrial photogrammetry to enhance field-based forest inventory across stand conditions. ISPRS Journal of Photogrammetry and Remote Sensing.• Nijland, W., Coops, N.C., Ellen Macdonald, S., Nielsen, S.E., Bater, C.W., Stadt, J.J., 2015a. Comparing patterns in forest stand structure following variable harvests using airborne laser scanning data. Forest Ecology and Management 354, 272–280.
https://doi.org/10.1016/j.foreco.2015.06.005• Nijland, W., Coops, N.C., Macdonald, S.E., Nielsen, S.E., Bater, C.W., White, B., Ogilvie, J., Stadt, J.J., 2015b. Remote sensing proxies of productivity and moisture predict forest stand type and recovery rate following experimental harvest. Forest
Ecology and Management 357, 239–247. https://doi.org/10.1016/j.foreco.2015.08.027• Nijland, W., de Jong, R., de Jong, S.M., Wulder, M.A., Bater, C.W., Coops, N.C., 2014. Monitoring plant condition and phenology using infrared sensitive consumer grade digital cameras. Agricultural and Forest Meteorology 184, 98–106.
https://doi.org/10.1016/j.agrformet.2013.09.007• Oltean, G.S., Comeau, P.G., White, B., 2016. Linking the Depth-to-Water Topographic Index to Soil Moisture on Boreal Forest Sites in Alberta. Forest Science 62, 154–165. https://doi.org/10.5849/forsci.15-054• Pickell, P.D., Coops, N.C., Ferster, C.J., Bater, C.W., Blouin, K.D., Flannigan, M.D., Zhang, J., 2017a. An early warning system to forecast the close of the spring burn window from satellite-derived greenness. Canadian Wildland Fire & Smoke
Newsletter 23–25.• Pickell, P.D., Coops, N.C., Ferster, C.J., Bater, C.W., Blouin, K.D., Flannigan, M.D., Zhang, J., 2017b. An early warning system to forecast the close of the spring burning window from satellite-observed greenness. Scientific Reports 7.
https://doi.org/10.1038/s41598-017-14730-0• White, B., Ogilvie, J., Campbell, D.M.H.M.H., Hiltz, D., Gauthier, B., Chisholm, H.K.H., Wen, H.K., Murphy, P.N.C.N.C., Arp, P.A.A., 2012. Using the Cartographic Depth-to-Water Index to Locate Small Streams and Associated Wet Areas across
Landscapes. Canadian Water Resources Journal 37, 333–347. https://doi.org/10.4296/cwrj2011-909• Willier, C.N., Devito, K.J., Bater, C.W., Nielsen, S.E., In prep. Evaluating changes in forest canopy structure in road-fragmented peatlands using airborne laser scanning. Forest Ecology and Management.
48
AcknowledgementsJohn Diiwu, John Stadt, Cosmin Tansanu, Lee Martens, Mike Wagner, Bev Wilson, Michal Pawlina, Dave Schroeder, Hilary Cameron, Brooks Horne, Jinkai Zhang: Forestry Division
Scott Nielsen, Phil Comeau, Francois Robinne, Jen Beverly, Hilary Cameron, Ivan Bjelanovic: University of Alberta
Nicholas Coops, Tristan Goodbody, Andrew Chadwick, and members of IRSS: University of British Columbia
Chris Hopkinson, Laura Chasmer: University of Lethbridge
Axel Anderson, fRI
Lee Benda, TerrainWorks
Paul Arp, Jae Ogilvie, University of New Brunswick
Question? [email protected]
49
Operational enhancements to the forest inventory program in BC
using airborne LiDAR
By Christopher ButsonForest Analysis & Inventory Branch (FAIB)
February 6th, 2020
Presentation Outline
1. BC Forest Inventory & Remote Sensing Overview
2. LiDAR Data Overview3. Calibration Library4. Current Project Update5. On the horizon…
6. Summary
VRI
Ground plot
Cost
Det
ail
and
/or
accu
racy
LiDAR
Delineation-Manual
Attribution-Photo
Digital air photo (0.3m)
LiDAR (0.1m)Attribution
Delineation-Semi
Calibration
Landsat(30m/15m)
Digital Camera(0.1m)
Delineation-Semi
Attribution-Photo
<$500,000 $500,000to $1 m
$2.5 m
$4 m
PFI
Digital Camera(0.1m)
Landsat 30m
2. LiDAR Data Overview• Industry driven collections for forest operations,
individual tree inventories, harvest planning & engineering.
• Point densities ranging from 1 pt/m2 to 12 pts/m2.• FAIB owns ~ 2.5 million
hectares (2011-2019) • BCTS owns ~ 6 million
hectares (2012-2019)• Other Licensees ~ 6 million
New LiDAR for forest inventory in 2018-2019
• Interior DF zone, Williams Lake & 100 Mile (850,000ha)• Boundary TSA (350,000ha)
3. LiDAR field calibration plot network:• 15 Mackenzie 2019
• 40 Haida Gwaii (Coastal) 2016
• 220 North Vancouver Island (Coastal) 2012
• 235 Kamloops/Okanagan (Interior) 2015
• 215 Cranbrook (Mountain Interior) 2017
• 60 TFL26 (Coastal) 2018• 200 Boundary (Interior) 2018-2019
LiDAR Calibration Plot Library
Reference data: • Ground sampling design
– Accessibility (<2km active road)
– Lidar pseudo density (GAP)
– Lidar pseudo height (P85)
– BEC stratification
• species specific crown architecture
– Structurally stratified random
– Systematic sampling intended to capture structural diversity
• Photoplot sampling design
– flight corridors 10km interval E-W
– Photoplot sampled at 5km along track
– Segments within ~500m selected for interp.
4. Current Project UpdateBoundary TSA: Predictive Forest Inventory (PFI)
258Reference data: • Ground sampling
– 197 modified CMI type-L (11.28m)
– Structurally guided sampling
– Quality GNSS
– Stem mapping
• Photoplot sampling
– 258 photoplot sample clusters
– 8,481 segments (µ: 3ha)
– 10cm RGBI stereo photo interpretation to VRI standards
– Treed attributes only
Boundary TSA – Tall trees
Stands delineated (n=480,223)
Treetops (n=145,372,791)
Tall tree inventory: 59.1m western larch, 1526459.5E 485664.5N (+/-50cm) epsg:3005
• LEFI specifications document created for use in Tree Farm Licenses (TFLs) and
Community Forests that have LiDAR and want to enhance the VRI.
FAIB is using three approaches to integrate LiDAR into existing VRI framework:
Tier 3 - Full integration, using calibration ground data, semi-automated delineation (structural & spectral) and photo sample estimation for species.
Tier 2 – More detailed integration using calibration ground data models.
Tier 1 – Simplest and fastest approach using CHM only.
VRI
Ground plot
LiDAR
New LEFI Tier 3 project…in
partnership with Tolko Industries Ltd. in TFL49 to
support 2022 Timber Supply Review (TSR). Area is 110,000 hectares and LiDAR acquired in 2017. Calibration data to be
collected in the summer of 2020.
5. On the horizon…
Interior Douglas Fir (IDF) zone in 100 Mile/Williams
Lake TSA’s. Area is 840,000
hectares. Objective is to:• Update forest inventory &
IDF management plan
FAIB will continue to acquire LiDAR to support forest inventory while also working toward the acquisition of provincial LIDAR data for application across the natural resource ministries.
Continue working with licensees, indigenous groups and contractors on creating & using Lidar enhanced forest inventories.
Keep developing the new hybrid forest inventory approach, titled the Predictive Forest Inventory (PFI), which incorporates methods found within previous inventory approaches: LVI, VRI and LEFI.
6. Summary
Questions?
[email protected] Analysis & Inventory Branch (FAIB)
Ministry of Forests, Lands, Natural Resource Operations & Rural Development