R3 TREES - Integrated Management of Urban Green Areas
GISrock1_DG_REwork
1. Team 1-2: Christopher Gray, Amanda Greenwalt, Sarah Rollins, Paige Terry, Ryan Witsell
Team 3: S Holly Russon, Nori Gilo,
Team 4: Nori Gilo, Connor Holmes, Holly Russon
Team 6: Michael Bartolo, Shelley Bayard de Volo, Allison Cech Murray
Team 7: Jason Nau and Tucker Werbach
Team 8: Melissa Daniels, Nick Hobson, Aaron Siegel
Integrated LiDAR, Aerial, and Satellite Data Analysis
for
Municipal Project Support
Instructor/coordinator: Peter Price, GISP FRCC-Boulder County Campus
Presenter: Valerie Dooley, Student Lab Assistant & Image Analyst
2. Integrated LiDAR, Aerial, and Satellite Data Analysis
for
Municipal Project Support
This Service-Learning project is an ongoing
cooperative effort between the City of Loveland,
Colorado, Department of Information Technology,
Application Services Division and the students and
faculty of the GIS Department at Front Range
Community College, Remote Sensing Class of Spring.
2015.
Special thanks to the
DigitalGlobe Foundation for
providing the WorldView2 and
WorldView3 imagery which was
absolutely critical to the success
of the project.
3. Integrated LiDAR, Aerial, and Satellite Data Analysis
for
Municipal Project Support
1.) Process and analyze a large volume of
data (Over a Terabyte) utilizing the technical
and student resources available.
2.) Develop workflows for extracting tree
canopy, height, and other characteristics.
Combine this tree data with municipal right-
of-way boundaries.
3.) Examine the possibility of applying the
data to the extraction of building footprints
and heights, drainage basin and channel
modeling, and ditch network configuration.
Project Goals - Municipal Partner:
4. Integrated LiDAR, Aerial, and Satellite Data Analysis
for
Municipal Project Support
1.) Provide a high quality learning experience for the class.
2.) Help students: Master basic and advanced skills for
image processing and analysis.
Integrate the products of analysis in a GIS environment.
Learn methods for recording processes and results.
3.) Give students the opportunity to: Experience a team approach to a technical challenge.
Create, present, and deliver professional-level materials that document their work.
4.) Further develop client relationships with the college and a compile a data archive that
will result in future service-learning projects and challenging class applications.
Project Goals - Student & Educational:
5. Project area: Loveland municipal boundary on LiDAR DEM
Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
6. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Examples of WV 3, Ortho Aerial,
and LiDAR (processed for height
above ground) - key data types
for this analysis.
7. Examples of
LiDAR products
used in project
analysis.
1. Height derived
from DSM minus
Bare Earth.
2. Point heights
from LiDAR
returns 2-5
3. Profile from all
returns (1st
returns in blue)
Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
1 2
3
8. Basic data from
WV3, Ortho, and
LiDAR height
with building
footprint overlay
Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Team Results – Complex Workflow
9. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Team Results – Complex Workflow
1.) Tree height
slope focal stats
on
LiDAR height
with building
footprint overlay
1
2
2.) Trimmed
Thiessen polygons
for trees
on
LiDAR height with
building footprint
overlay
10. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Team Results – Complex Workflow
1
2
2.) Thiessen Split Canopy center points
trimmed and buffered
on
Orthophoto with building footprint
overlay
1.) Thiessen Split Canopy Circles
Minimum Bounding Geometry
on
LiDAR height with building footprint
overlay
11. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Team 1 & 2 Canopy Footprints based on Tree Crown Center points that were found by finding the local minima in the
slope of the Height Raster. The green circles represent the estimated canopy footprint defined by local Thiessen
Polygons.
13. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Canopy height estimations
utilizing the Height derived
from DSM minus Bare Earth,
which is adjusted for
elevation.
Tree canopy rasters are clipped
to the municipal right-of-way
rasters giving us only what
trees fall within Loveland’s
municipal right-of-way.
14. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Multi-canopy estimations
estimating how many individual
tree canopies make up the
larger canopy areas.
Team 8 came up with 625 sq.
feet as the normalized tree
canopy average.
They took this 625 sq. feet and
divided the polygon areas by it
in order to create an estimated
tree count.
15. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Another aspect of
this project was
utilizing the 30cm
WV3 imagery from
October 2014 in
order to try and
differentiate tree
species.
I used multiple
band
combinations for
running iterations
of the
Classification
Workflow tools in
ENVI in order to
select out the
green, red and
turning trees.
16. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Here is a close up of the
level detail I went down
to while working these
classifications. The more
precise you are with
your pixel selection the
more precise the
classification will turn
out.
For this classification I
used a spectral subset of
bands 3, 4, 5 & 6 which
are Green, Yellow, Red,
and Red Edge. Out of all
the iterations I ran with
all the different band
combinations that I
tried, this was the best
one for finding the red
trees.
17. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
This is the product of the
classification. You can see
the different classes here,
some did better than
others - the red trees and
turning grass
classifications worked
really well. The others
classes ranged from
medium results to
straight up bad - An
Example of a bad result is:
the green trees category
didn’t really pick up the
green trees but it picked
up everything else,
including concrete and
bare earth. I have come
to the conclusion that
each classification type
needs its own workflow.
18. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Here are the classes that
worked the best. You
can see these areas that
are covered in different
stages of living and
dying grass. One future
possibility is to be able
to use this type of
imagery classification to
map irrigation patterns
and improve municipal
maintenance schedules
while improving water
conservation. We
discussed these results
with the city and they
were very interested in
looking into processes
that would allow them
to monitor and improve
municipal irrigation
practices
19. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
As we zoom
further into
this image
you can start
to see the
details better.
You can see
how well the
red tree class
covers the
red trees in
this area
excluding the
turning trees
and other red
objects not
trees - such
as cars and
red roofs.
20. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Just a note here that what
you are looking at is 30 cm
resolution pansharpened
imagery from the
WorldView3 Satellite. FRCC
received a grant from
DigitalGlobe for 8 band
multispectral + Panchromatic
imagery.
I ran my classification
workflows on the native color
image but pansharpened it
for this presentation.
The only reason I was able to
perform such detailed
classification is because of
the high resolution imagery
without it there would be no
way I could have achieved
this level of accuracy.
22. Integrated LiDAR, Aerial, and Satellite Data Analysis
Challenges1.) Data Volume:
289 LiDAR tiles just for Loveland Municip.
LiDAR tiles alone were > 1TB in size
Because of large files and our limited processing accesses we are only able to process a
maximum of 4 tiles at a time.
2.) Limited Processing Capabilities:
As a smaller college computing resources are limited.
It can take multiple hours to process 4 tiles. Ex: Spatial Joins
3.) Complexity of Data Models and Workflows:
Need streamlining More sophisticated users could take core work and adapt it. This is the
purpose of educational projects anyway.
4.) Tool Limitations:
Couldn’t afford to buy the specialized ENVI or ArcMap Tools and we weren’t too successful
with advanced experimentation with the free tools. Another problem is that we tried a
number of tools and none were user friendly enough for students to handle.
We are very encouraged by the inclusion of sophisticated LiDAR tools in ENVI 5.3 and we are
looking forward to applying them to this project.
24. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
An Example of raw
LiDAR point data
Color is Elevation
based on Mean Sea
Level (so - not
adjusted for
elevation)
8 Classes:
Orange: 4980.74 Ft.
AMSL to 4996.71 Ft.
AMSLf
Mustard Yellow:
4964.77 Ft. AMSL -
4980.74 Ft. AMSL
Green: 4948.8 Ft.
AMSL - 4964.77 Ft.
AMSL
25. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Same raw LiDAR
point data rendered
for “face elevation
with graduated color
ramp.”
You can see much
more detail coming
out in this image.
Topography is
starting to emerge.
26. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
Here we zoom into
the raw LiDAR point
layer and start to
see the individual
returns. This is still
classified by height
not return. You can
start to see things
like building
outlines and local
features.
You can start to see
how complicated it
is to derive usable
data from these
points.
27. Integrated LiDAR, Aerial, and Satellite Data Analysis for Municipal Project Support
As you zoom
farther into the
raw LiDAR point
layer you can see
the track that the
airplane took
while shooting
lasers out the side
of the craft.
I’ve been looking
at this data for so
long that I can tell
that the kind of
“blob” of LiDAR
points in the
middle are the
returns made by a
leaf off deciduous
tree.