Earth Science Applications of Space Based Geodesy
DES-7355
Tu-Th 9:40-11:05
Seminar Room in 3892 Central Ave. (Long building)
Bob Smalley
Office: 3892 Central Ave, Room 103
678-4929
Office Hours – Wed 14:00-16:00 or if I’m in my office.
http://www.ceri.memphis.edu/people/smalley/ESCI7355/ESCI_7355_Applications_of_Space_Based_Geodesy.html
Class 24
1
LIDAR
2
What is LIDAR?
LIght Detection And Ranging
3
What is LIDAR?
Optical remote sensing technology that can measure the
distance to, or other properties of, a target by illuminating
the target with light, often using pulses from a laser.
4
Uses frequencies in the near UV, visible, and near IR
5
LIDAR systems
Ground based
Airborne based
Space based
6
Types of LIDAR
Backscatter
Differential Absorption
Doppler
Flourescence
Raman
7
Backscatter LIDAR
most common
functions almost exactly like RADAR but at a different
wavelength
(typically looking at something solid)
8
Differential Absorption LIDAR (DIAL)
Calculation of molecular species in the atmosphere
Transmit pulses at two different frequencies determined by
the absorption line of the species you want to measure
(This and those that follow – typically looking up or
through gas to determine properties of something in gas)
9
Fluorescence LIDAR
induce fluorescence in the species you’re measuring
10
Raman LIDAR
Works on the principle of Raman scattering
11
Doppler LIDAR
Works on calculation of phase shift
Very similar to wind profilers
12
13
LiDAR – how it works
Each time the laser is pulsed:
Laser generates an optical pulse
Pulse reflected off object and returns to the system receiver
High-speed counter measures time of flight from start of pulse to
return pulse
Time measurement converted to distance (the distance to the target and the position of
the airplane is then used to determine the elevation and location)
Multiple returns can be measured for each pulse, up to 200,000+
pulses/second
Everything that can be seen from the aircraft is measured
14
Geodetic Applications
Backscatter used to locate (range, azimuth) of item
generating the backscatter.
15
How collected –
Airborne Laser
Swath Mapping
16
Airborne Laser Scanning
• ALS/LiDAR is an active
remote sensing technology that
measures distance with reflected
laser light.
• 1st developed in 1960 by Hughes
Aircraft inc.
• Modern computers and DGPS
make it practical.
• Typically used in very accurate
mapping of topography.
• New technologies and applications
are currently being developed.
Why use a Laser?
ALS systems take advantage of two of the unique properties of laser
light:
1. The laser is monochromatic. It is one specific wavelength of light.
The wavelength of light is determined by the lasing material used.
Advantage: We know how specific wavelengths interact with the
atmosphere and with materials.
2. The light is very directional. A laser has a very narrow beam which
remains concentrated over long distances. A flashlight (or Radar) on the
other hand, releases energy in many directions, and the energy is weakened
by diffusion.
Advantage: The beam maintains its strength over long distances.
3mrad divergence = 30 cm at 1 km and 1.5m at 5 km.
For an easy to understand discussion of how lasers work check out:
http://www.howstuffworks.com/laser.htm
ALS mapping concepts
• The position of the aircraft is known
(from DGPS and IMU).
• Measures distance to surfaces by
timing the outgoing laser pulse and the
corresponding return (s).
• Distance = time*(speed of light)/2
• By keeping track of the angle at which
the laser was fired: you can calculate the
X, Y, Z position of each “return”.
• Requires extremely accurate timing
and a very fast computer.
Basic ALS System Components
• Laser (usually NIR –1064nm)
mounted in an aircraft.
• Scanning assembly – precisely
controlled rotating mirror.
• Receiver for recording reflected
energy “Returns”.
• Aircraft location system
incorporating Differential GPS
and Inertial Navigation System.
• A very fast computer to
synchronize and control the whole
operation.
Two Distinct Families of ALS Systems
Waveform systems (a.k.a. large-footprint)
Records the COMPLETE range of the energy pulse (intensity)
reflected by surfaces in the vertical dimension.
Samples transects in the horizontal (X,Y) plane.
Waveform systems designed to capture vegetation information are not
widely available.
Waveform systems include SHOALS, SLICER, LVIS,
ICESat.
Discrete-return systems (a.k.a. small-footprint or topographic)
SAMPLES the returned energy from each outgoing laser pulse in
the vertical plane (Z) (if the return reflection is strong enough).
Most commercial lidar systems are discrete return, many different types
are available.
How collected –
Terresterial/Tripod
Laser Scanner
(TLS)
22
23
Advantages of LiDAR Technology
Provides a highly accurate means of elevation model
collection for 1’ or 2’ contours
Acquisition can take place day or night… shadows that are
problematic in mountainous areas are not an issue with
LiDAR
Unlike photography, acquisition can take place below cloud
cover… cloud shadows no issue
Very cost effective for larger projects
Does not provide break lines, nor is it imagery
24
Time to collect a million points –
Conventional Surveying: 15.5 years
Photogrammetry: 1.5 years
Lidar: 6.7 seconds @ 150 kHz
Lidar is expensive (as opposed to impossible)
25
Aircraft Requirements
Flying heights from 3,000 to 6,000 feet
Speeds ranging from 90 to 130 knots
Ability to carry equipment (relatively light), personnel, and
full fuel load
26
Aerial LiDAR System Components*
Aircraft
Scanning laser emitter-receiver unit
Differentially-corrected GPS
Inertial measurement unit (IMU)
Computer
LiDAR point data
colored by height
*components can be sources of error
Figures from McMcGaughey
USDA Forest Service--PNW Research Station
Scanning Mechanisms
Mechanism
sawtooth
Ground pattern
Most common
pattern (Leica,
Optech)
Figure modified from: Nikolaos 2006
What LiDAR is not –
Photography
We can shade the elevation and intensity data to create
“imagery”
Doesn’t capture breaklines
Doesn’t capture planimetric features
Advances in software may allow automatic feature
extraction soon
29
Traditional Photogrammetry vs. LiDAR
LiDAR
Day or night data acquisition
Direct acquisition of 3D
collection
Vertical accuracy is better
than planimetric*
Point cloud difficult to derive
semantic information;
however, intensity values
can be used to produce a
visually rich image like
product (example of an
intensity image)
Photogrammetric
Day time collection only
Complicated and sometimes
unreliable procedures
Planimetric accuracy is
better than vertical*
Rich in semantic information
*Complementary characteristics suggest integration
Why is LIDAR better than photogrammetry?
(It’s the trees)
Suppose timber allows 1
of 3 arbitrary rays to
reach ground; 1/3 of
ground can be surveyed
by LIDAR
Photogrammetry
requires 2 separate
views of a point; only
1/9 of ground will be
locatable
Shaded DEM
32
Cincinnati Airport – Aerial Photo
Overall Accuracy
(X,Y,Z) position of each return
50-100cm horizontal
10-15cm vertical
Ground surface (bare-earth surface)
What is the ground (grass, rocks, stumps)?
Tree heights
Underestimate tree heights by 0.5 to 2 m
Error is species dependent
LiDAR accuracy
Accuracy of elevation in range of 6 to 30 centimeters (0.20
to 0.98 feet).
Accuracy of XY position in range of 1.0 to 46 centimeters
(0.33 to 1.51 feet).
Accuracy depends on pulse rate, flying height, GPS
configuration, location of ground stations, and position of
the scanner with respect to nadir
35
LiDAR Data Characteristics
Raw return data are XYZ points (point cloud)
High spatial resolution
Laser footprint on ground ≤ 0.50 meters
Typical density is 0.5 to 20+ pulses/m2
2 to3 returns/pulse in forest areas
Surface/canopy models typically 1 to 5m grid
Large volume of data
5,000 to 60,000+ pulses/hectare
10 to100+ thousands of returns/hectare
0.4 to 5.4+ MB/hectare
36
Intensity and Multiple returns
Most units today have the ability to measure multiple
returns and the intensity of the returned signal for each.
This enables specialized applications using the LIDAR
data.
37
Multiple LIDAR Returns
Waveform LIDAR
This is the “waveform” graph of energy intensity
Notice that it follows the “shape” of the tree biomass
Return Density
In LiDAR the footprint size decreases with increasing post-spacing
and importantly the last return from a discrete return system is not
always the ground
LiDAR sensor systems vary in the number of returns from a surface
Figure Source: http://www.cnrhome.uidaho.edu/
Leaf-on vs. leaf-off
(A)
(B)
Cross section of LIDAR data through a single deciduous
tree (A) and coniferous tree (B) including bare-earth
returns. The green dots represent leaf-on returns and the
brown dots represent leaf-off returns
Waveform LIDAR: SLICER
Waveform systems have been used to accurately measure (r2 >.90) vertical
distribution of tropical and temperate forests (Lefsky 2001)
Discrete-return LIDAR
• Records data as X,Y,Z points.
• Spatial resolution is expressed
in terms of “post spacing” which is
the avg. horizontal distance
between points.
• Returns are “triggered” if the
laser reflects from a surface large
enough to exceed a pre-set energy
threshold.
• Minimum vertical distance
between returns is ~ 5 meters.
• New capability to record the
intensity of point returns.
Discrete Return Data:
Millions of X,Y,Z points
Area is approximately: 1 X 0.75mi.
includes ~ 440,000 returns
From Point Clouds to 3-D Surface Models
• Points are used to create 3D
surface models for applications.
• Triangular Irregular Networks
(TIN)s are used to classify the
points and to develop Digital
Elevation Models (DEM)s.
• Points must be classified before
use: “bare earth” points hit the
“ground”; other point categories
include tree canopy and buildings.
• Correct identification of “bare
earth” is critical for any lidar
mapping application.
Reconstruction
Building a 3D model from the point cloud
A polyhedral model uses edges and facets (like triangles), it should
“fit well” with the points
Types of models/approaches:
Meshes: many (small) triangles that use the points of the point cloud as vertices
(like a TIN but not just for a bivariate function)
Parametric models: pre-defined shapes where certain parameters can be set for
the best fit (e.g. the 6 planes of a house with a roof)
Hybrid methods
Reconstruction methods
Iso-surfaces and marching cubes
Medical imaging
DEMs, and parametric building models
Geosciences
Region growing and surface extraction
Urban reconstruction
"Bare Earth" model
Significant editing must be employed to create a “Bare
Earth Model” which models the natural ground.
Some automated procedures may be used.
Imagery backdrop may be necessary.
The 80/20 rule applies here as well.
In some cases, traditional photogrammetry may be
necessary to add breaklines
49
10m
USGS
DEM
DEM & Canopy Models
LiDAR
Strea
ms
IFSAR
Landsli
de
LiDAR
IFSAR
Strea
ms
Landsli
de
470’ Tall
Mississippi River
LiDAR is not only replacing conventional sensors, but also
creating new methods with unique properties that could not
be achieved before
57
Applications for Discrete-Return LIDAR
Digital Elevation Model (DEM)
4m DEM of PREF
Subset
USGS DEM (SRTM 1 sec?) vs LiDAR DEM
59
WV (MPBL source)
1-arc-second (30 meters)
1/3-arc-second (10 meters)
1/9-arc-second (3 meters)
60
LA (lidar source)
1-arc-second (30 meters)
1/3-arc-second (10 meters)
1/9-arc-second (3 meters)
Geomorphology
Mapping geomorphic features (Puget Sound LiDAR Consortium)
Wallace Creek, Carrizo Plain
63
Geology
Mapping Faults (Puget Sound LiDAR Consortium)
Other Applications
Measuring volume change in
open pit mines.
Utilities use lidar to map
power transmission line
curvature and clearance.
Discrete Return LIDAR for Forestry
•Can be used to measure
forest characteristics
including stand height,
biomass distribution, volume.
•DEMs are used in planning
for erosion and engineering
projects including roads.
•Mapping of forest stand
structure characteristics and
fire fuels conditions is under
development.
Finding faults with LIDAR in the Puget Lowland
Ralph Haugerud, Craig Weaver
U. S. Geological Survey
Jerry Harless
Puget Sound Regional Council
30 km
In some
places,
it is easy
to see
where
the
active
faults
are.
Seattle
Tacoma
30 km
In other
places,
it is not.
What are the salient differences?
SF Bay
area
Puget
Lowland
slip rate
3 cm/yr
strike slip
4 mm/yr
shortening
average tree
height
? 10 ft
? 100 ft
~106
18,000
years
years
age of
landscape
age  slip rate = feature size
18,000 yr  1 mm/yr = 18 m
106 yr  1 mm/yr = 1 km
In the Puget Lowland, to see a fault with the same
slip rate as in the SF Bay area, we have to look
more closely.
Start with traditional topo map.
Where is/are fault/s?
10-meter DEM from contours
12-ft DEM from LIDAR
PICTURE: OBLIQUE VIEW OF
S END ROCKAWAY BEACH
High-resolution LIDAR topography
Fly in winter, when leaves are off
Near-infrared laser; doesn’t penetrate clouds, rain
Errors
Largest are in angles—up to 1 m x-y error
Ranging error = ~15 cm z error!
2/3 of surveyed points on trees and buildings; remove with
automatic geometric filtering
Multiple reflections from one laser pulse = better filtering
Optimum working distance circa 1 km
Adequate reflection brightness
Keep laser eye-safe
Spot diameter: decimeters to meters
Spot spacing: 1 to 5 meters
Multiple passes
multiple look angles
higher point density
internal consistency check
$400 - $1,000 / mi2
15 km west of Seattle
Toe Jam Hill
fault scarp
Waterman Point
scarp
beach uplifted
during 900 AD
earthquake
landslides
southern Bainbridge Island
Carrizo Plain section of the
San Andreas Fault:
Paleoseismic trenches
provide access to
geomorphic and
stratigraphic records of
repeating earthquakes.
Fault-parallel trenches
provide access to intricate
geometries of
offset/deflected stream
channels.
0.5 m ALS-generated
DEM
Data collected by
NCALM for the B4
project, and processed in
OpenTopogaphy
Uses for high-resolution topography
Finding faults (earthquake frequency, kinematics)
Geologic mapping
Landslide hazards
Flood hazards, groundwater infiltration, runoff modelling
Fish habitat
?Precision forestry
?Noise propagation
Sources of Error
Acquisition
Processing
 Strip
adjustment
 Selecting ground points
 Thinning
Interpolation
Analysis/Visualization
Acquisition Scan Angle
LiDAR data should be
acquired within 18º of
nadir as above this angle
the LiDAR footprint can
become highly distorted.
Complex terrain can
exascerbate the problem
Strip Adjustment
Systematic Error (shifts & drifts)
- Wrong or inaccurate calibration of
entire measurement system (block specific)
- Limited accuracy of exterior
orientation (GPS- & IMU-related time- and
location-specific)
Result: Point cloud will not lie
on ground, but is offset in planimetric
view and height (10’s of cm)
For removing these discrepancies strip
adjustment algorithms require quantification of
these offsets at various locations
Improvements are needed in automatic tie
elements detection & 3D adjustments
Manual effort and labor are time consuming
Ditches & ridges are useful
Improves planimetric accuracy by about
40% and height accuracy by about
25%
Data correction and quality control tool
Overlap, across-track flight lines and ground
control are needed to fully adjust the
systematic errors
Create a seamless data set by correcting
for the systematic errors
Selecting Ground Points
Active area of research
Many algorithms
Project specific
Manual clean up necessary in most
cases
Result of 'slope threshold'
applied to an urban area (from
Vosselman 2000)
Getting Down to the Ground
89
Progressive Curvature Filter (Evans and Hudak 2007)
Filtering
Post ground point selection filtering is also performed to
reduce the size of the data sets
This type of filtering should only be applied in even terrain
Uneven terrain and densely vegetated areas are most
susceptible to removal of critical interpolation points
90
Terrain
Digital elevation model (DEM), digital terrain model (DTM):
“Ground”
Digital surface model (DSM): “top surface”
In open terrain, the separation surface between air and bare earth
DEM is different from measured laser points due to very different
reasons:
1 Filtering: classification of points into terrain and off-terrain
2 Basis for DTM generation, detection of topographic objects
91
Other Considerations
LiDAR derived DEM are not often hydro-corrected so
as to ensure a continuous downward flow of water (no
Digital Line Graph (DLG) hypsographic and
hydrographic data).
Water creates a natural void in LiDAR data and manual
addition of breaklines is necessary.
This type of processing is feasible with LiDAR data but it
adds cost
92
Visualization
Shaded relief & DEM
illumination can be used as a
simple visualization technique.
These methods are subjective.
Sensitive to hardware
parameters.
93
Further Analysis & Ground Validation
Once the DTM or DEM is available GIS can be utilized
for further systematic analysis and modeling
Accuracy assessment should always be attempted (best
approach is to do ground validation)
94
Issues with LiDAR data
LIDAR is indiscriminate… it places elevation points on
everything. This includes cars, houses, trees, etc.
LIDAR only places mass points, or random xyz points. It
does NOT pick up breaklines, or lines of abrupt change in
the ground elevation.
LIDAR is NOT imagery. LIDAR data can be shaded,
however, to offer a relief image
95
Reflectivity
Highly reflective objects may
saturate some laser detectors,
while the return signal from lowreflectivity objects may
occasionally be too weak to
register as valid
Minimum detectable object size
depends on reflectivity
A strong sunlight reflection off a
highly reflective target may
"saturate" a receiver, producing an
invalid or less accurate reading*
*most acquisition is done in a preferred range of angles to avoid this issue
Dust & Vapor
Laser measurements can be weakened by interacting with
dust and vapor particles, which scatter the laser beam and
the signal returning from the target
Using last-pulse measurements can reduce or eliminate this
interference
Systems that are expected to work in such conditions
regularly can be optimized for these environments
Background Noise and Radiation
The laser is not affected by background noise
Most systems determine baseline radiation levels to ensure
that it does not interfere with measurements
Cincinnati Airport – Intensity Plot
Intensity Image
Commonly unused bi-product
of a LiDAR acquisition and is
the intensity of object that the
laser pulse is striking. This is an
uncalibrated 8-bit (0-255)
image that is ortho-rectified as
therefore can be used as an
orthophoto
Not typically used in
quantitative analysis as image
gains always set to 'adaptive
gain' setting when images are
acquired
What are some of the LIDAR data products available?
Digital Ortho-Rectified Imagery
Some LiDAR providers collect digital color or black-and-white ortho-rectified imagery
simultaneously with the collection of point data. Imagery is collected either from digital
cameras or digital video cameras and can be mosaiced. Resolution is typically 1m.
Intensity Return Images
Images may be derived from intensity values returned by each laser pulse. The intensity
values can be displayed as a gray scale image.
LIDAR Derived Products
Topographic LiDAR systems produce surface elevation x, y, z coordinate data points.
There are many products that can be derived from raw point data. Most LiDAR
providers can derive these products upon request:
 Digital Elevation Models (DEMs)
 Digital Terrain Models (DTMs) (bald-earth elevation data)
 Triangulated Irregular Networks (TINs)
 Breaklines - a line representing a feature that you wish to preserve in a TIN
(example: stream or ridge)
 Contours
 Shaded Relief
 Slope & Aspect
http://calm.geo.berkeley.edu/ncalm/ddc.html
102
103