Airborne LIDAR
The Technology
Frank L.Scarpace
Professor
Environmental Remote Sensing Center
Civil and Environmental Engineering
University of Wisconsin-Madison
Slides adapted from a talk given by
Mike Renslow - Spencer B. Gross, Inc.
PRESENTATION OUTLINE
•
•
•
•
•
•
Background of LIDAR
Brief Technical Description
Quality Control/Assurance Procedures
LIDAR Data Characteristics
Processing LIDAR Data (2 Steps)
Data Set Characteristics
BACKGROUND
• LIDAR (Light Detection And Ranging)
– 30 Year Old Technology
– Became Cost Effective Very Recently
• System Components
– Laser Scanner, ABGPS, IMU, Precise Clock
(Multiple Planes of Reference)
– Robust Computer Support
– Requires Calibration (Bore Sighting)
BACKGROUND
• Capacity to Capture Multiple Return Values /
Pulse
– Derive many, many X, Y, Z Values
– Positional Data and Intensity Data
• Multiple Configurations Possible
– Remarkably Large Data Files
• Accuracy
– Standard Deviation 15-20 cm
– Vertical RMSE at 20 cm on Discrete ‘Hard Hit’ Points
– Horizontal Accuracy at 2X the ‘Footprint” Size
Rotating Mirror Scan Pattern
Oscillating Mirror Scan Pattern
LIDAR has Multiple Return
POINT CLOUD OF ALL LIDAR POINTS IN DOUGLAS FIR FOREST
LIDAR &Terrain Interaction
• For example; a calm still lake, will only
reflect energy back within a few degrees of
the nadir beam of the laser.
• A “wavy” lake on the other hand, will
reflect energy back from wider incident
angles.
• Diffuse surfaces (ground or tree) reflect
energy back omnidirectionaly.
LIDAR Intensity Collection
Laser Intensity Raster - Detail
TIN surface of Raw LIDAR Data
‘Raw’ FIRST Return LIDAR Data
Raw LAST Return LIDAR Data
Automatic Vegetation Removal
• Automatic programs begin the noise and
vegetation/surface feature removal process
• These remove approximately 80% of
vegetation (depending on the land cover and
terrain characteristics)
• This part typically uses about 20% of the
vegetation removal time budget
Trend Surface Analysis
Green Points = Vegetation
Brown Points = Trend Surface
Before
...after
Manual Editing
• Final vegetation and feature removal requires
manual intervention.
• Custom selection routines are used in 3D and
GIS Software to analyze the data and identify
target points.
• Accurate interpretation of the LIDAR data
requires supporting imagery.
• Removal of the remaining 20% of the
vegetation and features will account for about
80% of the time budget
...after
…final
LIDAR vs. Traditional Mapping
1”=100’ Scale Terrain Mapping Example
 Compiled Mass Points are more widely
spaced: 60 feet vs. 12 feet.
 Compiled DTMs use breaklines; LIDAR usually
does not (breaklines can be added from
photogrammetric techniques).
 Compiler can place points; LIDAR is
indiscriminate.
 Compiler must be able to SEE THE GROUND,
LIDAR is self-illuminating & ‘looks’ down into
the vegetation.
Typical Wooded Area Example
Detail with LIDAR Ground Points
Processed TIN Surface
DEM and Contour Generation
• Contours are a cartographic construct used to
visualize topography.
• Contours produced directly from the LIDAR
TIN are usually not aesthetically pleasing.
• LIDAR data can be converted into a DEM Grid
at the nominal post spacing which retains
fidelity to the original data and which
appropriately smoothes the contours.
Contours Generated from the DSM
Contours Generated from the DEM
Conclusions
• LIDAR is a powerful new technology for
determining terrain elevations.
• There are still questions as to the horizontal
accuracy.
• Appears to be a good companion
technology to the existing photogrammetric
methods of measuring terrain.