RECENT ADVANCES APPLYING TERRESTRIAL LIDAR TO EARTHQUAKE
LANDSLIDE RESEARCH
Robert Kayen(USGS, Menlo Park, CA & UCLA, USA) Scott Ashford,
(Oregon State University, USA) Jon Stewart (UCLA, USA) Brian
Collins (USGS, Menlo Park, CA, USA) International Conference in
Commemoration of 10th Anniversary of the Chi-Chi Earthquake, 2009.
Jungli, Taiwan, 21 September 2009 RECENT ADVANCES APPLYING
TERRESTRIAL LIDAR TO EARTHQUAKE LANDSLIDE RESEARCH
Terrestrial LIDAR Technology (Light Detection and Ranging) Methods,
Metrics, & Overcoming limitations Landslide analysis for change
detection: (Daly City, California, USA) Blind Comparison of Air-
and Terrestrial-LIDAR Accuracy: (Los Angeles Reservoir, California,
USA) High resolution change detection using LIDAR: (PARI
Liquefaction site, Ishikari, Hokkaido, Japan) LIDAR LIDAR (Light
Detection and Ranging)
Laser pulse backscatter off landslide Scanning mirror directs laser
beam Distance = (c x Tflight)/2 Rapid and SAFE Captures complexity
of surfaces Geometric analysis of landslide Change-detection
Permanent virtual archive Niigata Chuetsu, 2004 Laser Pulse RGB
Color Since the laser pulse travels at the speed of light, accurate
timing is necessary to obtain fine vertical resolution.For example,
a one nanosecond timing resolution allows a range measurement
accuracy of about 30cm.Timing technology allows for measurements of
< 5cm Scanning mirror directs laser pulse back and forth across
track.In shoals the track is an arc. Typical Terrestrial LIDAR
Setup
RTK-DGPS Base RTK-DGPS Rover Photogrammetric Camera Laser Benchmark
USGS Terrestrial LIDAR
Laser : Reigl z m z420i -1000m range Scan-to-scan registration
Reflector Registration Bare Ground Error Budget: Laser rangefinder
error = cm GPS error ~1 cm+ Registration Error = 2-5 cm
EstimatedRMSE cm vertical Heavily Vegetated Ground: Significant
positive residuals associated with inability to extract true bare
earth model Processing Procedures
B: Register. A: Scan Niigata Ken, Japan D: Quantify Deformations C:
Filter & Fuse Surface 5.69m multiple scans eliminate shadow
zones
Embankment Failure, Rt 252 Niigata, Japan multiple scans eliminate
shadow zones (2 false-colored scans in white and red) s Overcoming
Shadow-Casting Obstacles
High-altitude (20 m)Terrestrial LIDAR Downward view of
terrain/vegetation Improved grazing angle Flat terrain max range =
H / tan 3-5 USGS Space Needle GEOREFERENCED REFLECTOR
REGISTRATION:
RTK-DGPS registered best fit to common reflectors. BENCHMARK BASE
& Reflector Laser 1 - position Laser 2 - position Nantou
County, Taiwan Composite Lidar Data: Merged scans from all
orientations to eliminate shadows 100m Lake Sinizzo, Abruzzo, Italy
April, 2009 10 Scans Landslide analysis for change detection: (Daly
City, California, USA) 2007 Landslide reactivation of 2003 slide
site in Merced Frm
2007 Landslide reactivation of 2003 slide site in Merced Frm.
Weakly cemented sandstone with some slope failure in 1989
earthquake. Pre-2007 Event Data, (2003) Post-Event Data, Jan. 2007
Plan View
Oblique View Post-Event Data, Jan. 2007 Lidar Surfaces for change
assessment using cross sections GIS-map of lidar surface changes in
2007 (120,000m3, 1/3 of 2003 event)
Oblique photo and change map Relative multi-epoch vertical accuracy
(pre- and post-event imagery).
Blind Comparison of Terrestrial & Airborne LIDAR to Evaluate
Potential Landslide Displacements at Los Angeles Dept. Water &
Power Facilities Relative multi-epoch vertical accuracy (pre- and
post-event imagery). Coincident measurements with Total Station.
Assessment of bias and dispersion. (Collaboration between MCEER,
LADWP, UCLA, USGS) LADWP Study Site: Los Angeles Reservoir, Van
Norman Complex San Fernando Valley, CA Trench Study at Los Angeles
Reservoir is intended to evaluate bias and dispersion without
influence of vegetation Pre-Trench Post-Trench Post-Trench 120 cm
Benching at 10, 20, 40, 60, 80, cm Laser Rangefinder: Optech ALTM
3100 1064nm w/ integrated IMU & GPS.
Airborne LIDAR Laser Rangefinder: Optech ALTM 3100 1064nm w/
integrated IMU & GPS. Flight 1: Fixed wing at 900m (Airborne1,
El Segundo, California) Flight 2: Helicopter at 210m (Terrapoint,
The Woodlands, Texas) Laser rangefinder error = 2-3 cm GPS error =
5-10 cm IMU error = ~9 m Vendors quoted RMSE 10-18 cm vertical (95%
CI) m horizontal (1s) Trench Study at Los Angeles Reservoir:
Terrestrial vs. Airborne Data
Terrestrial Pre-Trench Terrestrial Post-Trench 900m Airborne1 210m
Terrapoint Airborne Pre-Trench Airborne-Trench RMS Elevation Error
Zlidar,i ZTS,i Bias - Residual Mean Dispersion - Residual Std. Dev.
For a large sample group (Plane) (Helicopter) RMSE=21.8 cm
RMSE=33.8 cm RMSE=6.8 cm RMSE=4.3 cm Los Angeles Reservoir
Trench
Normalized residuals Correlation Coefficient (uncorrelated) (weakly
correlated) Elevation change residuals measures bias and dispersion
in estimating topographic change
R,i = LIDAR,i - TS,i= (Ztl,i - Zt2,i )LIDAR - (Ztl,i - Zt2,i )TS
R,i = Rtl,i - Rt2,i RMSE = 8.5 cm RMSE = 22.6cm Blind Comparison of
Airborme and Terrestrial LIDAR
Terrestrial LIDAR had almost no bias with pre- and post-trench
Airborne LIDAR had negative bias in both pre- and post-trench = cm,
and = cm (errors exceeded both vendors specifications) Terrestrial
data had 1/4-1/10 mean error and 1/1.4-1/5 dispersion of airborne
data. Multi-epoch elevation change terrestrial residuals were 1/3
of airborne residuals. 2009 Journal of Surveying Engineering, ASCE
PARI Blast-Liquefaction Experiment: Ultra-fine scale/high precision
change detection using Terrestrial LIDAR PARI Experiment Tripod
Scans RTK-DGPS Reflector Registration Elevated Scans PARI Ishikari,
Hokkaido Test site, Japan, 2008
No Data/Shadow Elimination by Merging Data of Different
Orientations PARI Ishikari, Hokkaido Test site, Japan, 2008 60m 4
Scans PARI Experiment: Change maps of Improved and Unimproved
Ground on Tarmac area(5 cm contours)
BLAST DAY 2 DAYS AFTER BLAST 100 m Vibro-Compaction/replacement
Chem. Grout 4 DAYS AFTER BLAST 5 Months AFTER BLAST RECENT ADVANCES
APPLYING TERRESTRIAL LIDAR TO EARTHQUAKE LANDSLIDE RESEARCH
Terrestrial LIDAR Technology (Light Detection and Ranging) Methods,
Metrics, & Overcoming limitations Landslide analysis for change
detection: (Daly City, California, USA) Blind Comparison of Air-
and Terrestrial-LIDAR Accuracy: (Los Angeles Reservoir, California,
USA) High resolution change detection using LIDAR: (PARI
Liquefaction site, Ishikari, Hokkaido, Japan) RECENT ADVANCES
APPLYING TERRESTRIAL LIDAR TO EARTHQUAKE LANDSLIDE RESEARCH
Robert Kayen(USGS, Menlo Park, CA & UCLA, USA) Scott Ashford,
(Oregon State University, USA) Jon Stewart (UCLA, USA) Brian
Collins (USGS, Menlo Park, CA, USA) International Conference in
Commemoration of 10th Anniversary of the Chi-Chi Earthquake, 2009.
Jungli, Taiwan, 21 September 2009 34 Google Earth Visualization of
Lidar Analysis Airborne LIDAR Error Budget
Laser rangefinder error = 2-3 cm GPS error (single point moving
measure) = 5-10 cm IMU error = ,000 ft Usually flown around 2, ,000
ft Yields error of 5-10 cm LA Res. Flights were at 2000 (pre) and
700 (post) (i.e., as low as 2-5 cm error) Most vendors quote 15-18
cm vertical accuracy 0.5-1 m horizontal (depends on height)