HR: 0830h
AN: G11A-0239    [PDF]
TI: Operational Accuracy of LIDAR in Mountainous Terrain
AU: * McKean, J
EM: jmckean@fs.fed.us
AF: USDA Forest Service Rocky Mountain Research Station, 316 E. Myrtle St., Boise, ID 83702 United States
AU: Roering, J
EM: jroering@oregon.uoregon.edu
AF: University of Oregon Department of Geological Sciences, 1272 Univ. of Oregon, Eugene, OR 97403-1272 United States
AB: Conventional airborne LIDAR accuracy assessments are done by surveys of very flat surfaces: commonly tarmac areas or the roofs of hangar buildings at airports. These tests characterize the errors of the LIDAR hardware/software system including the laser ranging unit, opto-mechanical scanner, data recording devices, and aircraft GPS/INS. However, surveys over horizontal surfaces are not sensitive to data positional errors. Also, tests of the LIDAR detection of the positions of edges of horizontal surfaces are often unrealistic if the test flight pattern is designed to enhance discrimination of those edges. Furthermore, little information is available about the operational accuracy of LIDAR surveys in mountainous terrain where all the system errors combine with uncertainty related to general topography and local surface roughness to produce the total three-dimensional survey error. We have evaluated the operational LIDAR accuracy at a large landslide complex that has an extreme variety of topographic roughness. LIDAR data were compared with total station and RTK GPS ground points and profiles. Over gently sloping, smooth terrain adjacent to the landslide, single point elevation absolute errors were generally $<$15 cm, similar in magnitude to the system errors normally reported by LIDAR contractors. There were local vertical discrepancies of up to 45 cm, presumably due to temporary GPS errors. Within portions of the landslide that are rough over a spatial scale of meters to tens of meters, the point elevation median absolute accuracy remained better than 15 cm, but the range of errors increased to plus-or-minus several meters. Relative elevation accuracy was evaluated by comparing gridded surfaces of small identical areas measured twice by the LIDAR in the overlap zones between adjacent flight lines. Relative differences in sample area average elevation were less than 60 cm over smooth unfailed terrain, increased to 1 m over an active fine-grained earthflow, and 3 m in a portion of the landslide complex that included large angular limestone blocks in the slide matrix. The relative differences in standard deviation of elevations within the sample areas also increased from about 30 cm in the smooth area outside the slide to up to 2.2 m in the blocky portion of the slide complex. We were not able to distinguish LIDAR data elevation errors from positional errors. Work is continuing to quantify the absolute and relative accuracy of local slope and topographic curvature predicted by the LIDAR data.
DE: 1294 Instruments and techniques
DE: 1824 Geomorphology (1625)
SC: Geodesy [G]
MN: 2003 Fall Meeting