HR: 1340h
AN: G13B-0812    [Abstracts]
TI: Spatially-aware Processing of Large Raw LiDAR Data Sets
AU: * Strane, M D
EM: strane@email.unc.edu
AF: Department of Geological Sciences, University of North Carolina - Chapel Hill, Chapel Hill, NC 27599 United States
AU: Oskin, M
EM: oskin@unc.edu
AF: Department of Geological Sciences, University of North Carolina - Chapel Hill, Chapel Hill, NC 27599 United States
AB: An ultimate goal of LiDAR (LIght Detection And Ranging) data acquisition is to produce a regularly sampled accurate topographic view of the surface of the Earth. Last-return and inverse-distance weighted sampling of raw LiDAR data do not take into account the non-random distribution of raw data points. While elevation data produced by these methods is of high accuracy, gradients are not well-resolved and aliasing artifacts are produced, especially on low gradient surfaces. Because of the volume of data involved, resampling schemes that take into account the spatial distribution of raw data have been cumbersome to implement. We have developed a resampling method that uses the free open-source PostgresSQL database to store the raw LiDAR data indexed spatially and as its original time series. This database permits rapid access to raw data points via spatial queries. A robust and expedient algorithm has been implemented to produce regularly gridded resampled data with a least squares plane fit regression. This algorithm reduces aliasing artifacts on low gradient surfaces. The algorithm is also a proof-of-concept to show that complex spatially-aware processing of large LiDAR data sets is feasible on a reasonable time scale, and will be the basis for further improvements such as vegetation removal.
DE: 9820 Techniques applicable in three or more fields
DE: 8040 Remote sensing
DE: 1294 Instruments and techniques
SC: Geodesy [G]
MN: 2004 AGU Fall Meeting