TOWARDS ACCURATE BASALT LAVA FLOW BEDDING ATTITUDE MEASUREMENTS FROM AERIAL LIDAR DATA, LAPWAI DRAINAGE BASIN, IDAHO, USA

Accurate bedding attitude measurements can be difficult to obtain in a number of stratified lithologies. Lava flow units are commonly particularly challenging because they are typically massively bedded, and the few bedding features that do occur are commonly highly irregular such as surfaces bounding flow-top breccias. In these cases, traditional field measurements with a hand-held transit are very imprecise, and transit precision degrades even further for shallowly dipping strata. We have faced these issues in the Lapwai Creek, Idaho drainage basin, which is characterized by multiple semi-confined aquifers that occur in gently folded lava flows of the Miocene Columbia River Basalt province. The shallowly dipping strata strongly influence groundwater gradients, and field-based bedding measurements are too imprecise to use in developing an effective groundwater model. Our approach has been to improve bedding measurements by using aerial Light Detection and Ranging (LiDAR) data for the drainage basin that was recently obtained by the Nez Perce Tribe. The projected LiDAR data permit us to view basalt bedding from a distance and average irregular bedding surfaces. The >1 m vertical variance resulting from bedding surface roughness far exceeds the 11 cm 2σ absolute deviation in the LiDAR data. For a bedding attitude calculation, we select points at which bedding surfaces intersect topography within an analysis area of ~200 m². This area is large enough to minimize error due to bedding surface roughness, yet is significantly less than ¼ wavelength of folds in the region. Points are selected along multiple segments that are defined as the trace of a bedding surface across a slope of constant aspect. Two approaches are used to calculate a best fit plane representing bedding attitude in an analysis area. The first employs a planar least squares regression for all points in an area, and assumes that points along all segments are obtained from the same bedding surface. The second method involves a spherical distribution analysis of lines calculated between all pairs of points within each segment, and assumes only that points obtained from each segment are along the same bedding surface. Mean square errors for best fit planes average 2.4°, a considerable improvement on field-based methods for poorly bedded lithologies.