Airborne and terrestrial LiDAR data and methods have been demonstrated extensively for application to landslides and other earth surface processes. Fusion of these datasets provides new opportunities for monitoring change, quantifying mass transfer, and ultimately understanding the governing earth surface processes. Yet numerous challenges need to be considered to accurately model dynamic environments with LiDAR datasets. We present example considerations and solutions related to co-registration methods and errors of multi-temporal (from time series airborne LiDAR) and multi-source data (from fusion of airborne and terrestrial LiDAR). We demonstrate the spatially-explicit estimation of uncertainty of measured change, and the scale dependence of geomorphic features across a landslide using a slope-based surface matching and an adaptive iterative closest point (ICP) matching methods. Using such point-based registration allows for characterization of very fine-scale surface movements and associated co-registration uncertainties in the x, y, and z directions. We also demonstrate robust approaches to model the distribution of the errors across scales and their propagation during the surface analysis process using an ensemble learning method. Such error propagation model is essential for understanding and quantifying stochastic and systematic uncertainties associated with multi-source and multi-temporal LiDAR data for earth surface characterization. We will also discuss the upcoming zcloudtools project, a source for LiDAR analysis tools including tools used for change detection.
Available at: http://works.bepress.com/joseph_wheaton/77/