Propagated Uncertainty for Horizontal Ground Motion Derived from Multi-Temporal Digital Elevation Models

Quantifying measurement uncertainty is an important component of geospatial data analysis, including spatial displacement measurements computed from multi-temporal digital elevation models (DEMs). Uncertainty estimates provide context to the validity of reported measurements and enable hypotheses on the statistical significance of observed spatial motion. Although error propagation is sometimes applied to simple differencing of DEMs to generate uncertainty estimates in the computed vertical change, it is virtually non-existent in automated 2D and 3D change detection methods. We report on the performance of rigorous forward error propagation of source data uncertainty, i.e., Jacobian-based variance propagation, through an image correlation technique applied to DEMs to measure horizontal ground motion. We provide an example application and conclude with a brief outline of future work required to fully validate and automate the method.