Deriving Land and Water Surface Elevations in the Northeastern Yucatán Peninsula Using PPK GPS and UAV-Based Structure from Motion

Abstract

Abstract While UAV-based imaging methods such as drone lidar scanning (DLS) and Structure from Motion (SfM) are now widely used in geographic research, accurate water surface elevation (WSE) measurement remains a difficult problem, as water absorbs wavelengths commonly used for lidar and SfM feature matching fails on these dynamic surfaces. We present a methodology for measuring WSE in a particularly challenging environment, the Yucatán Peninsula, where cenotes – exposed, water-filled sinkholes – provide an observation point into the critically important regional groundwater supply. In the northeastern Yucatán, elevations are very close to sea level, the area is of low relief, and the near-vertical edges of the walls of the cenotes complicate the use of the so-called “water edge” technique for WSE measurement. We demonstrate how post-processing kinematic (PPK) correction of even a single Real Time Kinematic (RTK) Global Positioning System (GPS) unit can be used to finely register the SfM-derived point cloud, and present evidence from both simulations and an empirical study that quantify the effect of “dip” in SfM-based environmental reconstructions. Finally, we present a statistical analysis of the problem of “thick” or “fuzzy” point clouds derived from SfM, with particular emphasis on their interactions with WSE measurement.