High-resolution seafloor mapping on the U.S. Northeast continental shelf is increasingly used for geologic interpretation and offshore project planning. As grid resolution tightens toward the decimeter scale, the limiting factor often shifts from “can we see it” to “can we trust its position and change through time.” 

Autonomous underwater vehicles (AUVs) are changing how we map. Operating sensors close to the seabed, AUV surveys can produce very high-resolution data with minimal topside infrastructure and flexible deployment from smaller vessels. Because the vehicle is subsurface, AUV operations can also be less sensitive to some weather-driven surface effects that complicate acquisition. Bedrock’s AUVs carry a full geophysical payload, multibeam echosounder, side-scan sonar, sub-bottom profiler, and magnetometer, providing co-registered seafloor and subsurface context within a single survey line.

These benefits come with an uncertainty structure that differs from conventional surface-vessel surveys, and that difference matters for interpretation. Surface surveys typically have continuous GNSS positioning, so horizontal uncertainty is often comparatively stable while vertical uncertainty is driven by sound speed or acquisition parameters. Many AUV operations rely on inertial navigation aided by a DVL, with GNSS only during intermittent surface intervals and sometimes without acoustic positioning. This creates time-dependent horizontal error growth between surfacings that can dominate map usability for feature localization even when vertical uncertainty looks acceptable.

We present high-resolution AUV multibeam seafloor maps paired with uncertainty and accuracy analysis across diverse conditions. Using orthogonal check lines and comparisons to surface surveys, we evaluate how uncertainty evolves with underwater interval duration and survey geometry. Results show that AUV surveys can meet stringent hydrographic standards, (e.g. International Hydrographic Organization Special Order) but the uncertainty products cannot be assumed. 

We propose a practical framework for geologic use of AUV seafloor data that centers validation rather than paperwork. Treat Total Horizontal Uncertainty and Total Vertical Uncertainty layers as hypotheses to be tested, and require repeatability evidence at minimum (cross-check lines), with stronger validation coming from repeat surveys against an external source of truth, to demonstrate that mapped uncertainty reflects observed system accuracy. We suggest applying the same expectation to surface surveys, as uncertainty products that are not validated can still be confidently wrong.