1. Lawrence H. Haselmaier, Jr. Computer Scientist Naval Oceanographic Office 7 August 2014

2. The Problem Autonomous Underwater Vehicle (AUV) multibeam (MB) data require vertical corrections for : assessing internal consistency inclusion in hydrographic products Current options include: Predicted Tides – (potentially inaccurate) Observed Tides – (can be difficult/impossible to obtain)

3. An Example When NAVOCEANO attempts to assess AUV MB data accuracy, we compare it to surface vessel data collected over the same area. No observed tides available Predicted tides applied to everything AUV data collected 11 days after ship data

4. An Example, Continued AUV indicated bottom ~ 0.4 m shoaler than ship In approximately 50-m depths Violates error budget for IHO Special Order Jeopardizes Order 1 Second set of ship data collected (11 days after first) New ship data also shoaler than previous data By approximately the same amount Initial conclusion: predicted tides in this area are not useful for such a comparison.

5. Proposed Solution Applying a tide corrector derived from ellipsoidal measurements to AUV MB data would reduce the uncertainty present with predicted tides. Requires a continuous ellipsoid height (EH) data source to compute the solution. Theoretically, EH data could come from another nearby vessel.

6. Ellipsoidally Referenced Survey CS=RS + D - T (conventional correction) CS=RS + SEP – h + Z[a] (ERS correction) AUV includes depth corrector from surface Reference Ellipsoid Bottom Chart Datum RS MRP D ZaZa Water Surface Heave Datum heave CS ZtZt T h SEP

7. Advantages of ERS Allows for more rapidly assessment of data quality Can isolate other issues in real/near-real time Reduces reliance on shore-based tide gauges Only requires one period of gauge installation to determine geoid-chart separation Afterward, known separation can be used to derive a calibrated SEP surface Facilitates the inclusion of AUV data in products comprised of surface survey data

8. Method Install CNav 3050 GNSS receiver on R/V Meriel B. Antenna height above MRP measured with tape measure (-3.78 m) Drop from roll, pitch, settlement neglected PE-coated pressure sensitive adhesive layer Ratcheted, nylon-based retention system Collect navigation data while Meriel B. remains close to AUV location during AUV MB missions Collect navigation data at 1 Hz Post-process navigation data using PPP, 30-second clocks and ephimerides

9. Method, Continued Take unweighted moving average over 60-s period of the ellipsoid height data Period chosen from trial and error on earlier work Long enough to ignore small effects but short enough to capture long term water level changes

10. Method, Continued Apply ellipsoidally referenced correction to Generic Sensor Format (GSF) file if data are sufficiently close in time Interpolate linearly between discrete height measurements In general, if the GSF ping is less than a few seconds away from height measurement, it should apply Threshold we used here was |Δt| <= 0.99 s Tide corrector (TC) = SEP – EH Add difference to Tide Corrector field in GSF

11. Evaluation Requires test data to be brought to same vertical datum as the observed tides Accomplished using a calibrated SEP surface to determine ellipsoid/datum separation for area Found to be -29.24 m from nearby benchmarks Gridded data (1-m, average surface) will be differenced and assessed qualitatively and quantitatively Virtual tide data will be compared to NOAA Fort Point gauge data

12. Results from Grid Comparisons Day 1: Mean difference = 0.04 m (vtg results in shoaler depth than conventional gauge) Median of differences = 0.04 m Standard deviation = 0.05 m Day 2: Mean difference = 0.04 m Median difference = 0.04 m Standard Deviation = 0.04 m Far fewer samples

13. Results Fledermaus display of both surfaces and difference

14. Water Level Results Day 1: (N = 64) Mean: 0.04 m STD: 0.05 m Day 2: (N = 28) Mean: 0.05 m STD: 0.05 m Day 3: (N = 65) Mean: 0.00 m STD: 0.07 m

18. Conclusions With even basic GNSS capability on a nearby vessel, applying a virtual tide corrector is feasible Measurements from better-known survey vessel should further reduce uncertainty, but results are promising even with a basic setup Our distances between the navigation vehicle and the AUV were small (~1 k?) Further tests will be required to determine the horizontal limits of this method

19. Further Testing Repeat tests with survey launch, GPS Buoy, or autonomous surface vehicle Compute time series of horizontal distance between AUV and navigation vessel Compare horizontal distance to conventional/virtual disparity Develop routine to query calibrated SEP surface to determine separation at AUV location vice single SEP value Useful as survey area and horizontal distance become larger

20. Questions?