Due to the constant improvement in hydrographic survey technology, the best information on an area to be surveyed is the most recent information available for that area. If a survey line were just run in the area, then this would be the best source of information available. This is the premise upon which AutoSurvey was built, resulting in the advantage of near 100 per cent coverage in reduced time as compared with conventional approaches.

The AutoSurvey Advantage
The improved performance achieved with AutoSurvey results from the acquisition of real-time swath sensor data and the generation of adaptive line spacing from this data. The common ‘ladder’ approach to surveying, shown in Figure 1, uses uniformly spaced parallel lines, the spacing of which is based on swath width estimated from the minimum water depth of a survey area. Because this information may come from outdated charts or other questionable sources, and due to variations in sonar performance, the desired results of the survey may not be easily achieved. Data gaps may occur in areas where actual swath widths become too narrow for the given line spacing, whereas areas of overage (excessive coverage) may occur where the swath width becomes too wide. These variations need to be compensated for by planning new lines to fill in the ‘holidays’ and by trying to determine an appropriate line skipping plan to reduce the areas of overage.

Generating Lines with AutoSurvey
Because the undersea environment significantly affects multi-beam performance, using real-time swath sensor data from the previous line is a sensible basis for next-line generation. Given this data, optimal survey performance is achieved when each track line’s shape matches that of the previous lines’ swath edge. AutoSurvey provides three methods for calculating the next line to achieve this performance.
The Adaptive Parallel (AP) method most closely resembles a ladder survey. The AP method uses parallel lines but the constraint of the lines being uniformly spaced is removed. The method works by estimating the minimum swath width from the returned swath sensor data then bases its line spacing for the new track using this width. The result is a series of irregularly spaced parallel lines that ensure near 100 per cent coverage; however, as with a ladder survey, there is usually coinciding overage. Figure 2 shows an example of an AP survey.
The Linear Regression (LR) method uses a best linear fit approach to next-line generation. In addition to removing the constraint of having uniformly spaced lines, this algorithm also allows lines that are not parallel. The next-line generated by this method is instead parallel to a linear fit of the previous lines’ swath edge data. With this approach, the amount of coverage and the overall survey time are reduced as compared to the AP method.
The final line generation method used by AutoSurvey is the Piecewise Linear (PL) method. Instead of a linear fit to the previous lines’ swath edge, PL generates the next line based upon a piecewise linear fit, resulting in a segmented track line whose shape closely matches that of the previous lines’ swath edge. Using PL, data gapping and overages are almost entirely eliminated and overall survey time is reduced.

Line Method Comparisons
Examining first the simplest bottom geometry, that of a constant slope, it is straightforward to compute the time saved with AutoSurvey. For a constant slope, all three line methods yield the same results; with as little as a 1° bottom slope AP is 10 per cent more efficient than the ladder method for a survey extending 2,500m down slope. Much greater time saving can be realised over rugged terrain using the LR and PL methods. Simulations have been conducted using data from the East Flower Gardens area in the Gulf of Mexico that has multiple seamounts and depth variations from 18m to 140m. This area is shown in Figure 3 and the white box indicates the area where simulations were run. In these simulations it was shown that LR and PL provided time savings greater than 55 per cent as compared to a ladder survey.
At-sea comparison tests were conducted in August 2002 near the Dry Tortugas, 110 miles west of Key West. The survey area was 4,000m by 6,000m and varied in depth from 60m to 220m with a 70m drop in the middle. Four separate surveys were conducted using the ladder method and all three AutoSurvey methods. The ladder method yielded 95 per cent coverage of the area due to an over estimate of the swath width at the shallow end, and all three AutoSurvey methods yielded 97 per cent coverage or better. For this area, the LR method proved the most efficient and required 36 per cent less time than the ladder survey.

The AutoSurvey Planner
The AutoSurvey Planner is a survey-planning tool that uses the adaptive line generation methods of AutoSurvey. For a specified survey area the planner generates both graphical and numeric results, including the time required to run the survey, the distance covered, the waypoints generated, the overall percent coverage achieved, the location of data gaps and overages and the data density given a user-specified grid size. Figure 4 is an example of the percent coverage graphic where red indicates the coverage of a single swath, white indicates
data gaps and green indicates areas where swaths overlap. An example of the data density graphic is shown in Figure 5 and it displays the number of beams that fall in each 20m bin of the survey area.
Given a digital terrain map generated from historical bathymetry for the specified survey area, the Planner simulates a complete survey using either the ladder method or one of the AutoSurvey methods, depending on user selected options. The Planner allows specification of both sensor and vessel characteristics, including number of beams, angular coverage, maximum slant range, vessel speed and turning rate.

Conclusion
AutoSurvey was transitioned by NRL to the US Naval Oceanographic Office (NAVOCEANO) in 2002. It was installed into the Integrated Survey System (ISS-60) developed by SAIC and has been distributed to the NAVOCEANO TAGS-60 class survey ships and their associated hydrographic survey launches. A beta version of the AutoSurvey planner has been released and is presently undergoing validation and testing.