The Hugin autonomous underwater vehicle (AUV) was built at Kongsberg Simrad's Norway facilities in late 1998 and delivered to the US in August 1999. C & C Technologies conducted sea trials in the Gulf of Mexico over the next several months and the Hugin was officially commissioned as a commercial survey vehicle in January 2000. Since then, the AUV has completed more than 6,000 line miles of deepwater data acquisition on the continental slope and upper continental rise in the Gulf of Mexico.

Some of these projects have been rather challenging, particularly along the Sigsbee Escarpment in the north-central Gulf of Mexico. The topography of the Sigsbee is steep and irregular, with numerous faults and slumps occurring on the escarpment face, a challenge for the Hugin's collection of high-resolution geophysical data.

The data has provided a dramatic advance in the ability to map the seafloor. In fact, the AUV has proven so successful that the large and diverse data set it produces has challenged C & C's ability to process and manage it.

The Hugin 3000 carries three major remote-sensing systems. A Kongsberg Simrad EM2000 Multibeam collects high-density soundings in a swath perpendicular to the direction of motion. The AUV depth is recorded with a precision depth sensor. Sonar imagery is logged with an Edgetech dual-frequency sidescan sonar, and high-resolution seismic profiles are obtained with an Edgetech 216 chirped sub-bottom sonar.

Inertial navigation is used for positioning the AUV and a battery supplies power for mission times approaching 45 continuous hours. Numerous ancillary sensors monitor the vehicle and feed information to the artificial intelligence programs that control it.

Navigation Data Processing
The AUV's positioning implements the same type of intertial guidance technology developed for positioning precision guided missiles. This technology is relatively new to the survey industry and the results are much better than initially anticipated. The error in georeferencing of the post-processed data with system is ±5 meters.

In the early stages of development, some at C & C suggested developing swath-editing tools allowing the data processor to average or adjust the navigation data based on the alignment of seafloor features, which overlapped between adjacent swaths. This concern was based on the fact that the multibeam data sets would be extremely high in relative vertical resolution and any mismatches due to positioning would be very apparent.

Fortunately, the quality of the post-processed inertial navigation data is extremely accurate and little or no editing is required. This accuracy in positioning makes the acquired deepwater geophysical data very easy to process and interpret.

Being able to log AUV positions kilometers below the sea surface at resolutions approaching surface GPS accuracy is truly remarkable. Any geoscientist who has been assigned the task of mapping the seabed with conventional deep-tow data can appreciate this technological advancement.

The AUV navigation data are processed and then merged into the output formats of the acquired geophysical data. This requires all sensor data be time-synchronized for correct georeferencing. The standard deviation of the real-time navigation is on the order of ±15 meters. The Inertial Navigation System (INS) uses accelerometers as the primary positioning system on the AUV. Position fixes from a Kongsberg Simrad short-baseline acoustic system deployed on the mothership minimizes positioning drift. A fiber optic gyro also monitors heading and an acoustic Doppler profiler provides speed-over-ground input. The post-processed positioning solution is obtained utilizing a Kalman filter and differentially weighted inputs from applicable sensors.

Multibeam Data Processing
Hugin's multibeam bathymetry data allows pipeline and facility engineers to view large areas of the deepwater seafloor at a level of detail never before possible. The EM2000 multibeam system operates at 200 kHz and collects data in a swath width of about 220m. There are 111 beams, or individual soundings, collected on each ping.

Salinity and temperature measurements are sampled continuously at the transducer face for correct beam forming. An Ixsea-Oceano Octans compensator utilizes precision accelerometers to record the heave, pitch, and roll values, which are applied to the soundings and added to the values from a deepwater survey precision pressure sensor.

Multibeam data processing is performed utilizing proprietary software. The multibeam soundings are processed using binning algorithms. A three-meter bin size is the standard used for the AUV soundings.

Typically, six or seven raw soundings are recorded in the three-meter bin. The processing begins by conducting statistical analysis of the raw soundings within the bins. Any bins with high standard deviations are examined for noisy soundings, or outliers, and these points are eliminated from further processing. The soundings are then reduced using a median filter. The median sounding within each three-meter bin is used to produce a gridded dataset containing points that are equally spaced. The water depth value for the grid bin is calculated using a near-neighbor statistically weighted subroutine.

The gridded dataset is then used for geotiff and contour generation. Triangulation of the dataset can be performed if needed. Slope-gradient maps or images are easily generated and very subtle seafloor features are accentuated on these displays. Fledermaus 3D visualization software is used to view the multibeam data.

Sidescan Data Processing
Sidescan sonar data in the 120kHz frequency band are collected in raw Edgetech format and then converted to XTF file format for viewing and interpretation at a computer workstation. The sonar data are sampled more than 2,000 times over the duration of the receive time. In normal operation, the sidescan sonar transmits about three times a second, resulting in recording a per-channel range of 238 meters. The sonar range can be set to different intervals while in stand-alone mode. The sidescan sonar can be operated in a high-resolution 420 kHz mode. The same navigation dataset used for the multibeam is used to process the sidescan data.

Triton-Elics software is used for the playback, interpretation, and hardcopy generation of the sidescan sonar data. The software allows the interpreter to output ASCII or DXF files for import into CAD mapping software. Mosaics can also be constructed utilizing the program and output of the sonar data in TIF format is available.

Traditionally, one of the biggest problems with producing sidescan sonar mosaics is the editing of the navigation data. The AUV navigation data is processed and edited on the front end, prior to being merged with the sonar data. This results in the production of mosaics quickly, with little or no time spent on the editing of the navigation data.

Ocean Imaging Consultant's (OIC) software is generally used to produce mosaics with proper georeferencing and filtering. The mosaics can then be used to drape over the 3D model of the multibeam dataset for analysis in the Fledermaus software.

Sub-bottom Data Processing
The Edgetech "chirped" sub-bottom data are collected utilizing a frequency modulated seismic source in the 2kHz to 8kHz frequency band. The record time is limited to about 300 milliseconds, with time zero occurring at the altitude of the AUV, resulting in the recorded raw data being non-topographically corrected. The Hugin's depth information must be input into the final output files in order to produce a topographically corrected record. The survey precision depth sensor information must be incorporated on a ping-by-ping basis, as the AUV is constantly changing depth to follow the seabed terrain.

XTF and SEG-Y file formats are two outputs available in the processing of the seismic data. Triton-Elics software is used to read the XTF files, which allow for proper referencing of the seismic data in both the vertical and horizontal plane. These files are also constructed after the navigation data are processed. X and Y data, AUV depth, and event marks are incorporated into the final output by using time tags for each seismic shot.

The SEG-Y file format requires an integer millisecond value for the static offset of each ping. This integer value requirement is not resolute enough to properly reference the seismic data in the vertical plane and results in a blocky presentation of the seismic data in SEG-Y trace viewers. Seismic Micro Technologies worked with C & C on utilizing one of the unused records of the trace header for storing a number value resolute enough for static offset of each seismic trace.

Vehicle Maneuverability
The ability of the Hugin to maneuver without a tether results in a drastic reduction of survey time and significantly increases the quality of the remote sensing data. Conventional deep-tow survey systems are difficult to tow along a preplotted course due to the distance between the towfish and the survey ship, often more than 4km in water depths greater than 4,920 feet (1,500m). Course deviations made by the survey vessel to alter the towfish course are not immediate and crosscurrents may result in gaps in the survey coverage. The AUV continuously receives feedback from navigation sensors and can adjust the stern and rudder planes to quickly adjust attitude and course.

The ability of the Hugin to navigate curved lines drastically reduces survey time. The minimum turning radius is 15m. When line turns are necessitated, they are typically made in about 5 minutes, where deep-tow system line turns can take hours.

The lack of a tow cable attached to the AUV results in data collection that is virtually void of interference by weather. The wave action affecting the surface towed vessel in a deep-tow configuration is transmitted through the tow cable and reduces the data quality.

The Hugin is free from the effects of winching that is used to control the altitude of the towfish in deep-tow operations. Deep-tow system operators are always concerned about topography and usually fly the towfish higher than normal across significant topography due to concerns for the safety of the equipment. The response of the towfish through cable winching is not immediate and can take several minutes before a significant change in towfish altitude is observed. With the Hugin, instantaneous stern plane adjustments are made on feedback received by the acoustic Doppler profiler. The AUV maintains a relatively constant altitude of 40m, which results in high quality sidescan data collected across significant seafloor slopes.

The AUV maintains a constant speed approaching four knots, allowing a very consistent dispersal of the remote sensing data within the swath of survey coverage. In contrast, towfish are subject to increases and decreases in speed whenever cable is spooled in or out, which results in irregular data densities in the alongtrack direction.

Real-Time Data Transfer
One of the technical challenges of using AUV technology is how to transfer the data to the survey ship in order to make routing decisions regarding the remote sensing data stored on the vehicle without retrieving the system. An acoustic modem is used to accomplish this task using two discrete transmit frequencies. The high frequency band transmits the collected data from the AUV to the mothership for periods of 30 seconds. The low frequency band is then used for the next 10 seconds to transmit information from the mothership to the AUV. These transmissions include information for control and operation of the vehicle.

New mission plans or waypoints for "on-the-fly" course changes can also be sent, but the bandwidth allowed for these alter survey course points is less than 40 characters per 10-second transmission. This bottleneck has created the need for a unique binary survey command set to control the AUV.

The acoustically transmitted, decimated datasets of sidescan, sub-bottom, and multibeam data allow the shipboard geoscientists and engineers to make decisions regarding routing alignments. The transmitted soundings are dense enough to produce five-meter binned datasets.

3D Visualization
The development of the deepwater AUV with its high-resolution multi-sensor package, in concert with accurate navigation, has fundamentally changed the ability to map the seafloor. The high-resolution coverage in deepwater of relatively large areas of the seabed provides a new perspective that has the potential to revolutionize our understanding of seafloor processes. It also demands improved methods of presentation for interpretation and analysis.

Such a revolution does not come without a price, however, and in this case the price is one of data density. The massive amounts of digital data collected by the sensors present tremendous challenges, first in the individual sensor acquisition and processing, and then in terms of interaction, integration, and interpretation. If properly handled, however, the inherent density of the data available from these systems also presents tremendous opportunities.

The human visual system has an enormous capacity for receiving and interpreting data quickly and efficiently and therefore must be an integral part of any effort to understand complex data. The key is to be able to present the data in as intuitive a fashion as possible, and the more intuitive the presentation, the more rapidly data is interpreted, and the more new information can be extracted from that data.

These elements are incorporated in the Fledermaus interactive 3D software application, and allow the integration and analysis of the multi-sensor data sets from the AUV. Importantly, the accurate navigation of the AUV permits these complex data sets to be properly georeferenced in the 3D scene and presented in an intuitive manner that allows the simple integration of multiple components without compromiseing quantitative aspects of the data.

The software uses the C & C gridded data set in generating the seabed model that has a color map assigned. A lighting model is chosen, including artificial sun-illumination, shading, and true shadow, and the scene is then rendered to form a 3D image that is a natural and detailed view of the seafloor morphology. These scenes are easily interpretable, yet fully georeferenced and quantitative. All points are georeferenced and can be interrogated in the 3D scene for position, depth, and any other attribute. Measurements can be made and data sets profiled for interactive analyses.

Color, while used to represent depth in the images, can also be mapped to other parameters such as the sidescan sonar mosaic, and draped over the digital terrain model. The software also allows sub-bottom data to be imported as a SEG-Y or image file which can be co-located in the 3D scene as a vertical curtain that follows the track of the AUV. Each of these data is loaded at the best resolution. There is no need to resample any of the individual data sets or compromise their quantitative value.

Another significant advantage of the AUV sidescan and sub-bottom data that is not available from normal towed operations is that the position and orientation of the sonar and profiler is known as accurately as the multibeam sonar. This provides a superior result and allows, for the first time, these types of data to be successfully integrated in the 3D scene for an intuitive and "real" image of the seafloor processes. The user can interactively "fly" around the data, viewing it from all angles and with special LCD glasses - in true stereo.

3D visualization provides the ideal complement to the AUV and is a significant element of meeting the challenge of ever increasing digital data volumes. Visualization provides the complete picture of all the data gathered during the survey or available from other sources, and allows the interpreter to gain maximum value from seeing the complete picture.

A Revolution in Oilfield Surveying
The AUV surveys conducted across the Sigsbee Escarpment in the summer and fall of 2001 produced the most extensive and detailed high-resolution survey data obtained to date. The area of investigation encompassed approximately 147 square miles in the southeastern portion of the Green Canyon Area, about 120 miles south of Port Fourchon, Louisiana. Primary lines were run with 200m primary line spacing and the tie lines were spaced at 500m.

The Sigsbee Escarpment represents the seaward limit of the salt province of the Gulf of Mexico. The intrusion of a salt tongue has resulted in numerous seafloor faults along the intrusion area. The escarpment face is characterized by numerous gullies or slumps that have resulted from past sediment instability. The slump deposits at the base of the escarpment form aprons of sediment consisting of displaced and mixed sediments of primarily clay.

Mega-furrows were identified on deep-tow and 3D seismic surveys along the base of the Sigsbee Escarpment. The features identified along this portion of the escarpment are generally three to 10 feet (1m to 3m) in depth and range from 16 to 160 feet (5m to 50m) in width. The features extend for miles in some locations and were probably formed by helical flow of bottom currents. The features represent an engineering challenge for the flowline and pipeline alignments. The opportunity offered by full digital integration of these data for improved interpretation has long been a dream of industry leaders such as BP.

AUV technology has progressed from a research interest to a commercially viable alternative for the collection of remote sensing data in deepwater environments. Inertial navigation has been proven as a successful means of positioning an AUV in deepwater to accuracies that have never before been achievable. The ability of AUVs such as the Hugin to navigate curves and the lack of a tether results in a significant time savings.