Best Practices Guide for AUV Polar Operations

High Lattitude AUV Navigation

In essence, the navigation systems on an AUV to be used in high latitude are the same as those that would be used elsewhere. However, there are several points to bear in mind concerning AUV navigation in the Polar Regions. These are considered below under the headings of typical navigation sub-systems.

GPS
As GPS navigation satellites are polar-orbiting, there is no issue with coverage or performance at high latitudes.

DVL
Three options exist for operating DVLs in high latitudes:
  1. conventional bottom,
  2. water tracking, and
  3. tracking the underside of ice.
These will be considered separately.

Bottom tracking is the preferred mode of operation, as it offers a stable reference. However, the operating area may not always be sufficiently shallow for a DVL to have the necessary bottom tracking range. This can be especially true in the Antarctic, where, due to the ice loading on the continent, the continental shelf depths are deep - at typically 400-500m, compared with the more usual 200m or less. This implies that a low frequency DVL (e.g. 150kHz) is needed for bottom tracking on Antarctic shelves.

Water tracking is the only viable option in deep water. Its limitations as to error growth from unknown absolute currents are well known. Less obvious is the need to assess at what distance from the vehicle the velocity measurement is made. Polar waters, especially those under ice, can be very low in the animals contributing to sound scattering. Range cells closer to the vehicle may need to be used.

Under ice tracking offers a third option in Polar Regions. It has been used very successfully from the underside of an ice shelf (Nicholls et al., 2006), where the ice shelf motion of ~1000m a year (~0.003cm.s-1) was essentially zero. Tracking the underside of moving sea ice may or may not be successful. It depends on mission requirements. Where the need is to track beneath ice adjacent to a lead or polynya on short missions, and the relative motion between ice and lead or polynya is small, acceptable results may be achieved (McEwan et al., 2005). In such cases, the important factor is not necessarily the absolute position accuracy, but the position accuracy relative to the moving ice field, and being able to return to the lead or polynya for recovery. In other cases, such as with Autosub in the Amundsen Sea, tracking the underside of moving sea ice in deep water led to large position errors and prolonged search and recovery operations.

Heading
The two main types of heading sensor likely to be used in an AUV:
  1. magnetic and
  2. gyrocompass,
both have issues to be considered when operating in high latitudes.

Magnetic heading sensors, such as fluxgate magnetometers, will need careful calibration for the magnetic variation and local conditions at the AUV deployment site. Ideally, a heading-dependent correction table will be derived either via a calibration routine within the sensor, or by comparison with an external heading source with known characteristics. This may take the form of a GPS-based carrier phase sensor such as the Magellan ADU5. In addition to variation, the low value of the horizontal component of the earth's magnetic field near the poles may cause the magnetic heading sensor to exhibit higher heading noise levels. For long transects, the launch-point calibration would be updated en route using an on-board magnetic field model. Care is also needed to ensure that the heading output is independent of vehicle pitch. McEwen et al. (2005) reported pitch-dependent heading variations of up to 3° at 82°N off Svalbard, where the magnetic inclination was 83.5°.

The North Magnetic Pole was at about 82.7°N 114.4°W in 2005 and moving ~40km NNW each year. Daily, the position of the magnetic pole varies in an irregular elliptical path around its mean position, and at noon and midnight may be over 40km from the mean position. The South Magnetic Pole was at about 64.59°S 138.53°E in 1998 and moving ~5km a year to the NNW.

Gyrocompass heading sensors depend on sensing the rotation of the Earth about its axis, and, as a consequence, will cease to function at the north and south geographic poles. The dynamic error e is dependent on latitude, and is of the form: e = k S cos C sec L (Bowditch, 1977), where k is an instrument-dependent constant, S is the speed, C the course and L the latitude. Note that this error derives from the basic physics of the measurement, and hence is applicable to fibre-optic gyrocompasses as well as mechanical versions. While AUV operating speeds are generally low, and usually constant, heading will undoubtedly vary, and so the error will vary from zero on east/west courses (where cos C is zero) to a maximum on north/south courses. The dependence on secant of latitude means that the error at 80° is 3.7 times that at 50°. Hence, for high latitude operation, it is advisable to choose a gyrocompass with a low k. The Geonav SP2000/LFK95 fibre-optic gyrocompass has a k of 0.7°, while the Ixsea Octans unit has a k of 0.2°. Higher performance can be expected from ring-laser based inertial navigation systems (INS). McEwen et al. (2005) claimed that a Kearfott and RDI Seadevil integrated INS/DVL/GPS gave 1 milliradian (0.057°) heading accuracy at 82°N.

Homing
Where there is a need to traverse under 100s of km of ice, without the possibility of external position updates, from GPS or from acoustic navigation systems, some form of terminal homing system is advisable for successful recovery where the recovery position is a lead, polynya or seaward edge of an ice field. The homing system not only provides a means of coping with the position error growth by the time of recovery, it allows for the mission end point to be updated, or determined, from a support ship or other platform such as a helicopter. Hence, the ship or other platform finds open water near the expected final waypoint. The homing concept is also valid for ice-camp based AUV missions, where recovery is to a hole in the ice.

As implemented for Autosub, the acoustic homing system operates at 4.504kHz with a range of up to 15 km. A transmitter on the support ship emits regularly spaced chirps from a transducer array lowered on a cable down to a depth of up to 100m. Three spherical hydrophones on the vehicle provide signals to a DSP-based three-channel correlation receiver that determines the direction of arrival of the homing signal. To avoid false alarms, four consecutive transmissions with the correct spacing and chirp characteristics are needed before the AUV enters homing mode and heads towards the source of the signal.

Acoustic navigation
Under-ice navigation without periodic position updates adds to the risk of AUV missions. If a network of acoustic navigation and communication nodes were to be available, AUV operations under ice would be less risky and more scientifically productive. The challenges in developing such a network were reviewed recently at the ANCHOR workshop in Seattle (Lee and Gobat, 2006). The first requirement is seen as a consensus on the approach to design and implementation. Several ideas were considered, including: Similar concepts could be used in the Antarctic, beneath ice shelves, where acoustic sources could be placed through holes drilled in the ice. The high cost of such an infrastructure would only be justified if there were to be on-going monitoring programmes that used AUVs under ice shelves.

REFERENCES
Bowditch N. (1977) American Practical Navigator, Volume I. Defense Mapping Agency Hydrographic Center, 1386pp.
Lee, C.M. and Gobat, J.I., 2006. Acoustic navigation and communication for high-latitude ocean research workshop. EOS, 87(27): 268-269.
McEwen, R., 2005. Performance of an AUV navigation system at Arctic latitudes. Journal of Oceanic Engineering, 30(2): 443-454.
Nicholls, K.W., et al., 2006. Measurements beneath an Antarctic ice shelf using an autonomous underwater vehicle. Geophysical Research Letters, 33, L08612, doi:10.1029/2006GL025998.
© 2007