nearly 100% by mid-1991. Thus availability grew
rapidly during the early stages of the WHP, but
the earliest cruises still had gaps in coverage.
The importance of improved position measurement
Accuracy during the WHP has varied depending
on the status of ‘Selective Availability’ (SA), on the
type of receiver, and on the availability of a differential correction, either in real time or in delayed
reprocessing. SA is the deliberate reduction of
accuracy for civilian users by the US Department
of Defense. It was first activated in July 1991, and
was continuous after November 1991. The error
level was set to zero on 2 May 2000. Under SA,
position uncertainty (approximately 2 standard
deviations) is of the order of 100 m, equivalent to
velocity uncertainty of 47 cm s
91 for a 5-minute
average, or 4 cm s
91 for a one-hour average. This
was the best that was available with a standard
single-frequency civilian receiver using the Coarse
Acquisition or C/A code. This level of accuracy
is also known as the Standard Positioning Service,
or SPS.
By the time the WHP was concentrated in the
Indian Ocean in 1995, most US ships had military
GPS receivers. These could receive signals on both
the L1 and L2 frequencies, and could decode the
encrypted P/Y code, with about 10 m position
accuracy, and a corresponding 10-fold improvement in velocity accuracy.
An alternative method of improving accuracy
with just a C/A code receiver is Differential GPS.
In this technique, signals are received at a fixed
antenna of known location, as well as the mobile
antenna. At the fixed location, an estimate can
then be made of the error introduced by SA for
each satellite in common view of both antennas,
and a correction made at the mobile antenna. The
technique assumes a similar satellite geometry at
both locations. Since the GPS satellites orbit at an
altitude of approximately 20 000 km, DGPS provides useful corrections over baselines of several
thousand kilometres. Differential methods, when
available either in real time or in a delayed processing mode, provide P/Y code accuracy, or better, with C/A receivers. Uncertainty of as little as
4 m (one standard deviation) on a 2000 km baseline has been observed (King et al., 1996; Pierce
et al., 1999). As early as 1993 there were commercial systems for receiving DGPS corrections in real
time, but they were charged at a very expensive
daily rate that put them out of reach of research
budgets. Delayed-mode was the only feasible
option for research ships. In 1996 a commercial
receiver became available (the Ashtech GG24),
which was capable of receiving signals from both
GPS and the Russian GLONASS system; the latter
was by then not subject to SA or encryption. However, this potentially useful advance was overtaken
by the routine commercial availability of affordable real-time DGPS corrections via telecommunications satellites. Many non-US research
ships were equipped with continuous real-time
DGPS, providing effective high-quality positions
over most of the global ocean. On 2 May 2000,
the level of SA error on C/A code was set to zero,
immediately providing accuracy comparable to
P/Y code; the US government has declared its
intention not to reintroduce it. For precise applications, however, DGPS is still generally superior to
uncorrected GPS.
Even when degraded by SA, position fix accuracy usually is not the limiting factor for SADCP
applications; the error in section-averaged velocity
caused by fix error varies inversely with the length
of the section, and is usually below the ocean
signal (and internal wave ‘noise’) level for scales
larger than about 10 km. Fix accuracy is most
important for smaller horizontal scales and in
regions of weak currents. It is also important for
evaluating and calibrating the sonar system; we
will return to this point.
GPS heading measurements
In 1990, a far more serious problem than position
uncertainty was heading uncertainty. The error
characteristics of gyrocompasses were poorly
known and unmeasureable until the advent of GPS
attitude measurement methods early in the decade.
The sensitivity of shipboard ADCP measurements
to heading error, however, was always clear: a 1°
error on a ship underway at 5 m s
91 causes a
9 cm s
91 cross-track velocity error. Heading error
can be viewed as the sum of a constant error in
angular offset between the transducer axis and the
axis of the gyrocompass or other heading source,
and a zero-mean but time-variable error in the
heading measurement itself. The angular offset is
usually estimated in either or both of two ways:
(1) in shallow water, by comparing the velocity of
the ship over the ground as measured by GPS with
the velocity measured by the ADCP tracking the
3.1 Shipboard Observations during WOCE
113
King, Firing and Joyce
rapidly during the early stages of the WHP, but
the earliest cruises still had gaps in coverage.
The importance of improved position measurement
Accuracy during the WHP has varied depending
on the status of ‘Selective Availability’ (SA), on the
type of receiver, and on the availability of a differential correction, either in real time or in delayed
reprocessing. SA is the deliberate reduction of
accuracy for civilian users by the US Department
of Defense. It was first activated in July 1991, and
was continuous after November 1991. The error
level was set to zero on 2 May 2000. Under SA,
position uncertainty (approximately 2 standard
deviations) is of the order of 100 m, equivalent to
velocity uncertainty of 47 cm s
91 for a 5-minute
average, or 4 cm s
91 for a one-hour average. This
was the best that was available with a standard
single-frequency civilian receiver using the Coarse
Acquisition or C/A code. This level of accuracy
is also known as the Standard Positioning Service,
or SPS.
By the time the WHP was concentrated in the
Indian Ocean in 1995, most US ships had military
GPS receivers. These could receive signals on both
the L1 and L2 frequencies, and could decode the
encrypted P/Y code, with about 10 m position
accuracy, and a corresponding 10-fold improvement in velocity accuracy.
An alternative method of improving accuracy
with just a C/A code receiver is Differential GPS.
In this technique, signals are received at a fixed
antenna of known location, as well as the mobile
antenna. At the fixed location, an estimate can
then be made of the error introduced by SA for
each satellite in common view of both antennas,
and a correction made at the mobile antenna. The
technique assumes a similar satellite geometry at
both locations. Since the GPS satellites orbit at an
altitude of approximately 20 000 km, DGPS provides useful corrections over baselines of several
thousand kilometres. Differential methods, when
available either in real time or in a delayed processing mode, provide P/Y code accuracy, or better, with C/A receivers. Uncertainty of as little as
4 m (one standard deviation) on a 2000 km baseline has been observed (King et al., 1996; Pierce
et al., 1999). As early as 1993 there were commercial systems for receiving DGPS corrections in real
time, but they were charged at a very expensive
daily rate that put them out of reach of research
budgets. Delayed-mode was the only feasible
option for research ships. In 1996 a commercial
receiver became available (the Ashtech GG24),
which was capable of receiving signals from both
GPS and the Russian GLONASS system; the latter
was by then not subject to SA or encryption. However, this potentially useful advance was overtaken
by the routine commercial availability of affordable real-time DGPS corrections via telecommunications satellites. Many non-US research
ships were equipped with continuous real-time
DGPS, providing effective high-quality positions
over most of the global ocean. On 2 May 2000,
the level of SA error on C/A code was set to zero,
immediately providing accuracy comparable to
P/Y code; the US government has declared its
intention not to reintroduce it. For precise applications, however, DGPS is still generally superior to
uncorrected GPS.
Even when degraded by SA, position fix accuracy usually is not the limiting factor for SADCP
applications; the error in section-averaged velocity
caused by fix error varies inversely with the length
of the section, and is usually below the ocean
signal (and internal wave ‘noise’) level for scales
larger than about 10 km. Fix accuracy is most
important for smaller horizontal scales and in
regions of weak currents. It is also important for
evaluating and calibrating the sonar system; we
will return to this point.
GPS heading measurements
In 1990, a far more serious problem than position
uncertainty was heading uncertainty. The error
characteristics of gyrocompasses were poorly
known and unmeasureable until the advent of GPS
attitude measurement methods early in the decade.
The sensitivity of shipboard ADCP measurements
to heading error, however, was always clear: a 1°
error on a ship underway at 5 m s
91 causes a
9 cm s
91 cross-track velocity error. Heading error
can be viewed as the sum of a constant error in
angular offset between the transducer axis and the
axis of the gyrocompass or other heading source,
and a zero-mean but time-variable error in the
heading measurement itself. The angular offset is
usually estimated in either or both of two ways:
(1) in shallow water, by comparing the velocity of
the ship over the ground as measured by GPS with
the velocity measured by the ADCP tracking the
3.1 Shipboard Observations during WOCE
113
King, Firing and Joyce