The Use of a GPS-Equipped Buoy for Water Level Determination

Sunil Bisnath’, David Wells’, Stephan Howden’, and Gregory Stone2

’ Hydrographic Science Research Center, Department Of Marine Science, The University Of Southern Mississippi
1020 Balch Blvd.. Stennis Space Center, MS. 39529
[email protected]

Coastal Studies Institute, Louisiana State University

320 Howe Russell Geoscience Complex, Baton Rouge, LA 70803

Abstract - In recent years, Global Positioning System (GPS)

receivers have been placed on buoys to determine sea.surface
height, The Naval Oceanographic Office (NAVOCEANO) has
become interested in this application for tides work and has
sponsored a series of GPS buoy experiments in Mississippi
coastal waters. The primary goal of the experiments was the
determination of water level with GPS using the WGS84
reference ellipsoid as a tidal benchmark. Also of importance
was the determination of the minimum buoy sensor
.configuration (aside from GPS) required to accomplish this
measurement, and to evaluate RTK GPS performance over time
and distance.
The fully integrated buoy was deployed in 2001, 2002, and
2003. The 2001 RTK results were analyzed for solution
availability, solution quality, and solution correlation with other
sensor output. Solution availability was high during buoy
operational periods; however, the solution quality was hampered
by an apparent GPS filter-based anomaly in the receiver RTK
processing. Tide gauge comparison indicated subdecimeter-
level water level recovery was altainable. Initial processing of
2002 data produced cm-level differences between RTK and tide
gauge water levels, These initial results show great promise for . .
the use of RTK buoys in water level recovery, and more data
collection and analysis will be undertaken with the 2003
deployment.

I. INTRODUCTION Local tidal datum

A requirement for hydrographic surveying for marine chart
production has always been the determination of a physically
meaningful vertical surface with which to reference measured
depths. This chart datum is realized in the form of some low
water level in the survey area. Given tidal, meteorological, Tidal datum /
etc. effects the traditional approach has been to use local
water level gauge data or install gauges (if none exist) for the
. period of the survey. Since the gauges represent basically II
discrete measures of water level and the survey area can 1 Reference ellipsoid
cover considerable areas, it may be required to estimate
vertical datum information by partitioning the coverage area Fig. 1: Measurementsfrom a GPS-enabled buoy.
and modeling the water level. This procedure adds to the
vertical positioning error budget, which maps into the (I. GPS BUOY RESEARCH
accuracy of the chart depths.
The development over the past decade of commercial,
The Global Positioning System (GPS) has revolutionized carrier phase, differential, kinematic GPS. specifically Reai-
positioning and navigation. By measuring the travel time of Time Kinematic (RTK) GPS. has meant that an RTK-equipped
microwave electromagnetic waves from a number of GPS buoy can potentially be used to accurately measure water
satellite, all at known locations, a trilateration technique is level height above the WGS84 ellipsoid. A number of
used to estimate user position to the few meter level. By experiments have been successfully carried out for this
augmenting this procedure with data from other GPS purpose. The first example was the calibration of a satellite
receivers or GPS correction providers, position estimates as altimeter with a GPS buoy in the North Sea in 1988, and
precise as a few millimeters can be achieved under required the invention of the On-The-Fly (OTF) RTK
constrained conditions. processing method as a byproduct [I]. Others that have
followed include [2], [3], [4], and [5]. The ultimate goals of
GPS can aid hydrography in a number of ways. One these experiments and other potential purposes can be
obvious application is in the use of survey vessel positioning
classified as:
by onboard GPS. Perhaps less obvious is to place a GPS

0-933957-30-0 1241
 . Determination of water level with an RTK GPS buoy. NAVOCEANO deployed a buoy in 2001 and redeployed
Simply put, as an academic question, can this be the same in late 2002, and again in early 2003. .Planning
accomplished and with what caveats? And how would Systems Inc. (PSI) was tasked with integrating and operating
such a buoy compare with a water level gauge in terms of the buoy, and the Hydrographic Science Research Center
accuracy, cost, etc. (HSRC) was charged with assisting in the system and survey
design, and analyzing the collected data. Both deployments
Determination of water level at locations where were designed to zddress the above NAVOCEANO
installation of a gauge would be difficult. There are objectives.
scenarios where installation of a conventional tide gauge
is not practical. This can be due to distance from shore The buoy itself is shown in Figure 2. The internal
and / or water depth. or for safety concerns in a hostile configuration for the 2001 deployment is also shown in the
environment. In these situations an accurate buoy- figure. Table 1 describes the main components installed in
mounted sensor would be of great advantage. This the system. Dual freqiiency NovAtel OEM4 GPS receivers
raises the potential for buoydetermined water levels to were used for RTK. ;and a host of other sensors were
play a role in cotidal and hydrodynamic modeling. incorporated for redundancy comparison and attitude
determination. These! are three accelerometers, three
Determination of water level to establish the separation magnetometers, a tilt sensor, and a pressure sensor. For the
between tidal datum and GPS datum. Given that GPS is 2001 experiment, the tNuoy was deployed approximately 10
presently the ubiquitous global positioning and navigation km south of Gulfport, Mississippi, in the Northern Gulf of
technology, with both terrestrial and marine mapping and Mexico, and the reference station was located in Gulfport. For
navigation being referenced to its. datum (WGS84) both the 2002 test, the reference was again set in Gulfport, and the
directly and indirectly, there is great interest. in the buoy was placed 50 m away for an initial test, and placed
mapping community to develop transformations between approximately 15 km south of Gulfport. This latter location
local datums and WGS84. By determining water levels, was also used for the 2003 tests. Figure 3 illustrates these
and therefore chart datum, with GPS. the conversion of locations. Due to several RTK .and power problems
chart datum to the ellipsoid can be mapped out. If the encountered in the first and second deployment design, a
same is done with terrestrial datums, this can be a basis number of changes were made to the buoy design and
for development of seamlessness at the chart / map - operation for the second and third tests. These will be
sea / land interface. discussed in the results :section. . '

Determination of water level for calibration of altimetric

sensors on aircraft or spacecraft. Calibration of great and
growing interest given the rapid advancement of altimetric Table 1: Buoy system component descriptions 161.
sensors designed to determine sea state, sea surface DESCRIPTION
topography, etc. In situ water level measurements from
.. . RTKequipped buoys can provide for not only calibration Reference Station ' NwAtel OEM4 LI/L2, 12 channel, RTK
of these sensors at cross-over points, but perhaps also in and Buoy GPS differentialcapable
data assimilation models. They also complement
. . spacebome and airborne altimetric .data by providing
higher sampling rates at their moored locations. This can
be crucial to verifying if high frequency variability is Buoy GPS Antenna NovAtel Model 600. dual frequency
removed from the altimetry data properly so as not to antenna
alias the measurements.
FkeWav& DGR-115R SOOMHz, spread
Ill. NAVOCEANO RTK BUOY Sll0CtNm

The Naval Oceanographic Office (NAVOCEANO), which is Accelerometers Ciassbow Technologies CXLO2LF3 3-axis,
.charged, amongst other responsibilities, with fulfilling the U.S. f 2 g, l-V/g
- . Navy's hydrographic surveying needs, has recently become
.: . . . interested in using RTK buoys in its hydrographic operations.
NAVOCEANO is interested in GPS water level determination
1-1 r,
Watson Industries. Inc.. FGM-3013-axis,
7(1,mT, 20nThnV
*
. . as a tide- gauge proxy and for determining chart datum / Vertical Reference 'Watson Industries. Inc. ADS-C232-1A, dual
. . WGS84 ellipsoid separation. . Important engineering and . a,:is integrated sensor measuring tilt. d y
operational issues are the minimum buoy sensor configuration displacement and dy angular rate
. .requirements, the temporal and spatial performance variability rpxizzzsTl -- 7-7
r s ~ s ~ M z ~ i - s xo 3 300
of RTK GPS in coastal marine environments, and the buoy L l L I
power management strategy. The minimum equipment may
also be dependent upon buoy design.

1242
 V. 2001 BUOY DEPLOYMENT RESULTS
A primary task of the briefanalysis of the 2001 data was to
observe the performance of the RTK solutions. This meant
not only the apparent precision and accuracy of the estimated
positions, but also the integrity of the system. Figure 4
illustrates the availability of RTK solutions (as a percentage)
for each survey day in September 2001. There are significant
variations from day-to-day, with availability ranging from less
than 50 % to almost 100 %. The average was 82 %. Since
the GPS receiver was running continuously, as much as 12
hours of data could be lost on a given day. The loss is due to
an error in the sensor scheduling program that caused the
receiver to power down a number of times. The daily few
percent losses were mostly due to an intentionally planned
daily system reboot.
Fig. 3: GPS buoy deployments in Mississippi Sound.

IV. DATA ANALYSIS METHODOLOGY

The buoy collected approximately one month of
continuous RTK solutions and other sensor measurements in'
September 2001. A number of initial analyses were
performed on the data, including investigating RTK data
quality; correlating RTK height solutions with vertical
accelerometer data; correlating RTK height solutions with
local tide gauge data. A similar procedure was carried out for
2002 data and planned for the 2003 data.'
As will be shown, the RTK solutions from the 2001 dataset
indicated problems with the solution. Therefore for the 2002
deployment, receiver upgrades were made and the buoy was
collocated with a tide gauge to better control the solution
analysis. During the 2002 data collection, power problems
arose due to the buoy operating in the local winter. From
these experiences, the 2003 deployment is using a reduced
power consumption strategy. It is planned for the HSRC to
process all of this new data and complete the answers to the
Fig. 4: Buoy solution availability for September 2001 [7]
original queries posed by NAVOCEANO of determining water
levels and chart datum for hydrographic surveys. and to The RTK height estimates for a single day are shown in
determine the minimum sensor suite necessary to meet water Figure 5. 23 September was chosen. as it is a representative
level recovery specifications.

1243
sample of the entire dataset, but also contains a carrier phase To investigate the accuracy of the RTK height solutions
ambiguity problem. During the period of the shift in height and understand determination of chart datum / ellipsoid
estimation there appears to be incorrectly determined separation, the RTK-determined water levels were compared
ambiguities. Analysis of the solution quality and type records against data from a ni?ar-by tide gauge. The Waveland.
indicated that the processing engine could not resolve the Mississippi tide gauge was located approximately 10 km
ambiguities at the beginning of this shift. More problematic inshore from the buoy. To reduce the effects of the filter
were the filter artifacts visible on this day and all other days. artifacts and the overall noise of the RTK data, low pass
After ruling out the approximately 50 minute period Butterworth filtering wils performed. The results for 15
phenomenon as being caused by the natural environment, it September are produci?d in Figure 7. A bias has been
was concluded that the internal RTK processing filter was removed from the tide gauge and RTK values for plotting. A
most likely at fault. To further ascertain if this was a GPS distinct phase lag can tie seen, which is consistent with the
problem, the .vertical accelerometer data was analyzed. fact that the buoy was offshore and to the east of the
Figure 6 clearly shows that the accelerometer did not undergo reference tide gauge at Waveland, Mississippi. Even so, with
the dynamics determined by the GPS receiver. Discussion a mean bias between solutions of 47.9 cm, the standard
with NovAtel engineers brought no explanations for these deviation difference is clnly 6.5 cm, with a 2.5 % GPS data
artifacts. However, installation of the latest generation loss.
NovAtel receiver firmware resolved this problem, as will be
further discussed in the 2002 data analysis. As suggested in
Figure 3, the processing filter noise is a high frequency signal
riding upon the lower frequency water level changes.

5
i I

-5t I
0 6 12 ia 24

-5
S O
4
Fig. 7: Buoy height solutions verses tide gauge for 15
0.5 September 2001. (Tide gauge: blue circle: filtered RTK:
0 6 12 18 24 green squares; differances: red triangles.) [8].
UTC Houm
1
Fig. 5: Buoy height solutions for 23 September 2001 [7].

06
I i j

~.

R_-
0 5 10 15 20 25 30

0 10 15 20 25
Day of month

Fig. 8: Summary Statistics of buoy height solutions verses

tide gauge for September 2001 [SI.
Fig. 6: Buoy height solutions verses vertical Summary statistics for the entire month are given in Figure
accelerometer for 15 September 2001 171. 8. ’ However, for the most part, a bias of approximately 45 cm
was estimated for most days in which the solution loss was

1244
 ~

low, and for these days the noise of the difference between unavoidably mapped into the filtered water level. The
sensors was near the 5 cm level (1 a), even with the development and use of more robust filters will be carried out
mentioned phase lag. in future analysis.
To give a more realistic indication of system performance The comparison between the filtered RTK water level and
the 15 September 2001 data shown in Figure 7 is revisited. that observed at a tide gauge a few meters away was
An empirically determined phase shift of +49 minutes has excellent. With a bias of 112.5 cm removed, the standard
been applied to the RTK heights. (This requires the removal deviation of the difference between solutions was 1.2 cm.
of a portion of the comparison data at either end of the time These results are presented graphically in.Figure 11. It is.
'
series.) The resulting standard deviation of the difference worth emphasizing that in this deployment, unlike the earlier
between height determination methods is reduced from 6.5 one, the conventional tide gauge was closely collocated so .
cm to 2.5 cm. A value that is not much larger than the that no phase shift is expected.
specified precision of typical gauges.

40'
0 6 12 18 24
UTC b r s

Fig. I O : Buoy height solutions and filtered estimates for

Fig. 9: Buoy height solutions versus tide gauge for.15 15 November 2002. (RTK: thin red line: filtered RTK:
September 2001 with phase shin. (Tide gauge: blue thick green line.) [E].
circle; filtered R T K green squares; differences: red
triangles.)

VI. 2002 BUOY DEPLOYMENT RESULTS .. ..... ..,. . .. ;. ?.. .~ . ...._.-

The year 2002 data processing promised to be more
revealing than that of the previous year. Initial bench testing
with new firmware showed no indication of the filtering
artifacts previously encountered, and the installation of a 100
Mb hard drive meant all raw GPS receiver data could be
stored for later post-processing. The reintegrated buoy was
first deployed in Gulfport harbor, 50 m from the reference
station, to test the system and especially GPS receiver
processing firmware. After a successful test, the buoy was
placed approximately 10 km offshore, but experienced severe i
power drain over extended periods of continuous operation.
The cause was a combination of overcast weather conditions,
not enough winter sunlight hours, and solar panel mounting .
angle resulting in limited operation. It is intended that the
0 e 12 1s 24
buoy be again deployed in February 2003, but on a reduced UTC noun
duty cycle with the sensors on for one hour and off for
approximately three hours. This will allow sustained buoy Fig. 11: Buoy height solutions verses tide gauge for 15
operation and sampling throughout the local tide cycle over a November 2002. (Tide gauge: blue circle; filtered RTK:
month. green squares: differences: red triangles.)[a].
Data from the harbor deployment was analyzed for RTK
performance. Figure 10 illustrates the water level as VII. CONCLUSIONS AND FUTURE WORK
determined from the short baseline RTK. The data was again The determination of water levels with an RTK GPS buoy
filtered with a low pass Butteworth filter, which successfully can potentially provide data for a number of applications.
dampened much of the few-centimeter variability (attributed to NAVOCEANO understands this potential and has invested
receiver noise and wave motion) apparent in the original resources into an RTK buoy program. The developed buoy
estimates. However, gross errors in the RTK solution are

1245
 has been deployed three times. Brief analysis of the first
dataset indicated that the onboard RTK processing was
performing below expectations, but daily filtered heights
compared with nearby tide gauge data produced standard
deviation differences on the order of 5 cm. Standard deviation
fell to approximately 2 cm when the phase difference in the
tide between the buoy and the conventional tide gauge was
taken into account. Initial processing from the 2002 dataset
indicates that. with upgraded GPS RTK firmware and a very
: short RTK baseline, this difference is approximately 1 cm.
The HSRC plans to greatly expand-its analysis of the 2002
' and 2003 datasets to include correcting the RTK solutions
with data from the other sensors to improve the height
solutions, while determining the minimum sensor suite
required. Also, RTK solution precision and accuracy will be
investigated. This work entails analyzing all of the data
recovered from the continuing testing. It is also planned that
the HSRC will determine the complete integration procedure
for using a GPS buoy rather than, or in conjunction with, a tide
gauge in NAVOCEANO hydrographic survey operations. This
effort points to a significant near-term role for capable GPS-
enabled buoys in hydrographic applications.

Acknowledgments
The authors would like to thank SPAWAR for initial funding
for this research and. NAVOCEANO for present funding.
NAVOCEANO is also acknowledged for the use.of the data
from their GPS buoy. and Planning Systems Inc. for valuable
assistance in the data processing.

References
Seeber. G.. and G. Wubbena (1989).Kinematic positioning with
carrier phases and 'on the way' ambiguity resolution.
Proceedings of the Fifih International Geodetic Symposium on
Satellite Positioning, Las Cruces, New Mexico, 13-17March, pp.
600-609.
DeLoach, S.R.(1995). GPS tides: a project lo determine tidal
datums with the Global Positioning System. U.S. Army Corps of
Engineers Topographic EngineeringCenter, TEC-0071
' . .[3].Parke, M., J. Blaha, and C.K. Shum (1997).Precise sea surface
measurements using DGPS buoys. Proceedings of Workshop
on Methods for Monitoring Sea Level, IGS-PSMSL, 17-18 March,
Pasadena, California, pp. 117-123.
141 Mwre, T., G. Roberts, T. Veneboer. M. Pattinson, G. Close, and.
R. Mwre. (2002). New developments in river4evel monitoring
' . using GPS heighting. Proceedings of the Institute of Navigation
International Technical Meeting, ION GPS 2002. 24-27
September, Portland, Oregon, pp. 1819-1826.
151 Schueler, T.. B. Zimmermann, G.W. Hein (2003). Radar
altimeter calibration of the Envisat satellite: An autonomous
system of high-precision for instantaneous sea surface height
determination. Proceedings of the Institute of Navigation
National Technical Meeting, 22-24January, Anaheim, California,
in press.
Harrington, J. (2002). RTK GPS Water-Level Measurement
Buoy System. Presentation at The University of Southern
Mississippi RTK Workshop, 26-28 August. Stennis Space
Center, Mississippi.
171 Bisnath, S. and D. Wells (2002). GPS Activities at the
Hydrographic Science Research Center, Department of Marine
Science, The University of Southern Mississippi. Presentationat
The University of Southern Mississippi RTK Workshop, 26-28
August, Stennis Space Center, Mississippi.
Bisnath. S.:D. Dodd. D. Wells, S. Howden, D. Wiesenburg, and
G. Stone (2003).Water level recovery with an RTK GPS-
equipped buoy. Proceeding of the 2003 U.S. Hydrography
Conference, 24-27March, Biloxi, Mississippi.

1246