Time V Distance

ionosphere-free, double-difference-carrier-phase option as en-
coded in the PAGES (Program for the Adjustment of GPS

Ephemeris) software developed by the National Geodetic Survey
(NGS). Also, we used the final precise satellite orbits
(ephemerides) disseminated by the International GPS Service
(IGS), and we fixed phase ambiguities to integer values when-
ever we were sufficiently confident in these values.
We treated one station of each baseline as the control point,
denoted C, and fixed its 3D positional coordinates to the official
values adopted by NGS. We then computed corresponding coor-
dinates for the other (unknown) point, denoted P. For the true
position of P, we adopted the average of the ten non-overlapping
24-hr solutions. Then for each baseline, we determined the dif-
ferences
in the north, east, and up dimensions
between the
true position of P and the computed position of P for each ob-
serving session. We then computed the RMS (root mean square)
value for the collection of positional differences for each baseline,
each duration, and each positional component. Recall that the
T
he Global Positioning System (GPS) has dramatically
changed the way that surveyors, GIS/LIS professionals,
engineers, and others measure positional coordinates.
These practitioners can now determine the 3D coordinates of a
new point with centimeter-level accuracy relative to a control point
located several hundred kilometers away. That control point,
moreover, may already be associated with a GPS receiver that is
being continuously operated by some institution for any of sever-
al diverse applications. The National and Cooperative CORS (Con-
tinuously Operating Reference Station) network comprises such a
set of active control points (See: www.ngs.noaa.gov/CORS/ ). To
address the practicality of using either a CORS or a passive con-
trol pointsuch as those comprising the Federal Base Network
(FBN)for providing accurate positioning control, we studied
how the precision of an observed 3D relative position between
two GPS antennas depends on the distance between these anten-
nas and on the duration of the observing session.
For our experiment, we processed 10 days of dual-frequency,
carrier phase observations for each of 11 baselines formed by
pairs of sites in the National CORS network. These 11 baselines
range in length from 26 km to 300 km, and are widely distributed
throughout the coterminous United States (Figure 1). The data
for each baseline comprised 10 non-overlapping 24-hr sessions
that were further subdivided into 20 non-overlapping 12-hr ses-
sions, 30 non-overlapping 8-hr sessions, 40 non-overlapping 6-hr
sessions and, finally, 60 non-overlapping 4-hr sessions. Moreover,
the data for each baseline and each session was processed inde-
pendently from the data of other baselines and other sessions.
In addition to the length of the baseline and the duration of
the observing session, positioning precision will depend on sev-
eral other factors, including the methodology and the software
used for processing GPS data. Here, we used the static-mode,
GPS Precision with Carrier Phase Observations:
Does Distance and/or Time Matter?
Dr. Richard A. Snay, Dr. Toms Soler, and Mark Eckl
Figure 1: Baselines of the National CORS network involved in this
study. Baseline lengths are given in km.
Figure 2: RMS values for each baseline
and each value of T. Note that the ver-
tical scale of the bottom graph differs
from that of the other two graphs.
DISPLAYED WITH PERMISSION PROFESSIONAL SURVEYOR MAGAZINE October 2002 WWW.PROFSURV.COM ALL RIGHTS RESERVED
RMS value measures the scatter among a set of numbers. The
computed RMS values are plotted in Figure 2 as a function of
baseline length. In this figure, the graph for each of the three di-
mensions
north, east, and up
includes five paths, with each

path connecting all RMS values corresponding to a common time
duration. Each path visually indicates that the corresponding RMS
values do not grow significantly as baseline length increases. In-
deed, we conducted rigorous statistical tests that corroborate this
visual interpretation. These statistical tests do not negate the pos-
sibility that RMS values grow as baseline length increases, they
simply say that any such growth is statistically negligible.
We attribute this negligible growth to the fact that we used IGS
precise orbits, as opposed to the GPS broadcast orbits, for pro-
cessing our data. IGS orbits provide satellite positions that have a
1-sigma uncertainty smaller than 0.1 m, whereas broadcast orbits
provide satellite positions that have a 1-sigma uncertainty larger
than 2 m. The fact that relative positioning error is essentially in-
dependent of baseline length also indicates that the meteorologi-
cal effects on the accuracy associated with relative GPS position-
ing are statistically the same for baseline lengths ranging between
26 and 300 km. This is probably not the case for baselines short-
er than 26 km. That is, as baselines become much shorter than 26
km, the relative meteorological effects should approach zero and
the corresponding RMS values should be smaller than those pre-
sented in Figure 2.
While distance doesnt matter under the conditions of our
experiment, the results in Figure 2 indicate that the duration of
the observing session does. The RMS values for a longer ob-
serving session are generally smaller than those for a shorter
observing session. To quantify this pattern, we averaged the re-
sults for the 11 baselines to compute a single RMS value for
each of the five considered durations and for each of the three
components of relative position. These averages should be
more reliable than the results for the individual baselines if, in-
deed, baseline length does not matter. Figure 3 displays how
these averaged RMS values decrease as the duration of the ob-
serving session increases. We empirically fit these averaged
RMS values (when expressed in cm) to an equation of the form
where T denotes the duration expressed in hours and k is a free
parameter in units of cm hour. The usefulness of the above
equation rests in its ability to predict RMS values for other possi-
ble durations in the range between 4 hours and 24 hours. The
curves in Figure 3 illustrate such predicted RMS values. Figure 4
illustrates, in a different manner, the effect of increasing the dura-
tion of the observing session by comparing the scatter in the com-
puted horizontal positions for all 6-hr sessions with that for all 24-
hr sessions. In accordance with our equation, the scatter (or RMS)
is reduced by a factor of about two by increasing the duration by
a factor of four.
The results of our study, based on sessions from 4 to 24 hours,
indicate that highly accurate positional coordinates can be ob-
tained using CORS as control points even though the distance to
these sites may be a few hundred kilometers. One drawback,
however, is the need for rather long observing sessions to realize
FEATURE
Figure 3: Predicted RMS values for the vertical and horizontal
components of relative position as a function of session duration.
Fig 4a and 4b: Scatter of horizontal
positioning errors for sessions hav-
ing 6-hour durations (4a, right) and
24-hour durations (4b, left).
these accuracies. It is costly to remain at a single site for several
hours, and this cost may be prohibitive if several tens of sites need
to be positioned. A remedy is available if two or more GPS receivers
are on hand and if the points to be positioned are clustered within
a few kilometers of a site where one of these receivers can be placed
as a local base (hub) station. This scenario would allow the re-
maining GPS receivers to accurately position the various new points
relative to the hub station by spending a relatively short time at each
new point. Meanwhile, the GPS data being collected at the hub sta-
tion can also be used to accurately position this base station relative
to one or more existing CORS. Consequently, all visited points can
be effectively positioned relative to the CORS network. When ex-
cessive spacing occurs among the proposed survey points, then
users should follow the guidelines contained in the Technical Mem-
orandum, NGS TM 58.
In this article, we have considered only that precision associ-
ated with measuring the relative position between two GPS an-
tennas. The total error involved in positioning a new point also
depends on:
the accuracy of the vertical offset measurements that relate
each points position to the position of the GPS antenna
placed above it, and
the accuracy of the positional coordinates of the control point.
The use of fixed-height poles will help to obtain reliable offset
measurements. To help mitigate errors associated with control point
coordinates, we recommend that the new point be positioned in a
statistical manner relative to two or more control points. For exam-
ple, perform a network adjustment involving each baseline con-
necting the new point to a distinct control point, and (in this adjust-
ment) constrain the positional coordinates for these control points to
their adopted values. This strategy has been encoded into the OPUS
(Online Positioning User Service) software that allows a user to sub-
mit GPS data for a point to NGS via the Internet. This data set will
then be automatically processed on an NGS computer using the
PAGES software. In this process, OPUS computes three separate es-
timates of the new points coordinates by using GPS data from each
of three suitable CORS. OPUS then averages these three positional
estimates and emails the resulting positional coordinates (with ap-
propriate documentation) to the user-specified address. Additional
information about OPUS can be found at www.ngs.noaa.gov/OPUS/.
Additional information about our studies can be found in the De-
cember, 2001, issue of the Journal of Geodesy.
DR. RICHARD A. SNAY is Manager of the National Continuously Oper-
ating Reference Station (CORS) program and a geodesist with the Na-
tional Geodetic Survey.
DR. TOMS SOLER is Chief, Global Positioning System Branch, Spatial
Reference Systems Division, National Geodetic Survey.
MR. MARK ECKL is a geodesist in the Global Positioning System
Branch, Spatial Reference Systems Division, National Geodetic Sur-
vey. Mark also serves as the geodetic advisor to the State of Delaware.
FEATURE

Time V Distance

Uploaded by

Copyright:

Available Formats

Time V Distance

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Time V Distance

Uploaded by

Copyright:

Available Formats

ionosphere-free, double-difference-carrier-phase option as en-

coded in the PAGES (Program for the Adjustment of GPS

in the north, east, and up dimensions

north, east, and up

includes five paths, with each

You might also like