Performance of RTK Positioning in Forest Conditions:
Mieczysław Bakuła1; Stanisław Oszczak2; and Renata Pelc-Mieczkowska3
Abstract: This paper presents the performance of the real-time kinematic ?RTK? technique in very severe conditions. The measurements
were conducted in forest conditions during two different experiments. Two Global Positioning System receivers were used: an Ashtech
Z-Surveyor and a Z-Xtreme. Radiomodems were used for transmitting RTK corrections from a reference station to a rover. The RTK
measurements were performed during quite good satellite configuration. The use of RTK technology, based on the Ashtech Z-Xtreme
receiver as a rover in forest terrain, allowed centimeter accuracy to be obtained. However, practical experiments showed that there is a
need for redundant independent RTK solutions based on repeated independent ambiguity reinitializations. Although gross errors might
occur, the RTK technique can be very helpful technology for centimeter positioning in woodland areas.
CE Database subject headings: Global positioning; Geographic information systems; Satellites; Geodetic surveys; Kinematics.
Nowadays, it is well known that the real-time kinematic ?RTK?
mode of Global Positioning System ?GPS? can provide a user
with centimeter precision and accuracy in open terrain. It involves
a reference station transmitting its raw measurements or observa-
tion corrections to a rover receiver via a telemetry link. RTK uses
the carrier signal in addition to the code signal. The remote re-
ceivers use transmitted RTK data to compute a corrected position.
A communication link must exist between the base and remote
receivers such as a very high frequency or ultrahigh frequency
radio, a cellular telephone, or any other medium that can transfer
digital data. The data processing at the rover site includes ambi-
guity resolution of the differenced carrier phase data and coordi-
nate estimation of the rover position. That goal is achieved by the
phase observation of L1 and L2 frequencies. Carrier phase mea-
surements are extremely precise, but they contain an unknown
integer initialization constant called “phase ambiguity.” There-
fore, RTK positioning has to first resolve integer ambiguities to
achieve a high level of precision and accuracy. The process of
ambiguity resolution is referred to as initialization.
If there are no obstacles between satellites and a GPS antenna,
the RTK positioning is extremely efficient for surveying applica-
tions and navigation, but there are some situations where visibil-
ity to the satellites is limited due to buildings, trees, and the like
?Hasegawa and Yoshimura 2003; El-Mowafy 2000; Naesset 2001;
Naesset and Jonmeister 2002; Sigrist et al. 1999; Yoshimura and
Hasegawa 2003; Bakuła et al. 2006?. In such situations, a GPS
operator might have some problems with ambiguity resolution
and its validation.
Following practical problems with RTK gross errors and vali-
dation in forest conditions, this paper investigates the use of the
same equipment ?hardware and software? as was presented in the
paper by Bakuła and Oszczak ?2006? in a practical GPS RTK
project performed under very difficult observational conditions,
such as heavy shrubbery and under trees. When desired, RTK
technology could provide more precise positioning in forest envi-
?Rodríguez-Pérez et al. 2007?. The RTK technique has one big
advantage over postprocessing positioning—calculated positions
are available directly in the terrain. However, in the postprocess-
ing positioning techniques in forest conditions, it is never known
how long the static GPS sessions should last in order to obtain
approval results in terms of accuracy and reliability. In such situ-
ations, even with an experienced GPS operator, one point to be
measured needs to be occupied several times ?Oszczak et al.
2002? and acceptable accuracy is not guaranteed. Nevertheless,
the RTK technique seems to be a more economical and efficient
method in difficult observational conditions due to many addi-
tional reinitializations of ambiguity solutions allowing indepen-
dent redundant results to be obtained.
Hardware and Software Configuration
The GPS FieldMate is a real-time GPS for surveying applications.
The system operates in geodetic or local coordinate systems and
allows the user to perform layout or pickup surveys ?Ashtech
1998a?. Data can be stored in the handheld or the receiver,
1Assistant Professor, Chair of Satellite Geodesy and Navigation, 5
Heweliusza St., 10-724 Olsztyn, Univ. of Warmia and Mazury, Poland
10-724. E-mail: firstname.lastname@example.org
2Full Professor, Head of the Chair of Satellite Geodesy and Naviga-
tion, 5 Heweliusza St., 10-724 Olsztyn, Univ. of Warmia and Mazury,
Poland 10-724. E-mail: email@example.com
3Ph.D. Student, Chair of Satellite Geodesy and Navigation, 5 Hewe-
liusza St., 10-724 Olsztyn, Univ. of Warmia and Mazury, Poland 10-724.
Note. This manuscript was submitted on March 8, 2007; approved on
January 14, 2009; published online on July 15, 2009. Discussion period
open until January 1, 2010; separate discussions must be submitted for
individual papers. This technical note is part of the Journal of Surveying
Engineering, Vol. 135, No. 3, August 1, 2009. ©ASCE, ISSN 0733-
JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009 / 125
depending on the operational mode of GPS positioning. The sys-
tem can operate in Radio Technical Commission for Maritime
Services or carrier phase differential ?CPD? modes. The CPD
message can provide more complete information on position, ve-
locity, solution status, position root mean square ?RMS? and co-
variance, number of satellites, and position dilution of precision
?PDOP?, ?Ashtech 1998b?. In the real-time modes, the system
displays a constant precision value and ambiguity resolution re-
sults, i.e., fixed or float. The handheld software provides the user
with multipoint logging and offset options and offers several dif-
ferent modes of navigation. The GPS FieldMate system consists
of two main components: the base and the rover. The base system
is composed of a GPS base receiver, a radiomodem, and a radio
antenna; but the rover system is also composed of a GPS receiver,
a radio antenna, and an additional handheld controller, e.g., a
Husky computer ?Fig. 1?.
The base GPS antenna is located over a reference point that
the coordinates are known ?fixed coordinates? and entered into the
base receiver. Having calculated correction data, based on the
reference point coordinates and GPS measurements, the base GPS
receiver sends it to the base radio via a serial cable. Next, the base
radio broadcasts the corrections to the rover receiver. In real time,
the rover radio receives the correction information through the
remote radio antenna and sends the data to the rover GPS receiver
via a serial cable. The rover GPS receiver uses measurements
from the base and applies them to derive a corrected position for
the rover GPS antenna. The position information is transferred to
the Husky controller though a serial cable where the GPS Field-
Mate software converts the position data into local coordinates
using transformation parameters. Local coordinates are displayed
along with the current positional precision. The Ashtech Instant
RTK technique applied in Z-Xtreme receivers allows centimeter
accuracy to be obtained within a few seconds in open areas
?Abousalem et al. 2001?.
Practical RTK Tests
One point, LAS1, located in the middle of a park at the University
of Warmia and Mazury was chosen in order to perform RTK
positioning in forest conditions ?Fig. 2?. Around the point there
were many large trees. Two RTK observational experiments were
performed. The first was executed in April 15, 2005, from 15:30
to 16:52, and the second, 4 days later, from 11:18 to 12:39 local
time ?13:18–14:39 UTC?. There were no leaves on the trees on
the days of these tests, so the conditions under which the experi-
ments were executed were better than during other seasons of the
year. The best configuration of satellite geometry for that time
was chosen with the use of Mission Planning software ?Figs. 3
The Ashtech Z-Xtreme GPS antenna was set up on the tripod
and centered at Point LAS1 as the rover receiver. Precise coordi-
Fig. 1. Base and rover configuration during tests
Fig. 2. Location of Test Point LAS1
126 / JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009
nates of Point LAS1 were known previously based on classical
The Husky handheld computer and GPS FieldMate program
determined the rover RTK configuration. That architecture of the
rover RTK station also allows the monitoring of the accuracy of
positions calculated every second. When the Ashtech Z-Xtreme
works alone without a handheld computer, the RTK positions are
recorded on a PC card into c*. * files every second, but without
the added accuracy of the determined coordinates. However, po-
sitions recorded into the handheld controller have a frequency of
about 4–5 s and give horizontal and vertical accuracies.
The base RTK station was at a distance of 430 m and situated
on the roof of the building of the Faculty of Geodesy and Land
Management, with unobstructed sky. In view of the numerous tree
branches, the measuring conditions during GPS observations
were very difficult. The obstructions caused changes in values of
PDOP, considerably limiting the accessibility of satellites as well
as causing frequent losses of continuity of satellite signals. It
caused gaps in GPS observations for all satellites, which deterio-
rated their overall geometry. Therefore, the real availability of
satellites ?Figs. 5?a and b?? as well as their geometrical configu-
ration were much worse than in the Mission Planning software
anticipated before executing RTK experiments.
On the first day of RTK tests, there were 1,200 positions, but
only 143 ambiguities were fixed ?Fig. 6?a??. On the second day of
measurements, 992 RTK positions were obtained, with integer
ambiguities for 558 positions ?Fig. 6?b??.
It can be seen ?Figs. 6?a and b?? that positions with float solu-
tions are within a few meters as in the case of fixed solutions. A
system of grid coordinates ?x, y? in which all the RTK positions
are presented is obtained from the Transverse Mercator projec-
The results are deployed quite around the true positions of the
observed point, LAS1, during the first experiment, but in the sec-
ond one, the positions are much closer to the “true” position of
LAS1. Note that satellite availability during the RTK measure-
ments ?Fig. 5?b?? ?second day of experiment? was better than in
the first experiment ?Fig. 5?a??. It should also be noted that posi-
tions with fixed solutions might be arranged into separate groups
with a scatter value of every group of about a few centimeters
?Fig. 7?. The values in brackets present the number of fixed po-
sitions, but the values before brackets—the number of the group.
It can be seen that scatter values between groups were in the
range of a few meters. Some statistics of the groups are presented
in Tables 1 and 2. The accuracy for east ??y?, north ??x?, and
height ??h? components was estimated with the use of true coor-
Fig. 3. Satellite trajectories during the RTK tests: ?a? April 15, 2005; ?b? April 19, 2005
Fig. 4. Satellite availability and values of PDOP: ?a? April 15, 2005; ?b? April 19, 2005
JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009 / 127
dinates of the rover point, but precision was estimated based on
the average of every component. As it is well known, accuracy
refers to the closeness of the observations, to the true value, but
precision refers to the closeness of repeated observations, to the
sample mean. It follows that accuracy includes precision plus
systematic and other errors. In RTK positioning, ambiguity reso-
lutions have a strong influence on the final accuracy. If validation
procedures accept a misleadingly wrong set of ambiguities, only
the precision is higher but the accuracy is biased by gross errors.
On the first day of the experiment ?Table 1?, there were only
53 RTK fixed positions with an accuracy of about 1–3 cm in
three-dimensional space, for the 13th group. In other groups, the
accuracy was different for every group and every component was
in the range of meters. However, the precision in those groups
was within centimeters except for the first group, where the height
component had a precision of 0.378 m.
On the second day of the RTK experiment ?Table 2?, there
were as many as four groups ?1, 2, 7, and 10? where the accuracy
was reliably estimated, i.e., ambiguities had been properly re-
solved and, therefore, the precision was very similar to the accu-
racy, in the range of centimeters. In other groups, the accuracy did
not correspond to the precision, even though the precision of
every group was within a few centimeters.
In RTK positioning in forest conditions, the number of satel-
lites and the values of PDOP have an influence on the accuracy
and validation of ambiguity resolution as in open areas. It should
be noted that the accuracy of RTK positioning increases if the
value of PDOP is steady for a longer time and its value is as low
as possible. In this situation, short-term jumps in PDOP do not
have an influence on accuracy. Subsequently, low values of hori-
zontal RMS and vertical RMS allow us to fix ambiguities, but the
accuracy of these solutions is similar to float and is not always
reliable. It means that the ambiguity resolution strongly depends
on obstacles and multipath errors.
The entire history of fixed and float RTK solutions and their
accuracy versus the values of PDOP are presented in Figs. 8?a and
b?. The time-to-fix ?TTF? of the ambiguity varied from a few
seconds to 25 min in the first experiment ?the first ambiguity so-
Fig. 5. ?a? Number of satellites and values of PDOP during the first
experiment; ?b? number of satellites and values of PDOP during the
Fig. 6. ?a? Horizontal deployment of RTK positions during the first
day of experiments; ?b? horizontal deployment of RTK positions dur-
ing the second day of experiments
128 / JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009
lution was obtained after 3 min? and from seconds to 12 min in
the second experiment ?the first ambiguity solution was obtained
after 100 s? ?Table 3?.
It was noted that short TTF ?10–15 s? coincided with closer
coordinates of float and fixed solutions and, therefore, that “reini-
tializations” remained within that group ?Table 3?. Losses of fixed
ambiguities ?up to 30 min? were usually connected with evident
change in accuracy. The first positions with centimeter accuracy
were obtained after 80 min in the first experiments and after only
2 min in the second experiment.
During the second experiment, more satellites were at higher
elevation than in the first experiment, although the PDOP had a
lower value. Higher elevations of visible satellites allowed much
more efficiently fixed solutions to be obtained and four groups of
fixed RTK positions obtained at different times allowed treating
those positions as closer to the true coordinates. In the first ex-
Fig. 7. Deployment of RTK-fixed coordinates: ?a? April 15, 2005; ?b? April 19, 2005
Table 1. Analysis of the Precision and Accuracy of Fixed RTK Positions
during the First Experiment ?April 15, 2005?
Table 2. Analysis of the Precision and Accuracy of Fixed RTK Positions
during the Second Experiment ?April 19, 2005?
Fig. 8. ?a? Accuracy of float and fixed RTK solutions versus PDOP
values in the first experiment; ?b? accuracy of float and fixed RTK
solutions versus PDOP values in the second experiment
JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009 / 129
periment, despite the different accuracies of every group, it was
not possible to decide which of them was correct; in this situation,
more RTK observations should be performed in order to gather
further independent RTK positions.
Summary and Conclusions
The main aim of the experiment was to investigate the accuracy
of RTK positioning in forest conditions. Based on the research, it
is clearly possible to obtain centimeter accuracy of a rover GPS
RTK antenna in forests, despite a weak GPS signal and many
permanent obstacles. Forest observational conditions made ex-
tremely difficult conditions for GPS observations owing to lim-
ited access to the sky. The RTK technique has one major
advantage over postprocessing GPS methods; positions are di-
rectly determined on the terrain and many independent ambiguity
values might be resolved, allowing redundancy of RTK position-
ing. Although many fixed RTK solutions are biased, a GPS op-
erator can gather as many independent RTK solutions as needed
with the same precision and accuracy. If there are many indepen-
dent groups of RTK ambiguity solutions where positions are the
same within a few centimeters, the results in those groups can be
treated as true and valid. It is clear that the time of RTK position-
ing in forest conditions should be chosen during the best satellite
constellation and in the season when there are no leaves on trees.
The influence of trees located near the GPS antenna had a strong
influence on ambiguity resolution in terms of the time-to-fix the
ambiguity resolution and its reliability. This problem is well
known and has been investigated by many researchers. More re-
search is still required in terms of ambiguity validation proce-
dures. From a practical point of view, an RTK positioning
operator should be very careful during RTK positioning near the
trees and additional reinitialization of ambiguity should be carried
out as redundant observables.
Integration of different global navigation satellite systems—
GPS, GLONASS, and, in the future, GALILEO—might be prom-
ising in terms of validations of RTK positions in forest conditions
because of the use of independent reference frame systems.
Abousalem, M., Han, S., Qin, X., Martin, W., and Lemoine, R. ?2001?.
“Ashtech instant-RTK: A revolutionary solution for surveying profes-
sionals.” 3rd Int. Symp. on Mobile Mapping Technology.
Ashtech. ?1998a?. “GPS FieldMate.” Operation and reference manual,
Magellan, Sunnyvale, Calif.
Ashtech. ?1998b?. “Z-family.” Technical reference manual, Magellan,
Bakuła, M., Oszczak, S., Pelc-Mieczkowska, R., Suchocki, M., Chros-
towska, M., and Rudziński, M. ?2006?. “Analysis of precision and
accuracy of GPS measurements in forest conditions.” Polskie To-
warzystwo Informacji Przestrzennej, 4?3?, 23–32.
Bakuła, M., and Oszczak, S. ?2006?. “Experiences of RTK positioning in
hard observational conditions during Nysa Kłodzka River project.”
Reports on Geodesy, 1?76?, 71–79.
El-Mowafy, A. ?2000?. “Performance analysis of the RTK technique in an
urban environment.” The Australian Surveyor, 45?1?, 47–54.
Hasegawa, H., and Yoshimura, T. ?2003?. “Application of dual-frequecy
GPS receiver for static surveying under tree canopy.” J. Forest Re-
search, 8?2?, 103–110.
Naesset, E. ?2001?. “Effects of differential single- and dual-frequency
GPS and GLONASS observations on point accuracy under forest
canopies.” Photogramm. Eng. Remote Sens., 67?9?, 1021–1026.
Naesset, E., and Jonmeister, T. ?2002?. “Assessing point accuracy of
DGPS under forest canopy before data acquisition, in the field and
after postprocessing.” Scandinavian J. Forest Research, 17?4?, 351–
Oszczak, S., Bakuła, M., Surowiec, S., Czarnecki, T., Kiszkiel, S., and
Zaleski, K. ?2002?. “GPS measurements of forest control points.”
Proc., XII Scientific Conf.: Geographic Information Systems, Polish
Association for Spatial Information, Warsaw, 133–141.
Rodríguez-Pérez, J. R., Álvarez, M. F., and Sanz-Ablanedo, E. ?2007?.
“Assessment of low-cost GPS receiver accuracy and precision in for-
est environments.” J. Surv. Eng., 133?4?, 159–167.
Sigrist, P., Coppin, P., and Hermy, M. ?1999?. “Impact of forest canopy
on quality and accuracy of GPS.” Int. J. Remote Sens., 20?18?, 3595–
Yoshimura, T., and Hasegawa, H. ?2003?. “Comparing the precision and
accuracy of GPS positioning in forest areas.” J. Forest Research,
Table 3. Analysis of TTFs of Ambiguities during the Experiments
April 15, 2005April 19, 2005
Number of reinitializations
in each group
Number of reinitializations
in each group
130 / JOURNAL OF SURVEYING ENGINEERING © ASCE / AUGUST 2009
Page 7Download full-text