Professional Documents
Culture Documents
NRCan CSRS-PPP-v3 - Tutorial EN
NRCan CSRS-PPP-v3 - Tutorial EN
Simon Banville
Canadian Geodetic Survey, Surveyor General Branch, Natural Resources Canada
Last update: 2020-08-25
Motivation __________________________________________________________________ 1
New Features ________________________________________________________________ 2
Ambiguity Resolution __________________________________________________________ 3
Principles _________________________________________________________________________ 3
What to Expect ____________________________________________________________________ 3
Ambiguity Status Plot _______________________________________________________________ 5
Example 1 _______________________________________________________________________________ 6
Example 2 _______________________________________________________________________________ 7
Example 3 _______________________________________________________________________________ 8
Motivation
The Canadian Spatial Reference System Precise Point Positioning (CSRS-PPP) service was
launched in 2003. It allows users of global navigation satellite systems (GNSS) to collect data in
the field and upload this data to Natural Resources Canada (NRCan) servers. Within minutes,
users receive an estimate of their position, or trajectory, along with quality estimates and a report
for visual quality control.
CSRS-PPP uses precise satellite orbit, clock and bias corrections derived from a global network of
receivers to determine accurate user positions. This strategy differs from other online positioning
services based on differential positioning, i.e., using nearby base stations. CSRS-PPP enables up
to millimeter-level accuracies for long observation sessions (24+ hours) in static mode, but an
accuracy of a few centimeters can typically be achieved in an hour.
On 16 August 2018, a new version of the software supporting CSRS-PPP was launched. This
software replacement was the first step in the CSRS-PPP modernization plan which includes PPP
with ambiguity resolution (PPP-AR), faster convergence using external ionospheric information
and processing of multi-GNSS observations. CSRS-PPP version 3, described in this document,
implements PPP-AR.
In the past decade, PPP-AR algorithms have matured from experimental to operational solutions.
Ambiguity resolution offers significant benefits for users by transforming ambiguous carrier-
1
phase observations into precise ranges. As a result, centimeter-level accuracies can often be
obtained more rapidly. Furthermore, due to satellite geometry, resolving carrier-phase
ambiguities results in improved estimates for the longitude (east) component.
This document guides the user through the new features of this software. It explains basic
principles of ambiguity resolution and presents, through examples, what CSRS-PPP version 3
users should expect in terms of performance.
New Features
The main novelty of CSRS-PPP version 3 is the implementation of PPP-AR algorithms. Apart from
this substantial change, other modifications could impact results with respect to CSRS-PPP
version 2, such as:
Changes to outputs: the tropospheric zenith delay and receiver clock estimates were
moved from the ‘.pos’ file to the ‘.tro’ and ‘.clk’ files, respectively. The ‘.tro’ and ‘.clk’ files
are IGS standards.
In static mode *only*, the reported standard deviations for the receiver clock and
tropospheric zenith delay estimates are scaled to obtain more realistic values. The applied
scale factor is the same as for position standard deviations, which were already being
scaled in CSRS-PPP version 2. As a consequence of this scaling, it is possible that reported
sigma values in static mode are larger than in kinematic mode.
The use of observable-specific satellite biases: to minimize errors that could impact
ambiguity resolution, CSRS-PPP version 3 only accepts signals for which it has explicit
code-bias corrections. For this reason, it is strongly suggested that users submit GNSS
data in RINEX 3 format.
Improved day boundary crossing: when observations spanned more than one GPS day,
users of CSRS-PPP experienced a discontinuity at midnight due to clock alignment issues
between days. These discontinuities have been reduced in CSRS-PPP version 3, although
they could still occur (see section Day Boundaries).
Modified cycle-slip detection thresholds: tighter thresholds are now used to improve
detection of half-cycle-slips. As a natural consequence of this change, more cycle slips
could be reported in CSRS-PPP version 3.
2
Ambiguity Resolution
CSRS-PPP version 3 introduces ambiguity resolution capabilities for data collected on or after 1-
January 2018. This section first describes, at a high level, the basic principles of this feature. Then,
examples are provided to guide users through a better understanding of their results.
Principles
When applying a specific set of corrections to carrier-phase observations and estimating phase-
bias parameters, the ambiguity parameters set up in the PPP filter have an integer nature. This
means that, using sophisticated algorithms, it is possible to identify these integers and constrain
their values. Adding this additional information to the solution typically improves accuracy and
reduces reported uncertainties.
In CSRS-PPP version 3, ambiguity resolution is performed on the backward run. This means that
epochs are first processed in a chronological order to ensure that all information is available prior
to resolving ambiguities. Then, starting with the last epoch successfully processed, ambiguity
resolution is attempted. This process is repeated for all epochs in reverse order (from last to first
epoch). It should be noted that this sequential algorithm is not theoretically optimal. While
attempting ambiguity resolution on all ambiguities simultaneously should yield better results, it
is subject to significant computational load constraints for datasets with several hundred
ambiguity parameters. The sequential algorithm implemented is fast and often provides results
identical to the optimal algorithm.
Once the backward run is completed, a final back-substitution is performed to obtain the final
values for all parameters: tropospheric zenith delays, receiver clocks and positions (in kinematic
mode).
What to Expect
While successfully resolving ambiguities is beneficial for the solution, incorrectly resolving some
or several ambiguities can have disastrous consequences. For this reason, several mechanisms
were implemented to reduce the risks of incorrectly resolving ambiguities:
Ambiguities associated with less than 5 minutes of data are not resolved.
Partial ambiguity resolution is performed, meaning that not all ambiguities need to be
resolved.
3
The validity of the stochastic model is taken into consideration in the ambiguity validation
process. This check implies that if ambiguity residuals (i.e., the difference between the
float and integer estimates) are much larger than their precision, the process is aborted.
While ambiguity resolution can sometimes be successful with as little as 15 minutes of data,
longer observation sessions are usually associated with more ambiguities being resolved.
The ‘.sum’ file contains a tag called “IAR”, reporting the percentage of fixed ambiguities. This
value is computed by considering carrier-phase observations with fixed ambiguities rather than
arcs with fixed ambiguities. For example, let us consider a simplistic example with two satellite
arcs:
Satellite G01 is observed for 90 epochs and has its ambiguities resolved
Satellite G02 is observed for 10 epochs but ambiguities could not be resolved
In this case, the ambiguity success rate is 90% since it represents the percentage of carrier-phase
observations with fixed ambiguities.
The percentage of fixed ambiguities is also reported on the front page of the PDF report, as shown
in Figure 1.
4
Figure 1: PDF report showing the percentage of fixed ambiguities
Olive: float ambiguity. This ambiguity could not be resolved to an integer, either because
the software could not do it with enough confidence or because ambiguity resolution is
not activated for a system (e.g., GLONASS).
Cyan: datum ambiguity. CSRS-PPP version 3 implements the ambiguity datum concept of
Collins et al (2010). To estimate the additional receiver phase-bias parameters set up in
the PPP filter, the filter must fix a priori a certain number of ambiguities. These selected
ambiguities are called datum ambiguities.
Green: fixed ambiguity. Integer ambiguities that were validated by the software’s
algorithms are indicated in green.
Let us look at a few examples of the ambiguity status plot to learn how to interpret them.
5
Example 1
This example shows data from a 24-hour static session collected by high-quality geodetic
instruments. As expected, most GPS ambiguities have been resolved (98.52%) and are indicated
in green. Since ambiguity resolution is not enabled for GLONASS, all ambiguities are float and
plotted in olive. If your ambiguity status plot looks like this, it is quite likely that the PPP-AR
solution is of good quality.
6
Example 2
This example shows a short observation session lasting 15 minutes in which ambiguity resolution
was successful. Since CSRS-PPP version 3 performs partial ambiguity resolution, not all
ambiguities were resolved. The PDF report indicates that 68.54% of ambiguities were resolved.
Even though this number may seem low, there is no reason to believe that the solution
experienced any issues. Satellite G06 was not observed for the whole session, resulting in too
short an observation period to confidently resolve its ambiguities. While G31 was observed for
the whole session, this satellite was tracked at a low elevation angle and its ambiguities could,
again, not be resolved confidently.
7
Example 3
This example shows a 30-minute observation session in which no ambiguities were resolved. The
reason should be obvious when looking at the ambiguity status plot: many cycle slips (indicated
in red) contaminated the data which resulted in very short arcs. In this case, the software could
not confidently resolve ambiguities and provided a float solution.
Figure 4: Ambiguity status plot for a session with several cycle slips
8
Day Boundaries
A benefit of the “integer” satellite clock corrections used in PPP-AR solutions is that they can be
precisely aligned from day to day. As a consequence, position jumps occurring when observations
span two GPS days can be minimized. As an example, data from station ALGO located in Canada
was processed from 2019-08-14 23:00:00 to 2019-08-15 01:00:00 using CSRS-PPP version 2
(Figure 5) and version 3 (Figure 6).
9
With CSRS-PPP version 2, position jumps of approximately 6 cm in latitude and 7 cm in longitude
can be noticed, even though the receiver maintained continuous lock on all satellites. This is a
consequence of IGS clock products not being aligned between two consecutive days. With CSRS-
PPP version 3, no jump can be seen in latitude and the discontinuity in longitude has been
reduced to 1.5 cm. The latter is caused by satellite orbit interpolation errors associated with the
use of two orbit (SP3) files. Still, the benefits of PPP-AR (and the products used) are obvious.
10
References
Collins P, Bisnath S, Lahaye F, Héroux P (2010) Undifferenced GPS ambiguity resolution using the
decoupled clock model and ambiguity datum fixing. Navigation 57(2):123–135
11