Hướng dẫn bình sai GPS Huace X20 - chương V

5 Chapter V Static Baseline
The Baseline processing software has an important effect on the precision of the
relative static measurement, the reliability of the relative static measurement and
observation time. Therefore, desirable commercial baseline processing software must
be able to process baselines accurately and must be user-friendly and easy to handle.
Our software applies complicated baseline process theory to an easy-handling
interface. Normally, to have a precise result, you do not need to do any manual work on
a well-observed data. As for those unqualified data, you can make them meet your
requirement by doing a manual re-processing according to the processing information
the software gives.
5.1 Baseline Processing Flow
When you input all observation data in pointed format, the software will analyze the
relationship among collected points to form the static baselines. Then, the processing
will start.
[w8]:
The Following steps are involved:
5.1.1 Parameters Setting
The Control parameters of the baseline computation are for choosing the processing
method used for the baseline computation. It is a crucial step to set the parameters so
that can optimize the processing.
You can set these parameters in the ‘Computation Settings’. The Parameters available
here are data-sampling interval, elevation mask, reference satellite and its (ionosphere)
ionosphere, and computation model as well.
76

5.1.1.1 Fieldwork Data: Check and Modify
Before the baseline, computation is inputted, it is necessary, when fieldwork data is
entered, to check loaded observation data to avoid disoperation during fieldwork.
Checking the items include: observation station name and number, coordinates of
station and antenna height, etc.
5.1.1.2 Baseline Computation
The Typical process of computation is automatically carried out without any manual
work. The Following steps are involved:
5.1.1.2.1 Computation Self-check
the
Before doing the computation, the software will check
parameter settings,
observation data, ephemeris file and coordinate of starting pointed.
5.1.1.2.2 Read Ephemeris Data
The format of an ephemeris data could be either RINEX format, binary format *.HCN or
precise ephemeris data format SP3.
5.1.1.2.3 Read Ephemeris Data
The first step of baseline processing is reading the original observation data derived
from GPS receiver. The original data can be processed directly in the software that
comes with the GPS receiver from the same producer. However, software developed
by a third-party company may not be able to process data from different types of
receivers.
5.1.1.2.4 Triple Difference Computation
The triple difference is the difference between two double differences of different
77

epochs. The Triple Difference Computation is derived from the equation based on the
triple difference. Since in terms of short edges, the precision of results of such
computation is not high enough
, its typical function is getting approximate baseline
1
edges to process cycle slips reparation.
Single Baseline Computation Process:
Begin
Self-Check
Cycle Slip Repair
Read Ephemeris
Double-Diff. Computation
Read Obs. Data
Ambiguity
Triple-Diff. Computation
Double-Diff. Fixed
End
72. Process Chart
1. Normally, fixed double difference provides higher precision for short edges, while triple difference is
sometimes used for long edges.
78

5.1.1.2.5 Cycle Slip Reparation
It is the key point of the baseline computation to find a correct integer ambiguity.
Continuously tracking of the carrier phase by receivers provides the possibility of
having such integer ambiguity. However, blocks and disturbances may interrupt such
tracking and destroy its continuity. Therefore, So-called cycle slip, occurs in the carrier
phase observation results between different epochs. Then the main issue of the
baseline processing software becomes, how to detect and repair cycle slips.
5.1.1.2.6 Float Double Difference Computation
Suppose that signals from N satellites are observed, there would be N-1 extra
unknowns in the double-difference system-of-equations compares to that of
triple-difference. Therefore, double-difference computation derives advanced
coordinates of unknown points and float-marked integer ambiguity. Normally, this
method derives a float, which results from the absorbing of noises and others
un-modeled errors, instead of a theoretically literal ‘integer’ of the integer ambiguity.
Therefore, biases often exist, even up to several cycles, between such float and correct
integer.
5.1.1.2.7 Integer Ambiguity Solution
Under the circumstance of obtaining enough data in sufficient time, it is quite possible
to derive a correct integer ambiguity simply by rounding. Nevertheless, it is still true that
Fast Ambiguity Resolution Approach (FARA) and LAMBDA can largely accelerate
observation time and enhance working efficiency.
5.1.1.2.8 Fixed Double Difference Computation
The software will carry out Fixed Double Difference Computation when the integer
ambiguity is fixed correctly. The highest precision, will not come out without a correct
integer ambiguity, although it could be achieved by using this approach
79

5.1.1.3 Check Quality of Baseline
Only qualified baselines can be used later, thus quality checking will be required when
the baseline computation is done. The Checking process includes: RATIO, RDOP,
RMS, closed loop synchrony-sum/difference, closed loop asynchrony-sum/difference
and overlapped baseline correction
80


5.2 Settings of Baseline Processing
You need to set up baseline processing before you do the processing.
Select ‘
Baseline > Settings
’
from the Menu Bar. Below a dialog-box will appear, as
shown below:
73. Settings of Baseline Processing
There are four panels on this dialog-box
:
general settings, input rules,
troposphere/ionosphere settings and advanced settings.
For a Brief introduction of each panel see below:
81

5.2.1 General Settings
Epoch sampling interva
l
The Epoch sampling interval is the frequency, when doing baseline processing, of the
data sampled from the original observation data by the software. As the below shows:
74. Epoch Sampling Interval
Suppose that the sampling interval is 5 seconds for the static observation on two
devices. Such a high-density sampling interval often increases the time consumed
without deriving a significant improvement of the precision during the indoor work.
Therefore, a larger interval can be used to quicken the baseline processing.
What is a proper sampling interval? A common agreement is that relatively small
interval should be applied to short edges in short observation time, while larger interval,
to long edges. For example, you should set the sampling interval as 5 seconds for
static baselines shorter than 2 km within 20 minutes of observation time. However, you
should set a larger interval, which could reach 60 seconds or 120 seconds.
The reason you should set smaller interval during the fieldwork is that it is quite
possible to improve the processing result, which is derived from the limited-functional
software with imperfect and sometimes-random data, by reprocessing baseline with
modified epoch interval.
The Default epoch interval value is 60 seconds in our software.
82

5.2.1.1 Elevation Mask
The purpose of the elevation mask is to exclude the data collected from low elevation
satellites to the baseline computation.
Signals of low elevation satellite are complicatedly influenced by aerosphere and
difficult to correct with models. Additionally, many factors such as multi-path and
electromagnetic wave can put impact on them. Therefore, these signals are often poor
and should be deleted during the post processing.
Simply in terms of atmospheric refraction, the shorter observation range is the better
atmospheric refraction can be evened. Thus, a smaller elevation mask should be
applied to the short-range observation and larger value should be set on long range
ones. Off course, you should take into account the condition of the surroundings into
account.
To bring convenience into post processing, you should reduce elevation mask in
fieldworks to collect more data.
75. elevation mask
Default elevation mask is 20 degree.
5.2.1.2 Reference satellite
The Double difference is the difference of the two single differences between different
satellites. The software selects reference satellite to bring convenience to the
processing when the double difference is generated.
83

The Default setting AUTO makes the software pick a satellite obtaining most data with
the highest elevation angle as a reference. However, due to the influence of the
observation condition, it is not always the best choice. An improper reference satellite
will put a bad impact on the result of the processing, so you need to reset reference
satellite according to your data when the default option is not suitable.
You should select a new reference satellite according to the satellite alert, fieldwork
records, former baseline processing result comprehensively. It would not make any
sense to select an undetected satellite as your reference.
5.2.1.3 Gross Error Tolerance
It is often necessary to delete some unqualified data as gross error during data
processing. Data of which the observation offset exceeds (gross error tolerance × RMS)
will be taken as gross error.
5.2.1.4 Minimum Epoch Number
The Observed carrier phase must be continuous to ensure the good quality of the
baseline processing. Therefore, data with frequent cycle slips are poor-quality data and
in most cases, should be deleted. The software will consequently delete those data that
contains less continuous epochs than the minimum epoch number requires.
Minimum epoch number the software requires is two or bigger than two and the default
number is five.
5.2.1.5 Maximum Epoch Number
The Maximum epoch number is determined by the assigned memory in processing.
The Default value is 999.
84


5.2.2 Troposphere and Ionosphere Set Up
The Following figure shows the
Troposphere and Ionosphere Set Up
dialog-box. In
most cases, you do not have to make any change.
76. Troposphere and Ionosphere Set Up
85


5.2.3 Advance Settings
The figure below shows the
Advance
dialog-box. Normally, you do not have to make
any change for single frequency data. However, for dual-frequency data, you may often
change your options in the
Observation Method Combination
Combo Box.
77. Advance Settings
86


5.3 Baseline Processing
When the preparation mentioned above is done,
Select ‘
baseline
’>’
process all
’. The software will process all the baselines and an
information dialog-box will appear, shown in the figure below (figure 8-7).
The name of each baseline, computation progress and computation information of
each baseline are available in this dialog-box.
Multithreading is applied to the baseline computation.
You can click the ‘Stop’ button to stop the computation.
78. Advance Settings
87


When the computation is finished, the results of the computation are shown in the
figure below:
79. Compuatition Results
You can also view these results in the
Computation Result Details
or in the
Baseline
Properties
. For more Details, see relevant chapters
5.4 Recheck Process Results
5.4.1 Baseline Quality Control
There are several criterions of the quality of a baseline: RATIO, RDOP, RMS and Data
Deleting Rate.
A common agreement is that the larger RMS, the worse the quality of observation value
is, and the larger the RDOP, the worse observation conditions are. However, this is
comparatively true. That is to say, values of parameters above should not be taken as
absolute criterions of the baseline quality.
88

5.4.1.1 RMS
Root Mean Square (RMS) is:
V
PV
T
=
RMS
n
f
Where
V, residual error
P, observation value weight
N-f, overall observation value - an unknown
RMS shows the quality of the observation value. The smaller RMS, the better the
quality of observation value is, and the larger, the worse the quality of observation
value is.
According to the statistics theory, the probability is up to 95% of having an observation
error within the range from zero to 1.96 times of RMS
5.4.1.2 RATIO
RATIO is the ratio of the second minimum RMS and the minimum RMS when the
integer ambiguity solution came out, say:
RMS
=
RATIO
s ec
RMS
mi n
RATIO, of which the value depends on many factors, reflects the reliability of the
integer ambiguity. It relates to both the observation value quality and observation
conditions.
As a crucial criterion of the baseline quality, it is normally required to be larger than
three are.
89

5.4.1.3 Data Deleting Rate
An observation value, of which the adjusted value is larger than certain threshold value,
would be deleted as a gross error-contained value. Moreover, the ratio of deleted
values and overall values is called the data-deleting rate.
The Data deleting data, on certain aspect, reflects the quality of the original observation
data. The higher this value is, the worse the observation quality.
5.4.1.4 RDOP
Relative Dilution of Precision (RDOP) is the square root of the trace of the least
squares cofactor matrix (GPS World/UNSW), say:
=
RDOP
tr
(
Q
)
The RDOP value depends on the baseline position, satellites distribution and moving
tracks (observation condition). However, as long as baseline position is determined, it
becomes observation condition-related only. Since observation condition is expressed
as a function of time, the RDOP value is actually, for certain baseline vector,
observation session related.
The RDOP reflects the influence of the situation of the satellite on relative positioning; it
depends on the observation condition rather than on the observation value.
Apply
You should check the static baseline when process is done. As shown in the table
below , check whether all variance ratio values follow the ‘Fix’, and are larger than three.
In this table, some values are smaller than three are. That is to say, this baseline has
unqualified fixed double difference solutions.
To solve this problem, you can
•Change relevant values of the baseline processing setting
90

•Click this baseline to select it, then select ‘Process This Baseline’(or press F2)
•Repeat the two steps above until all baselines are qualified
In the result of each baseline, the line ‘Tri’ shows the fixed triple difference computation,
‘Flt’ shows the float double difference computation and ‘Fix’ shows the variance ratio of
the fixed double difference computation. Items shown along left to right are mean
square error of an edge, DX, DY DZ and the baseline length.
5.4.2 Closed Loop Checking
5.4.2.1 Conception of the Closing Error
Closed Loop Checking is a powerful tool to evaluate baseline quality.
Closed Loop includes synchrony-closed loop, asynchrony-closed loop and overlapped
baseline.
Closing Error of closed loop is theoretically zero, but it is allowed to have certain
deviation values. For more information about the allowed closing Error, see relevant
document or appendix.
There are several types of the closing Error:
1
.
Branch Closing Error:
=
X
X
=
Y
Y
=
Z
Z
2. Relative Closing Error:
+
+
2
2
2
=
X
Y
Z
S
91

Where
S
Stands for the loop length
5.4.2.2 Synchrony-closed loop, asynchrony asynchrony-closed
loop and overlapped baseline.
5.4.2.2.1 Synchrony-closed Loop
The Synchrony-closed loop closing Error is formed by simultaneous observation
baselines.
Because of certain inner relationship between simultaneous observation baselines,
synchrony-closed loop closing Error is theoretically zero. It is always true that if the
synchrony-closed loop closing Error exceeds allowed value, there is at least one
unqualified baseline among those formed by the loop. However, the opposite
hypothesis may not be true that the closing Error-within-allowed-value implies good
quality of every baseline. You can only confirm that most of baselines are qualified.
5.4.2.2.2 Asynchrony-closed Loop
If the Close loop is formed partly by the synchrony-observed baselines is called
asynchrony-closed Loop. The closing Error of the asynchrony-closed Loop is the
asynchrony-closed Loop closing Error.
If the asynchrony-closed Loop closing Error is smaller than allowed value, the
baselines forming the asynchrony-closed Loop are qualified. When the closing Error
exceeds the allowed value, there is at least one unqualified baseline. To pick
unqualified baseline vectors, you can check the adjacent
asynchrony-closed loops or
overlapped baselines.
92


5.4.2.2.3 Overlapped Baselines
The So-called overlapped baselines refer to the observation results in different session
between the same couple of the observation stations. The difference among these
results is the overlapped baselines correction.
5.4.2.3 How Do You Check?
The Following functions are available in Menu Bar ‘Check’, as shown in the figure
below:
80. Closing Error
The Closing Error values are stored in the closing Error file in the project folder, which
is a .txt file and named as ‘ProjectName.ISC’, ‘RINEX.ISC’, as example.
Note: It is for users’ continence of finding the unqualified baseline and comparing
baseline process results with each other to add the closing Error computation results of
the closing Error file ordered by the corresponding baseline computation results.
Therefore, if you do not need the previous results, you can delete your closing Error
records.
The figure below shows a closing Error file:
93


81. Check Report
For Details of the Closing Error files Formats see Chapter 13.
94


5.4.3 Check Free Network Adjustment
82. Network Adjustment Checking Result()
5.5 Distinguish Different Influence Factors
5.5.1 Influence Factors
The Following items are the influence factors of the baseline Computation:
1. Imprecise starting point coordinates: imprecise starting point coordinates leads to
bias on scale and direction.
2. Insufficient observation time leads to the failure of integer-unknown identification: if
the integer-unknowns of the satellites that involved in the baseline computation cannot
be identified precisely, the overall computation result would be spoiled.
3. Cycle slips reparation cannot be well achieved if there are too many cycle slips in
several sessions.
4. Serious multi-path effect exists within total observation session. It leads to big
residuals at large.
95

5. Troposphere or ionosphere reflections are too large.
6. Electromagnetic waves influence is too big.
7. Poor data quality due to some problems of the receiver itself, such as low-precision
phase determination and low-precision receiver clock, etc.
96

5.5.2 Influence Factors: Distinguish and Solve
5.5.2.1 Distinguish Different Factors
1. Summarize
Some of the factors are easy to distinguish, such as insufficient observation time,
redundant cycle slips, serious multi-path effect and large troposphere or ionosphere
reflections, etc., while others are not so obvious, imprecise starting point, for example.
2.Distinguish: Imprecise Starting Point
There is no easy way to distinguish the impact on the quality of baseline computation
made by Imprecise Starting Point at present. Thus, the only way to solve such
imprecision problem is making your start point as precise as possible.
3. Distinguish: Insufficient Observation Time
It is an easy-distinguished factor. You can simply check the observation data records to
find the number of observation data of each satellite. Satellites visibility figure is
available in our GPS software, which make these numbers more distinct.
4. Distinguish: redundant cycle slips
You can analyze this factor according to the obtained residual error of the observation
data. The Double difference, which is used by most typical baseline processing
applications, has an obvious integral-multiple-enlargement on its residual errors related
to an unprepared-cycle slip of a satellite observed by certain station.
5. Distinguish: Serious Multi-path Effect and Large troposphere / ionosphere reflections
Residual error of the observation data is also used to distinguish these factors. It has a
non-integral-multiple-enlargement of the residual error instead of the integral-multiple
one in the last case. Such enlargement is normally limited within 1 cycle, but it is still
bigger than that of common residual errors.
97

5.5.2.2 Solutions
1. Imprecise Start Point
To solve this problem, you should have a start point with higher precision. To do this,
you can do your single positioning in longer time or have the start point positioned
together with a more precise point in WGS-84 system, or you can derive, at the stage of
the overall baselines computation, all start points of all baselines from one single point
coordinate, to make all baseline results of the processing, having certain systematic
bias and then eliminate it, at the network adjustment stage, with imported parameters.
2. Insufficient Observation Time
To solve this problem, you can simply delete the data of the satellite with insufficient
observation time to exclude these data in the computation.
3. Redundant Cycle Slips
If the cycle slips occur in the same session of different satellites, you can delete such
session as an attempt of improving computation result. If it is a certain satellite that
often has cycle slips, you should delete the data of this satellite.
4. Serious Multi-path Effect
This factor often leads to a large residual error of the observation value. You can solve
it either by reducing the edit threshold to get rid of the values with large residual error,
or by deleting seriously affected sessions/satellites.
5. Large troposphere/ionosphere reflections
You can solve this by the following methods:
1) Increase the elevation mask to eliminate those data that tend to be effected by
troposphere/ionosphere. However, blindness exists in this method, since not all
low-elevation angle signals are seriously affected by troposphere/ionosphere.
2) Correct troposphere/ionosphere delay by models
3) For dual-frequency observed value, use ionosphere reflections-free observation
value to run baseline computation
98


5.5.2.3 A Useful Tool of Refined Process: Residual Error Chart
The Residual Error Chart is quite helpful to estimate influence factors and to
determine problematic sessions and satellites. A Residual Error Chart is a
chart drawn according to obtained residual errors, as shown in the figure
below.
Click
‘
up
’
/
’
down
’
to view each double difference residual error.
83. Residual Error Chart
Above figure is a typical chart of the double difference residual error chart, of which the
X-axis represents observation time and the Y-axis represents residual error.
SV14-SV07 on the top-right is the double difference residual error of satellite SV12 and
satellite SV15. Normally, the swing refers to 0 of residual error values, would not be
larger than 0.1 cycles.
99

The Figure below shows a cycle slip:
84. cycle slip
The data in the figure below shows the influence of unknown factors; they may be one
or more factors among multi-path effect, troposphere/ionosphere reflections, and
serious electromagnetic disturbance and so on.
85. Unknown Factor(s) Disturbance
100


5.6 Re-process a Baseline
When the influence factors are estimated. You can re-process a baseline by modifying
its processing settings or by editing the baseline sessions.
To select the data you want to delete, drag the pointer in the observation data chart. As
shown in the figure below, data framed with broken-line would be blocked and would be
involved in the computation.
86. Delete Observation Data
In some cases of unqualified baseline processing, you probably need to modify the
processing settings and edit sessions more than once. In some cases of
fail-to-have-qualified-solution may even happen on certain baselines. In this case, you
should delete them or block them from the network adjustment. If such baselines are
curial for your network, you have to measure them again!
101