Sect_4.4.3

4.4.3 Testing GPS Surveying Systems:

CERTIFICATION TESTS

An attempt to test the performance of the GPS instrumentation (hardware and software) under realistic conditions.
The implicit assumption is that the contribution of all GPS system biases to baseline results is largely a function of baseline length and not a function of time or geographic location.
Some tests are conducted once, for example in the case of newly developed GPS surveying instrumentation, while other tests may be conducted as deemed necessary.
Some tests are conducted as a form of certification, or validating the claimed performance of GPS survey instrumentation.
Some tests involve specific GPS instrument units, others are tests applied to an entire class or brand of instrumentation.

Three types of tests in this category shall be discussed:

FGCS Tests

In January 1983, the U.S. Federal Geodetic Control Sub-committee (FGCS) conducted the first in a series of tests of GPS satellite surveying systems and associated commercial software. The first system tested was the Macrometer^TMV1000, a codeless carrier phase survey system. Many other GPS survey systems have been tested since late 1985, including: Texas Instruments TI4100, Trimble 4000S (4-channel), ISTAC 2002, Trimble 4000SX (5 channel), Wild-Magnavox WM101, Motorola Eagle, Sercel NR52, Ashtech XII, Ashtech LD-XII (24 channel), Trimble 4000SST (16 channel), as well as recent models of the Trimble, Ashtech, Leica, as well as other companies.

The following additional comments can be made:

The FGCS have tested all GPS surveying instrumentation as it has been released onto the market, and continues to do so. These receivers have included single frequency and dual-frequency instruments. No other country has systematically tested GPS surveying receivers in this way. It therefore functions as a de facto Certifying Agency.
All tests are performed over a network of up to ten stations located in the vicinity of Washington D.C. The baselines for the network vary in length from 186m to about 105km. The standard for evaluating the GPS baseline results are coordinate differences compared with those determined using 1-2 part per million (ppm) terrestrial survey techniques, and a data base of accumulated baseline measurements from all the FGCS test surveys.
The data collection is carried out by survey personnel provided by the GPS manufacturer, and is reduced by the same personnel using the manufacturer's software. The raw GPS observations are immediately processed using the broadcast ephemerides, under campaign conditions as close to those that would normally be encountered as possible.
Such a network was initially used only to test conventional static GPS surveying techniques, however modern high productivity surveying techniques (section 5.5.1) have been recently tested as well, but only for the short baselines. Some real-time GPS surveying tests have been conducted.
The analysis of the results of the short baselines (< 2km), estimates the base uncertainties.
The analysis of the results of the medium length baselines (ranging in length from 8 to 105km), estimates the line-length dependent uncertainties.
GPS results are evaluated in terms of: Cartesian coordinate difference (dX, dY, dZ), baseline lengths, ellipsoidal height differences, and azimuth.
Computations are analysed according to: repeat baseline measurements, loop miscloses, minimally constrained 3-D Least Squares adjustments, comparisons with the terrestrial standard, and comparisons with past FGCS GPS test survey results.

Summary of data analyses and results:

Fixed orbital coordinate solution using the broadcast ephemerides (assumed to have introduced an uncertainty no larger than 1-2 ppm).
Ionospheric refraction correction accounted for in dual-frequency systems by using the L3 "ionosphere-free" combination for long baselines (section 6.4.2), and ignored for short baselines. Other systems which used L1 phase data only could not ensure the same accuracy results for long baselines.
Internal consistency (or repeatability) of the baseline results were characterised by the following estimated uncertainties in each component (at the1-sigma, 67% confidence level):
- "Base" error: 0.3 to 1cm
- "Line-length dependent" error: 1 to 2 ppm
- Combined uncertainty therefore is (cm):
where B is the baseline length in kilometres.

The Issue of "Legal Traceability"

There has recently been a move to establish procedures in Australia and other countries for the regular testing of GPS survey instrumentation and staff. This initiative derives from the need to define "legal traceability" of GPS survey results for cadastral surveys (BOEY & HILL, 1995), in a similar manner to the calibration of EDM on official baselines. It has been recognised that there may need to be several levels of testing:

An "integrity monitoring" network that continuously tests the health of the GPS system.
Regular "zero baseline" tests administered by the user.
The definition of a set of test procedures and the provision of test network facilities by State Lands and/or Survey Departments. Such a network could consist, for example, of a number of high precision GPS survey control points, in a configuration that may be similar to that of the FGCS network.
Network and procedures may have to be defined for both the conventional static GPS survey technique, and the modern "stop & go" and "rapid static" GPS techniques.
An alternative approach to the above could be based on the principles of "Total Quality Management", by making the surveyor responsible for ensuring that his/her GPS instrumentation is in satisfactory compliance with the operational performance standards defined by the Certifying Authority. In this case, a user-pays service for rigorous testing of GPS receivers may be provided.

The following additional comments can be made:

Each State or Country would develop its own guidelines for testing. For example, such tests maybe specifically concerned with validating GPS for cadastral surveys. Alternatively, a general certification may be given.
It is important to define not just the network of test points to be surveyed, but also the field and processing procedures to be used.
If modern high productivity GPS procedures, such as "stop & go" and "rapid static", are to be tested, a typical survey environment must be ensured. Hence the inclusion of trees and other obstructions is important in order to realistically test ambiguity reinitialisation procedures.
There should be a variety of baseline lengths within the test network. In the case of networks to test conventional static GPS, the baselines could be up to 50km in length (or greater). However, "rapid static" and "stop & go" GPS surveys are generally carried out only over baselines less than 10km in length.
The problem with such campaign style GPS testing is that they provide merely a "snapshot" of GPS performance. The challenge is how to investigate the impact of varying observation conditions:
- changes in receiver-satellite geometry (as experienced through the day),
- different numbers of available satellites (as a result of signal obstructions),
- day and night observations (to induce differences in atmospheric conditions),
- varying temperature conditions (seasonal effects?), and
- varying site conditions (to induce a variety of multipath and signal jamming scenarios).

See BOEY & HILL (1995) for a discussion of the options being considered for the state of Victoria, in Australia.

Pre-Mission Testing

If there is any doubt concerning the proper functioning of the GPS equipment, or there is unfamiliarity by the field staff with the equipment to be used, and to verify that the hardware and software is functioning to the required accuracy, then some localised testing could be undertaken. Such a test could be carried out over:

a "zero baseline", or
a micro-network, to test instrument functioning, antenna connections, battery and power requirements, etc., or
a "standard" validation test network, as described above (under "Legal Traceability"), to verify the average performance of GPS.

Such testing is strictly the prerogative of the surveyor, hence he/she could make the testing as sophisticated and comprehensive as they wished. Furthermore, regular testing could be considered prudent if the surveyor is operating under the principles of "Total Quality Management". At present there are no test networks established for these purposes in Australia, although the new coordinates determined in the Geocentric Datum of Australia reference frame could serve this purpose. The Australian "Standards & Practices for Control Surveys" (ICSM, 1994 -- see section 10.2.4) sets out some "test procedures" for GPS equipment, however these are just suggested practices and do not relate to "certification".

Back To Chapter 4 Contents / Next Topic / Previous Topic