１５．一般的な水路測量（General Hydrography - GPS）
Q. RTK GPSの特別な使用方法(Do Not Fudge RTK GPS)[英語]
A few years ago, in Hypack conferences, we were discussing about "fudged" differential systems. What is a "fudged" system? To get rid of the annoying and possibly inaccurate datum transformation, you enter local datum coordinates in the base station of a differential GPS system and you expect the rover GPS to output coordinates directly in the local datum. All GPS experts consulted at the time seemed to agree that such a technique is usable in a fairly large radius around the base station (over 100km if I remember correctly).
At the time it was a very useful technique for many of our users as datum transformation parameters were not readily available and Hypack didn't provide any support for finding even approximated parameters. Times have changed and now a user within continental US can calculate those parameters directly in the "Geodetic Parameters" program (using the CORPSCON button) and the rest can try to find an approximated value using the "Datum Transformation" program so the "fudged" systems have become somewhat obsolete.
Recently however I've seen someone trying to use a "fudged" RTK system and not with very good results: the rover GPS would drop out of RTK mode and it not come back even at very short distance from the base station. To see if it is safe to use "fudged" RTK systems let us examine closer how an RTK system works.
An RTK system achieves its incredible accuracy by measuring the "carrier phase" - this is the phase of the 1.52 GHz signal sent by satellites. The wavelength of this signal is approximately 19cm and its phase can be measured with a precision of about 1%. The only problem is that this signal repeats itself, well every 19 cm, so to find out your position the GPS receiver needs to calculate the number of integer cycles of 19 cm that fit along the path between itself and the satellite. This process is called "solving the integer ambiguities" and it is not an easy job (if it would have been, RTK receivers would have been available long ago and they would be much cheaper). Help comes from different sources:
- The pseudorange solution (the classical way of calculating a GPS position) gives an approximate position of the receiver thus narrowing the range of integer cycles that yield a possible solution.
- Knowledge about satellite's movement - a nice smooth ellipse - also helps if we are to sit longer in the same position (roughly speaking, if the receiver picks the wrong solution, the satellite doesn't follow it's predicted path so the receiver tries a different solution).
Assuming a receiver has found a good solution (the RTK loop is locked in GPS lingo) it still needs to differentiate between it's own movement and the satellite's movement and this is where the RTK base station comes into play: it transmits it's own carrier phase measurements so that the roving receiver can figure out how much of the change is due to it's own movement and how much is due to satellite movement. This is one of the main differences between differential and RTK GPS: the classic differential solution is a "discrete" solution that is calculated completely at every cycle with little information about the previous state of the receiver, while RTK solution is a "continuous" one in which information about the previous position is built into every new one.
What happens now when you "fudge" the system by entering coordinates in another datum than WGS-84? All of a sudden satellites don't follow their predicted path so your GPS receiver is going to be confused. More often than not it will not be able to properly solve the integer ambiguities and it will drop to a standard differential solution.
So our recommendation based on the little knowledge and experience we have with GPS systems is DO NOT TRY TO FUDGE AN RTK SYSTEM! Go out and find the correct datum transformation parameters, set your base station to correct WGS-84 coordinates and Hypack will obligingly calculate correct coordinates on your local datum.
I tried to get more information and opinions by posting this question on USENET discussion groups. Initial reactions seem to validate our opinion. If we are at it, you should try to visit sci.geo.satellite-nav and sci.engr.surveying. There you will find a wealth of information about GPS and surveying as well as informed people willing to answer your questions.