As promised in my post "Problem with new GPS Satellite SVN-49/PRN-01"here some details regarding what we observe when we analyse the data of this satellite.

Below are two plots. The first one showing the pseudo range residuals of a "normal" GPS satellites. We randomly picked satellite SVN-38/PRN08 for this. The residuals are based on all the observations taken by the stations observing this satellite. The data used was from day 100 in 2009, or in normal date April 10, 2009. The residuals are plotted as a function of the elevation of the satellite above the local horizon of the observing station. That means that at 90 degrees elevation the satellite stands right above the station whereas at 10 degrees elevation the satellite is very low on the horizon. In this figure one can clearly see the increase of the noise of the observations at low elevations which is a well known phenomenon mainly caused by so called "multipath" effects.


The second figure shows the same picture but for our the new GPS satellite SVN-49/PRN01. The behavior of the residuals shows a clear signature which is obviously elevation dependent. Interestingly enought the carrier phase observation residuals do not show such a signature. This indicates that the problem is in the pseudo range observations, e.g., in the so called group delay.


However, the residuals show that the problem is only at the few meter level. But somehow the GPS operators see much larger problems. How and where the see those is still unclear to me. However, to my understanding it is possible to measure the differences between the observations on board of the satellite. Possilbe the >100 meter effects are observed in such measurments. However, for normal users as us there is only a few meter problem which we could live with, more or less. The really big problem comes from the fact that the GPS system operators are trying to solve this problem by changing the broadcast ephemerides. This becomes clear if we compare an orbit estimated by ourselves with the orbit information broadcasted by the GPS satellites. Note that our estimated orbits have a precission of a few centimeters whereas the broadcast ephemeris is normally at the few meter level. The figure below shows such a orbit comparison of the GPS satellites with PRN 1, 2, 3, 4, and 5. The differences for the "normal" GPS satellites is at the few meter level. For PRN 1 the differences amount up to 60 kilometers!!! This of course makes the satellite completely useless for any processing.


All in all the problems with this satellite do not seem to be very severe except for the broadcast ephemerides which are truely horrid!

So whatever you do with GPS do not use the data of the new SVN-49/PRN01 unless you know what you are doing!

(ps. I appologize for the poor quality of the plots. Have to find a better tool to do these plots...)

Labels: GPS, L5, SVN-49 anomaly

It looks as if the newest GPS satellite has a significant problem. This is the GPS satellite which also broadcast signals on the new L5 frequency to ensure that frequencies for the GPS system. The L5 signals were planned to be available from the new GPS Block II-F satellites but due to delays in building these satellites it became necessary as kind of an "emergency rescue" of the L5 frequency band to launch launch some other satellite to use the frequency. The "perfect" solution was found to add the L5 capability to one of the Block II-R satellites. So this was done for the SVN-49/PRN-01 satellite that was launched on March 29, 2009.

However, it seems that something has gone wrong with the signals of this satellite. In my work where we try to get the highest precission out of the GPS system, meaning orbits at the ~20mm level and receiver positions at the ~1mm level, we can see a clear pattern in the pseudo range residuals of this satellite. Now since we typically use the carrier phase observations and not the code this is not really a problem for our type of work. However, the few meter (!!!!) pseudo range residuals are disturbing and they show a clear dependency on elevation.

However, what is more disturbing is that on the GPS system side they seem to be getting even larger effects then the few meters we see in our work. So far this is something I have not yet been able to understand. It seems the GPS system operators see problems with this satellite at the 100 to 200 meter level!!! A true catastrophy! As work around for this problem they GPS system is applying a huge antenna offset when broadcasting the ephemerides and clock corrections of this satellite. These problems also explain why this satellite is still not declared healthy. In fact there is some speculation that because of the observed problems it may never be set healthy.

So in summary the facts are:

1. There is a significant problem difference the code and the phase observations
2. The phase observations seem to be fine as we can do integer ambiguity resolution without too much problems (but of course one can not use the code observations for aiding the ambiguity resolution)
3. The observed differences point in the direction of a significant "group delay". This is a delay which effects the code observations but not the phase observations
4. The GPS system is observing even larger problems (>100m)!? We do not see those problems however!?

In any case it may be that this GPS satellite will never be set healthy making the L5 experiment a rather costly one...

As soon as I find out more details I will let you know!

Cheers,
Springinhetveld

Labels: GPS, L5, SVN-49 anomaly