Bill or anybody else,

Just wondering what your opinion is on using multiple base station during a kinematic survey.

After reading the thread on the main board it seems that except for extending range to avoid a logistical "timeout" there is no reason to use two bases during a K session.

Any thoughts ?

Jimbo
Modified By James Webb on 6/9/2001 at 7:44 AM

James, with Locus we use 2 or 3 base stations when doing K, although I've only used the K option a dozen or so times.

When doing K you have to have 4 or 5 COMMON satellites between units for the vector to process. Just because the unit is recording 6 or 7 sats doesn't mean you are getting common sats.

With 2 base stations you increase the probability that you will have at least one base station recording the same sats as the rover.

When processing, I've seen several vectors which processed fine from 1 base station, but failed from the other. So I think using 2 base stations for Locus K has saved me a trip back to the site a few times.

In addition, it creates an adjustable network for all points that you located. Even though the "trivial vector" comes into play on that idea, I still like to see the adjustment results and look at the residuals.

Just my thoughts ...

-Greg

Greg,

Excellent point on the COMMON sats. I guess the guys on the main board or so used to using the $40k stuff with 12+ channels the obvious is sometimes missed !

But then I hadn't thought of that reasoning myself, but I always use two bases during the few K sessions we do.

Jimbo

Greg,

Using two base receivers in the kinematic mode is what I like to do also. The "trivial vector" thing is put to rest by excluding the base-base vector of the session in the adjustment. Use another vector from a different session for the base-base vector.
Maybe Mr. Geodosist is absolutely correct in his opinion of the two base stations being overkill or not necessary. But I think that 2 independent vectors with 10 or 15 seconds worth of data, based on a previously established static network where I have two points with good redundency is the way to fly.
I think I am getting a lot more "bang for the buck" than any RTK system when you go this route.
But thats just my opinion. Yours could be wrong!

Jimbo,

The important point you must remember when using multiple base stations in a kinematic survey is you are not getting independent multiple vectors into your rover points. Why? Because each vector to each rover point is computed using the same sample of rover data. This makes all the vectors into each point dependent or highly correlated by this common data. An undetected problem with the rover data at a point COULD find it's way into each vector going into this point. So multiple base stations does not provide you with independent checks for your rover points. BUT, it is my opinion that muliple base stations do have value. The multiple vectors into each point my be highly correlated but they are not useless. They do provide some level of redundancy, albeit not as good as truly independent observations. You will find situations were the solution from one base station is poor yet from another is good. Reasons for this include:

1. One base station may be seeing a slightly different constellation of satellites compared to the other. This is primarily true in situations were your base points have some amount of obstructions. In this case, different data is being used to compute the base to rover vectors.

2. GPS accuracy is distance dependent. The closer you are to a base station, the more accurate the solution and the less likely you are to be fooled by a bad solution looking good (i.e. probabilities of incorrectly fixing ambiguities go down the shorter the vector). With multiple base stations, you can position them to limit the distance between the rover and a base.

So, by having multiple base stations around the project site, you are adding value with these multiple vectors to your rover points, even though they are not independent solutions.

Hope this helps.

Bill Martin
Ashtech Precision Products.

Bill,

Your post answers my questions.

JImbo

Ok, I stand corrected. I see Bills' point on this. I was thinking multiple sessions in static mode.
But I still feel a bit more comfortable with having 2 known points processing with the unknown kinematic "roved" points.
Like 2 total stations and a 360 degree prism, perhaps? ;0