I appreciate your help.

Kind Regards Pavlo.

]]>1) All individual epoch were recalculated and all failed observation removed from deformation data set.

Okay.

Point 90P is outside a wall and located on nearby building. So I updated setup.

2) Point 90P is no longer a wall point but with point 101 a reference point. (datum group)

The wall points are the points to analyse and the stations are non-homologous points. Thus, your reference point field is defined by two points i.e., 101 and 90P. This is - more or less - like a minimal configuration. As pointed out in my last posting: What is happens, if 101 or 90P is unstable and shifted? The current configuration is somewhat *weak* (I don't know a better English term).

An example: In a leveling network, the number of datum points should be three or more. If one of the datum points is unstable (while the other points are stable), it is possible to *identify* this unstable point. If the network has only two datum points, it is possible to detect the discrepancy between both points but you cannot identify the unstable point itself.

3) Stations are new points. (new group)

Okay.

4) Wall points are datum points, together with point 101 and 90P.

The wall points are the points to analyse, thus, normally, I would define these points as new points.

If this a reasonable thing to do ? Failing component is sigma_a but sigma_a is defined by the theodolite manufacturer, that why I increase sigma_c is this reasonable?

Yes, it is. The uncertainty sigma_a and sigma_c are not independent in variance-component estimation. If you change sigma_c, the value for sigma_a will be changed in variance-component estimation, too.

2) Can this configuration be considered as "correct" configuration for deformation test ?

I would define the wall points as new points. (And I would try to increase the number of points, which defines the datum for the upcoming epochs)

With this new setup is this still true ? Should point 90P be excluded from datum ?

In my configuration (datum: 101 and 90P), I cannot exclude one of the points but JAG3D detects a (small) discrepancy between both points. The test statistic for `Tpost`

is exceeded (the critical value from F-distribution is `4.2`

) and the estimated point shift is `> 2 mm`

.

reagrds

Micha

I updated my project as follow.

1) All individual epoch were recalculated and all failed observation removed from deformation data set.

Is it impossible to add more reference points to define the datum outside the wall?

Point 90P is outside a wall and located on nearby building. So I updated setup.

2) Point 90P is no longer a wall point but with point 101 a reference point. (datum group)

3) Stations are new points. (new group)

4) Wall points are datum points, together with point 101 and 90P.

Question.

1) During individual epoch run I had failed zenith angle for epoch02 zenith angle observation group. To fix this I increased centering error from 0.5 mm to 1 mm.

If this a reasonable thing to do ? Failing component is sigma_a but sigma_a is defined by the theodolite manufacturer, that why I increase sigma_c is this reasonable?

I included an updated project with screenshot of the network configuration.

2) Can this configuration be considered as "correct" configuration for deformation test ?

3)

Normally, I would now exclude unstable points from the datum starting with point 90P.

With this new setup is this still true ? Should point 90P be excluded from datum ?

Kind regards.

1) In my job setup I have only one point fixed.

In most cases, this is not a solid initial situation because your detected deformations dependents on the position (*and stability*) of the single point. What is happens, if this point is shifted?

2) I used this fixed point and station points which I calculated from individual epoch solutions as stochastic points.

In classical deformation analysis, the network is adjusted as free network adjustment.

a) First solve individual epoch and derive coordinates for stations (new points), use wall and one fixed points as datum points.

We adjust the single epochs as free network. Here, only datum- and new points are valid. The single epoch adjustment is used to check the stochastic model and detect outliers in observations. In general, we remove outliers from the data sets.

b) Use derived coordinates for station and one fixed point as stochastic in deformation analysis.

As noted, we use a free network adjustment (only datum- and new points).

In your case, it is hard to make a suggestion because you haven't a reference point field (expect point 101). Of course, you can define all wall points as datum. This will work until the complete wall is shifted. Is it impossible to add more reference points to define the datum outside the wall?

For your information: This is my pre-result. Normally, I would now exclude instabile points from the datum starting with point 90P.

regards

Micha

I manage to resolve problem but

want to confirm with you if my thinking is correct:

1) In my job setup I have only one point fixed.

2) I used this fixed point and station points which I calculated from individual epoch solutions

as stochastic points.

3) As you suggested I set my (wall) points as new points.

4) Solution converges.

Question:

1) Is my assumption and work flow is correct:

a) First solve individual epoch and derive coordinates for stations (new points), use wall and one fixed points as datum points.

b) Use derived coordinates for station and one fixed point as stochastic in deformation analysis.

I attached a data set which I used in my adjustment.

Data set.

Kind regards.