Topocad 23

Known points

If we have carried out a forced adjustment (adjustment with known points locked) and had several observations designated as incorrect, this does not always need to be due to the error in the observations. It could instead be that the known points have incorrect positions. This could be due to them moving, that you have use the wrong error point, or that we have specified the wrong coordinates. All known points are calculated in the adjustment as perfect and any errors they may have are interpreted as observation errors instead.

In order to test the observations without any influence from coordinate errors, you should therefore carry out a free adjustment (all points treated as new) in order to remove all errors in the observations. This assumes that the net is linked in loops as far as possible traverses to connection points produce uncertain results for free adjustment.

If you have removed all the observation faults in the net, it simply remains to test the positions of the known points. You do this via the following steps:

• If you have selected Plane or Plane and height under Netadj.|Settings|General the known coordinates in plane are tested. If the selection is Height, the Z coordinates are tested instead.
• The test starts by selecting Tests|Known points. The following window appears:

1. Here we select the points we want to test in the list first Lock/release known points. The points that are pre-checked will be included in the test. If we click the Extents button, all points will be included. The None button releases all points allowing you to make your own selection. This gives us the option of testing known points in a certain part of the net, which can be useful in expansive nets.
2. The program can then be set to stop when a calculation has been made (Only release point with greatest error) or release the worst point and recalculate until all points meet the threshold (Release points until the net is approved). The latter is as quick and easy as an initial test, but the final check should preferably be carried out point by point where you make a thorough analysis before proceeding.
3. When the program calculates length observations, you can specify under Corrections if the lengths are to be corrected for Ellipsoid and Projection. If you select Use project settings, the corrections apply that have been set generally for the project. Settings can be checked under File|Settings|Project settings|Coordinate. If you select According to settings, the settings are used for each individual observation's corrections (the Projection and Ellipsoid columns) in the observations tab. Note that these selections apply regardless of what you have set as speed settings under Net adj.|Settings|Advanced.

In order to describe other settings, we go through what happens if you start the test by pressing Calculate:

• A free adjustment is carried out. For the points to be tested, the coordinates are picked that the points were given in the free adjustment. These are incorrect in that they originate from a free adjustment, but if this is correct the points will be right in relation to each other.
• The program then takes test points coordinates from the free adjustment and transforms them so they fit as well as possible with the known coordinates for the same points.
• This is done to test in plane by moving in X and Y, rotating and, if you have selected it in the program, scale changing. Do this by selecting Congruent or Helmert as Transformation. The latter type also adjusts the scale of the free net, which means that you remove the influence of the scale error at the length gauge. If you are sure that the scale of the lengths is correct, you should use Congruent, which retains the scale of the lengths. Otherwise there is a small risk of fitting errors at the points being partially interpreted as scale errors in the calculation instead.
• For heights, the transformation takes place via the program calculating the average values for both the known and the adjusted points. The mean value is then removed from known and adjusted coordinates making both averages zero (center of mass reduction).
• For heights, mean errors are also calculated for connection height fixes even though they are not part of the free adjustment. The program then looks up the nearest adjusted height and uses the mean error's law of error propagation for the connection observations and the nearest adjusted point to set a mean error for the height fix you have connected to. Naturally, this value does not have the same certainty as the height mean error that is included in the free adjustment. However, excluding them would mean that you would not get any connection height fixes at all in the test, which is often a major disadvantage as this measurement situation occurs quite often.
• In plane position only the known points that are included in the free adjustment, i.e. connection points are excluded from the test unless the observations are over-determined in relation to them. This is due to them being uncertain in relation to the other net, where at least two unchecked observations (angle and length) are used. However, it is normal in plane mode that the connection observations are over-determined to ensure the points are included in the free net. We also have situations when just one angle is measured in relation to a known point that is a backsight. In that case this point is impossible to test and is excluded from the test.
• If the known coordinates are correct (and also the observations in the free adjustment) the adjusted and known coordinates fit exactly with each other for a transformation. If any point is incorrect, this is noticeable by it having a fitting error between the free and known coordinates. The fitting error is reported as an error divided into X and Y as well as radial (total) errors. The problem now is where to draw the boundary line for when a point is incorrect and, in connection with this, take into consideration the error sources included in the calculation. These are primarily the mean errors of the points from the transformation and the free adjustment. A point that is at the edge of the net will be more uncertain in the transformation than one in the middle.
• In order to have a tool that is as certain as possible when identifying errors, a test quota is calculated. This specifies how large the fitting error is compared to the total mean errors of the point from the transformation and the free adjustment in the direction of the fitting error. This test value can be compared with standardized improvements (sigma levels) for observations. Following this, HMK's three level principle can be applied in order to assess if a point is wrong or not. You can set the program if the limit for errors is set at factor 2 (95% error probability), 3 (99.8%) or your own level.
• When the calculation is complete, the number of points is reported that are locked or released following the calculation. In the Current point box you can see the worse point's ID and test quota together with the error in X and Y, radial (total) and the direction (bearing) in which the point has moved.
• If you click Edit, the program jumps to the point tab and positions itself on the row of the current point. This is to enable you to quickly check and, if necessary, correct any wrong coordinates for the current point. If you click Next, the second worse point is displayed and so on. Previous then goes in the other direction.
• We can also tick the box if the point is to be known (Locked) or released in the next calculation.
• You get a summary of a calculation by clicking Report. You then select the report template you want to use (normally Standard) and then get a summary of the calculation. The report shows the following details first:

• The standard mean error is then displayed, HMK's approval limit, over-determinations and K-Value for the free adjustment that form the basis of the test. Following this the same parameters are shown for the forced adjustment with all points locked and finally a forced adjustment with only the remaining locked points as known. The idea here is that you can see if the deleted points improve the net as a whole at the last adjustment.
• The data is then displayed for the point(s) that have been released. The following data is displayed:

• The next part of the report is a record of each individual search and its results. If we have set the program to only make one calculation, it is shown here. If we have selected Release points until the net is approved all the separate calculations are reported. The following data is included:

• When you have finished analyzing the results, you can print or save the results file in various formats using the icons top left. To return to the test settings, close the results window and select OK, whereupon you return to the test's initial window. If points have been released during or after the latest calculation, they are now released in the list Lock/release known points. We can now choose to change the settings, release or lock points, and recalculate.
• When we have finished with the test, we press Apply. We are then asked if we want the points that have been released in the test to be released under the point tab as well. To give known points new coordinates could be delicate and you should be aware of the consequences. The danger is that you could easily have different coordinates for a certain point in different projects, so the points that are released should not be uncertain.