Best practices for running a fabric least-squares adjustment

This topic applies to ArcEditor and ArcInfo only.

When running an adjustment, it is good practice to examine the adjustment report to ensure that there are no blunders and that the control coordinates are correct. Examining the close point and line point errors will help to reveal data inaccuracies and connectivity problems in the fabric network. Close point errors indicate that there are some points that most likely should be merged into single points. Line point errors may indicate data inaccuracies as the line points have become offset from their lines at a distance greater than the specified tolerance. Close point and line point errors should be repaired before any further adjustments are carried out.

Another useful way to assess fabric quality or the validity of control points is to perform the adjustment with a few control points left inactive. If the corresponding fabric points of the inactive control points adjust to the locations of the inactive control points within the expected tolerances, the adjustment is performing well. Large discrepancies at the inactive control points is cause for concern: either the control coordinates are inaccurate, some parcel dimensions are suspect, or there is insufficient connectivity in the network geometry. If some parcel lines are the culprit, they will usually appear in the adjustment report. Work should not proceed until the cause of the problem is uncovered and remedied. Often fabric data or control problems reveal themselves as unexpectedly large errors in most adjustment reports.

Once problems are resolved, the adjustment should converge and give useful information about the real quality of the fabric dimensions. Least-squares adjustment convergence occurs when coordinate shifts become zero or do not change after each successive adjustment iteration (running and accepting the adjustment repeatedly).

Maximum and average coordinate shift after adjustment
Maximum and average coordinate shift after adjustment

A satisfactory adjustment not only yields parcel geometry commensurate with the real precision of the survey dimensions but also the most probable coordinates for the parcel corners.

By invoking least-squares adjustments at an early stage and running them often, you can be alerted to any problematic data as soon as it enters the fabric.


A redundant parcel network allows the least-squares adjustment to determine the best solution as well as flag those lines that may be statistically suspect. On the Least Squares Adjustment Summary dialog box under Adjustment Statistical Summary, the value for Redundancy should be greater than the value for Number of Unknowns.

Adjustment Statistical Summary

Area of adjustment

A fabric least-squares adjustment is run on a group of parcels selected in the map or on the parcels in a fabric job. When running a least-squares adjustment, the best results are obtained when the adjustment area is a well-balanced geometric shape with redundant measurements and evenly distributed control. Long, narrow areas without adequate control and areas with minimal redundancy (connectivity) can give poor results. These problems can be solved with more strategically located control and tighter parcel networks with higher degrees of connectivity. As more survey data and control are added to a poorly conditioned parcel fabric, readjustment will improve the accuracy and stability of the parcel fabric over time.

Entering new parcels into the parcel fabric

In general, when entering new parcels into the parcel fabric, it is recommended that you perform a least-squares adjustment after 20 or 30 parcels have been completed. Misclosures in parcels cause the shape of the fabric to be determined by the order in which the parcels are assembled. This is evidenced by the fact that as more parcels are joined to the fabric, the residuals during the joining phase begin to grow larger. By running an adjustment, these errors are distributed and new parcels will fit more closely with the adjusted fabric. Moreover, running adjustments fairly frequently during the parcel assembly process reduces the number of iterations needed for the adjustment to converge to an optimal solution.

Running least-squares adjustments when starting out with poor quality or unreliable data

It is not recommended that you run a least-squares adjustment on data with dimensions that do not match the plan or record of survey. If dimensions do not match the record of survey or plan, it is impossible to tell how inaccurate or accurate the dimensions really are. A least-squares adjustment could be run to preliminarily examine the network and identify those parcels with lines that do not fit in a best fit solution of the network. The results of this least-squares adjustment should not be applied (that is, the Accept button on the Least Squares Adjustment Summary dialog box should not be clicked).

When starting out with poor quality or unreliable data, it is best to first enter new parcel data into the parcel fabric before running and applying your first least-squares adjustment. If you have one or two large, new subdivisions entered with dimensions that match the plan, a least-squares adjustment can be run on these subdivisions and surrounding data. In this way, you can apply an appropriate accuracy level (an accuracy level that makes sense) to your newly entered subdivisions and apply a low accuracy level, for example, accuracy level 6, to parcels surrounding the subdivisions. The least-squares adjustment then has reliable survey dimensions with corresponding accuracies to work with and has a benchmark of good data versus the surrounding unreliable data. The reliable subdivision data will have a greater influence on the outcome of the adjustment results than the unreliable data, and more probable, realistic coordinates will be generated.

Because the adjustment now has a benchmark of reliable survey data to work with, it is now able to better identify which parcel lines in your migrated data are accurate, that is, which parcel lines fit well into the solution and which do not. If you supply no benchmark of good data to your least-squares adjustment, the results of the adjustment will be unreliable.

Thus the best approach is to adjust your parcel fabric in sections as new parcel data is entered. Over time, with the newly entered parcel data, the results of your adjustments will become more reliable and accurate and your network will stabilize.

Related Topics