Betty,
Thanks for starting this discussion. It sounds like you’ve made a good start by updating your lines to the original record data. Having the adjustment fail when first running it is fairly normal/expected; the reported results can be used for QA and to better understand your data, and to choose a Distance Residual tolerance, as described below. Of course you should not expect to see things in the report get worse after making changes to the parcel data. I’d like to help you figure that out, and I may contact you to get your data, but in the meantime here are some answers to your questions.
With fabric adjustment you’re making a best fit of all the record data by matching it to the control points. The adjustment will create a balance between the control point positions of a few good known locations in a network, and the COGO record values on all of the lines used in that adjustment. If there is a significant mismatch in making this balance, it could be due to one or more of the control points, or one or more of the record distances or bearings. The control point may be out of sync if it is not referenced to the correct fabric point. I’ll go through the information that is important for the adjustment, and the attributes that affect the adjustment versus attributes that do not directly impact the adjustment.
The following attributes impact the adjustment directly: Accuracy (on plan, parcel and lines tables), Bearing, Distance, Radius, and Category, (on lines) the X Y Z coordinate attributes (on control), and true-mid-bearing flag (on plans). All the other attributes are computed information, or are non-spatial cadastral information. The line category attributes are related to parcel structure (example boundary lines, connection lines etc.); the adjustment will report parcels without the correct structure, but it is not very common to run into this problem.
For the plan table, the accuracy value is required, and it has a direct impact on fabric adjustment results. Aside from true-mid-bearing setting, it’s the only attribute for Plans that can be changed to affect adjustment results. You should set that accuracy to the level that matches the survey date of the subdivision/plat/deed etc. (see doc). Similarly, for parcels, the accuracy category is the sole attribute that will affect adjustment, however the attribute is null by default and the accuracy is inherited from the plan when running an adjustment. You will most likely not need to manually add an accuracy value to the parcel, unless you want to explicitly over-ride the value inherited from the Plan (this should be rare). Also for lines, accuracy category is null by default and is inherited from the Parcel or Plan, but it can be set directly for a line in order to over-ride the value it’d otherwise inherit. (Setting accuracy on a line is a bit less rare, see tips below on Accuracy values.)
Tips on using Accuracy category values (1 through 7)
Setting lines to accuracy 7 is like removing the line completely, so while this is OK for some lines, for others it may be a problem as it could weaken the network, or break network connectivity. In general, setting accuracy should be done at the Plan level, and should be set based on the date of survey, as defined in the accuracy table description (see doc: About accuracy—Help | ArcGIS for Desktop )
You typically should not have to change accuracy on individual parcels or lines. However, if you’d like to reduce the influence of a particular line to the lowest possible without breaking connectivity, then use accuracy category 6 instead.
More about accuracy category 7: There are some valid cases for setting accuracy 7, such as for long lines that you did not directly COGO in, such as radial lines that are not part of a cul-de-sac, or part connector lines that are (typically) system generated, but as mentioned, be aware that this can break connectivity.
Tips on control points
Add elevation data to the control points. Make sure that the control points that you are using have elevation values (z values) populated. They do not have to be very accurate elevations, so if you have control points that do not have a Z, then you can add the standard topographic base map to your map, and use the elevation from the nearest topographic layer contour, nearest spot height, or bench-mark. Note that the units of the elevation may have to be converted when using this technique. This is important because the record distances are presumed to be ground values, and the adjustment will use the elevation information on the control points to compute the combined scale factor. Without this factor applied, you are more likely to have lines that are detected as outliers, even though they are actually good record values. The distribution of the control points should not be such that the control is bunched together in one area of the network. Rather they should be well distributed and, as a very approximate rule of thumb, ratios of control points to parcels can range between 1:20 to 1:200.
Tips on overlapping parcels
When you have Lots and Units, and the Tax Parcels are a direct duplicate of these Lots and Units, and share the same points, then it is not necessary to select the lots as well as the tax parcels when running the adjustment.
This just increases the amount of data that the adjustment processes. Although it is important for there to be lots of redundant information in the adjustment, it is not meaningful if it is just duplicated overlapping lines, with the same record values. Since the fabric points are shared between the lots and tax parcels, all parcels connected to those points will still be adjusted even though they may not have been directly participating in the adjustment. So for these cases you can choose to minimize what the adjustment processes by turning off selection for the layers that have the duplicate parcels, and only selecting the Lots, for example.
Tips on Connectivity and line categories
Sometimes you can have data where there are fabric points very close together and that are not connected with a line. These points should be averaged/merged using the mean point tools, or else the parcel should be re-joined so that this connectivity is created.
Tips on reading the report
After running the adjustment, it’s useful to temporarily save the report file, and refer to it when going back into the data to check on the reported items, and make changes as needed. As mentioned before, the adjustment report is a good way to QA your data, and find possible data entry problems, or connectivity (topological) problems. The double hash “##” next to a value in the report indicates that the record value for that line does not match well with the other measurements in that part of the network. This means that there may be a possible mistake in the data at this location, such as a mistyped bearing or distance, or a transposed number on the actual record, and so on.
If you see the “##” next to bearing values, then it is likely that it is a radial line, a connection line, origin connection line, or a part connector line that has a bearing that does not have a correct bearing with respect to the parcel it’s attached to. This can happen when there is a rotation on a parcel.
Tips on the adjustment tolerance
The Distance Residual tolerance is the value that is most typically changed to get an adjustment to complete. The default value is 0.348 feet (0.1meters.) If you see a ‘##’ next to a distance value, look at the residual value, and divide by 3. This is a rule of thumb for determining an initial value to use. However, if you find you are making this value very high (more than 25 feet) then there are likely other problems that need to be resolved.
There are a few things that may need to be expanded on, and this should hopefully keep the discussion going.
-Tim