e-survey Forum

We are having issues with Reg 3.1(b)when adopting work off old plans with less than ideal closes. In particular where an old traverse (eg an old centre-line) runs close to a more modern traverse with infrequent connections between the two. There is a very high likelihood that the short vectors between the two series of marks will fail the regulation 3 tests. This may only be an error of 0.04 but at present LINZ are requisitioning it.

Their response is of more concern. In the Landonline FAQ they advise surveyors to show the old traverse on a calc sheet and re-calculate the boundary vectors so that they fit and capture those. In earlier days we used to call this 'applying Cooks Constant" and was considered fudging.

This suggestion concerns me greatly as I worry about the long term effect of re-calculating boundaries to make them fit least square adjustments simply to meet some arbitrary limit.

It was a major backwards step when the allowance for old survey tolerances was thrown out. The result is a re-calculation and a loss of integrity of data for the sake of mathematical theory.

This approach also prevents us from capturing old traverse to beef up the LoL topology. In one area where we are working the miscloses are often in the 1 to 2 metre range (rf ~1:1000). This is a remote location and the LINZ topology is lucky to be within 20 metres of the correct position.

This is a good example of definition being driven by technology rather than the other way around.

It was a mistake to drop the old allowance for old adoptions and this should be rectified.

These concerns require separate responses for boundaries and for non-boundaries.

Firstly, the comment about the need for the accuracy of boundaries to meet some arbitrary limit is unfounded. The accuracy of boundaries is a fundamental requirement of the cadastral system. An old boundary that does not meet the specified accuracy standards is simply ‘wrong’ and should not be allowed to persist. Landowners and surveyors need to have confidence in these boundaries and their dimensions. It should be noted that the Rules have set a variety of standards at fairly forgiving tolerances, as well as the ‘Accept’ provisions, to generally match the risks in different areas. Enhancing the accuracy of a boundary does not produce a ‘loss of integrity of data’ or ‘fudge’ the boundary, but rather provides that integrity where it was previously lacking.

However the concerns about the equivalent requirement for adopted non-boundary points [part of rule 3.1(b)] may well be justified. We are now reconsidering this requirement and will let you know shortly what we find.

Notwithstanding that reconsideration of the rule, the capture and testing of poor quality adopted traverse data is currently problematic. We recognise that there are currently some issues around the accuracy tests that Landonline applies and how these are actioned through requisitions. We have started developing improvements to those tests and the procedures around them. However these are technically complex and would need to take account of any changes we make relating to rule 3.1(b), so will take a little time to develop. We are also working on some short term responses including the possibility of not requisitioning.

In the meantime there are some steps that surveyors can take to address these issues:

(1) avoid unnecessary capture of poor quality adopted traverse data. Normally such data provides more reliable evidence to assist in determining the location of an old boundary, such as in the case of an old road centreline traverse. Having made such determinations which result in corrections to original dimensions that exceed the accuracy standards, the recording of that original data in Landonline does little to enhance the accuracy of the cadastre and would be of little use for future surveyors. Surveyors should continue to calculate these boundaries as they have done – however the adopted data that does not meet the standard does not need to be captured. This would avoid many of the current validation test failures. The recalculated information could be provided on attached calculation sheets. Note that there is no requirement in the Rules to recalculate poor quality traverse vectors (and this is not recommended).

(2) Consider whether additional vectors between key points can be captured. This is not always possible, as these vectors should be either adopted or calculated from reliable data, noting that the surveyor is taking responsibility for these new vectors.

Clearly there is room for discussion on this topic. I am not sure that many read this forum so this might not be the best place to do it.
However to add my thoughts to Anslem's comment about boundaries that do not meet the required standard being wrong, I beg to disagree. The purpose of the surveying is to determine the location of title boundaries such that the title can be supported and there is no overlap or conflict with adjoining titles. In the case of a boundary between two titles not meeting the 'required' accuracy standard does not mean that the mathematical definition is wrong, simply that its position is less certain. In some circumstances that level of uncertainty my be fine (eg high country runs) in others it may not.
I use the term 'arbitrary' simply because our concept of what is sufficiently accurate seems to change with technology rather than with the clients requirements.

Some thoughts while I wait for Plan gen ...
Having taken on capturing a heap of remote rural traverse and some survey fixes to support an heap of easements I have had a good play with Rule 3.1 - or more particularly LoL's Business House Rules for testing Rule 3.1
It is my firm view that much of the problem is in the testing. My adopted data has generally quite good bowditch closes (in 12D) and looks good. Initially I got 12% failure of 3.1(a) (new work) and 21 vectors failed 3.1(b) The 3.1(a) test was the most interesting because the survey data was pretty solid (GPS closes between trig of 1:50-60k and nothing worse than 1:10K for shorter edm). All the outliers disappeared with the removal of a redundant GPS vector and the addition of a calculated vector (missing line calc) to brace a loop. Note that no data was changed to achieve this. What did this tell me? There was nothing wrong with the data but the method of testing it was at fault.
I was taught that least square networks are not good at testing long narrow chains of measurements.
The 3.1(b) stuff was pretty similar; most of the failures were in two or three areas. The first two were fixed by capturing more calculated vectors in the area that failed. The third area I discovered (after a bit of playing around) was because it was a section of older traverse that I had flagged as 'Old Adopted'. That was a mistake. What happened, I believe, was that the adjustment put more correction into that area which resulted in many vectors failing. Flagging them the same as all the others (Theo/EDM) spread the errors evenly (same weight) and the problem went away.
Passing the LoL tests seems to have nothing to do with how good your data is but how you lay out your traverse or which extra vectors you choose to calculate.
The weighting seems to work in reverse. Reducing the weighting of old work is the right thing to do but the tolerances need to be similarly enlarged. In other words we need to have tolerances related to the age of the adopted work - as in the 2002 reg's.
Okay Plan gen is ready - on with it.

On August 13th 2010 the SG issued an exemption from Rule 3.1(b) which eliminates most of the problems we have been having. I expect that this is a temporary measure while a more long term solution is found.
See http://www.linz.govt.nz/about-linz/news-publications-and-consultations/search-for-regulatory-documents/DocumentSummary.aspx?document=295

We have issued a Ruling that deals with most of this issue. See http://www.linz.govt.nz/survey-titles/cadastral-surveying/publications/adopted-traverses/index.aspx
Changes will be required to Landonline to implement this Ruling. Other enhancements to the accuracy validation tests are also being considered.