This page provides information to consider when determining reference benchmark heights in terms of a Local Vertical Datum (LVD).

The introduction of NZVD2016 and NZGeoid2016 provided tools that allow customers to better understand the local accuracy and network accuracy of heighted marks. They also allow for LVD heights to be computed anywhere in the country through the application a Vertical Datum Relationship (VDR) grid.

One of the potential consequences of a national vertical datum and geoid is that customers may find differences between heights which have been determined via different methods, ie published LVD heights may differ from heights computed using NZVD2016 + VDR

When deciding which height is ‘best’ for the project that you are working on, there are a number of factors to consider, such as:

  • What are the orders and survey date of the marks in question?
  • What is the accuracy required, and is relative or absolute accuracy of the heights more important?
  • Has this area been subject to deformation? (examples: landslides/slips, earthquakes, changing hydraulic conditions)
  • Is this work going to be connected with other past or future works? 
  • Is it possible to have all heights in the project determined from a consistent method?


Below are some items that you should consider:

Published NZVD2016 Heights

The NZVD2016 heights have been computed using the results of a National Geodetic Adjustment (NGA). All geodetic vectors, including GNSS, terrestrial observations and 1st order levelling, held by the LINZ have been adjusted in a single least squares adjustment. All marks in this adjustment have been rigorously tested and assigned orders based on the computation statistics. This means that each mark is tested against every other mark in the adjustment – whether or not there is an observation directly between them. This gives us confidence that if we level between two published NZVD2016 marks that the observations will close within tolerances. 

In addition, the NGA is constantly evolving as additional survey data is incorporated, allowing NZVD2016 heights update as vertical deformation is observed between measurements or as inaccuracies in previous measurements are identified.

Published LVD heights

LVD heights have come from many sources over the years. In general 1V and 2V heights have come from precise levelling, and 3V and 4V from less accurate methods such as trig heighting. Most of the precise levelling was completed in the 60s and 70s, and has not been maintained or repeated by LINZ. During this time these marks may have been subjected to measureable, but unknown amounts of vertical deformation.

The observation methods used to determine these LVD heights are very good at determining the uncertainties on lines which have been directly measured, but not so good at testing absolute accuracy or testing accuracy between nearby marks for which there are no direct observations.

For example, the relative accuracy component for 3V is 10cm per kilometre. As marks heighted using trig heights are often several kilometres away, it is not unreasonable to have +/- 20cm uncertainties on 3V LVD heights.

LVD Transformation Grids

The transformation grids were created by taking all the marks which had both 1V/2V LVD heights and orders 0-5 NZGD2000 heights and comparing the offsets. These offsets were then used to create transformation grids with a resolution of 2’ x 2’ arc minutes. The accuracy of the transformation grid is at the cm level. However when transforming heights it is important to consider that the uncertainties associated with order of the mark in question may be greater than this.

For example, applying a transformation grid to a 3V LVD mark which has uncertainties of +/- 10cm may result in discrepancies at the decimetre level with the published NZVD2016 height on the same mark.

Last Updated: 8 May 2018