Many organisations use land parcels in their Geographic Information Systems and Web Mapping displays. On the their own, they aren't particularly useful, but they do make for a great backdrop for contextual purposes. Everybody knows where they live, and the extent of ownership is a good surrogate for location in the absence of other data. It is also the basis of lots of other data that is captured relative to it, for example planning zones or buildings.

Land parcels are a data set that is rather taken for granted. It's just there. It's always been there - well, ever since the first spatial data was captured. Surveyors create them for us and LINZ makes them freely available. So far, so good.

Photo Credit: Mike-Campbell via Compfight cc

SEISMIC COMPLICATIONS

The trouble is, we've never gotten our heads around the fact that in a tectonically active country, land parcels are constantly on the move, either gradually (a few millimetres per year), or suddenly, such as in a magnitude 7 shunt.

I guess we always knew there was a problem, but imagined it would somehow be at least 10 years away. At 50mm per year, it would take 10 years to accumulate a 0.5m shift, which was the approximate accuracy of a GPS receiver 10 years ago. In the meantime, the accuracy of GPS receivers has increased ten-fold, there are more satellites enabling greater precision, and a huge proliferation of smart phones and tablets to capture data.

This data is being captured, of course, where it occurs relative to the GPS satellites, not where the land parcel was at the time it was surveyed. In addition, the annual rate of land deformation is now close to GPS receiver accuracy.

New Zealand is a long diagonal country halfway between the equator and the South Pole, making it a difficult country to map using conventional projections. NZMG, implemented in 1973, was a great response to this challenge, but by 2000 it was becoming increasingly obvious that it had served its useful purpose and needed to be replaced. NZGD2000 was created in response to the needs of changing technology.

HOW NZGD2000 WORKS

At this year's ALGIM GIS Conference in Auckland, Dr Chris Crooks from LINZ gave great insight into the detail and thinking behind the NZGD2000 coordinate system, and in particular, how the effects of the Canterbury earthquakes have beeb subsequently incorporated.

In selecting NZGD2000, LINZ opted for a semi-static coordinate system that places NZ at a location for the year 2000 with a velocity model included. The theory was that any data collected could be shifted back to where it would have been in 2000 by shifting it in the direction of the annual correction, multiplied by the number of years since 2000.

By taking this approach, LINZ had effectively implemented an untested system, and brought several problems along for the ride.

• The commercial GIS tools weren't (and still aren't) equipped to handle the solution

• The parcels data model didn't support the management of the data using this approach

• There was no plan or process in place to manage the data

LINZ had committed to periodically releasing updated velocity models, but not one release had occurred, signalling that the above issues continued to be a problem.

Additionally, the solution was based on gradual shifts of a uniform nature in a predictable direction, and could not account for the chaotic changes created by the September 2010 earthquake. The 2009 7.8 Fiordland earthquake also caused significant deformation throughout Southland, but this had been tolerated due to the largely rural nature of the area.

Christchurch, being urban and in need of immediate redevelopment, created an urgent need for a solution to be found.

POST-QUAKE UPDATES

LINZ recently adjusted all those land parcels that had been deformed by more than 5 mm. Presumably they were able to exclude the normal drift that applies to the rest of the country, as this is a stepwise correction related to seismic events, rather than a global correction. The net effect is that the land parcels have changed to where they would have been, had the earthquakes happened prior to 2000.

LINZ still hasn't released an adjustment nationally for gradual shift, but the fact that they have successfully dealt with the Canterbury issue, signals that they are prepared to deal with this issue as well. Chris reported that they now have much better data on what the actual velocity is around the country, whereas when the system was implemented, it was much more of a guess.

HOW DOES THIS AFFECT YOU? WHAT ISSUES SHOULD YOU CONSIDER?

If your data is coincident or relative to data in the South Island, then you need to think about adjusting it soon. You'll also need to think about metadata.

Have you recorded whether your datasets are subject to tectonic shift at a dataset level? You'll also need to think about whether you should be recording tectonic shift at a record level - has this record been adjusted? Which velocity model was it adjusted to? And if you are not in the South Island, you'll need to deal with this issue eventually - do you have a plan to do so?

With LINZ embarking on upgrades to existing or new data sets, including Addresses, ASaTs (Automated Survey and Titles system), the Property Data Management Framework, and 'scaleless' topographic data, it is a timely reminder that you should be providing input and feedback that ensures that sufficient record level metadata exists to effectively manage the sharing of data with a temporal tectonic component.

It's also a reminder that such systems should be tested end-to-end for integration before being put into production. As users of the data you can offer to be involved in the testing.