[Aust-NZ] NZ Geo Placenames

[Peter]

Thanks to those putting me on the right path. Having completed a preliminary tracing of the urban areas (openlayers plugin thanks) to use as a check against better data, Ive ended up splitting this task into 2, deal with the towns first, and leave unpopluated places to a second task altogether.

In the absence of any other metric, SNZ defines 'urban areas' as follows:
   
   city         >50K
   main UA      >30K
   secondary UA >10K
   minor UA     >1K
   rural center <1K

I dont have any gripes with that, and my look at SNZs own "experimental" algorithim to determine urban areas from meshblocks (MBs), shows they err on the side of scooping up quite a bit of rural landscape around each UA. It is quite interesting though studying the way urbanism changes over time. Questions like is South Bay an outlying suburb of Kaikoura or a town of its own, or Taipa + Cable Bay + Mangonui, becoming over time one 'place' (Taipa-Mangonui Community according to SNZ).

Anyway the SNZ concordance mapping for UA<-->MB only includes the larger UAs, main and secondary i think, hence id have to build that as well. I figure it will be best to start from first principles, and either snap my tracings to MBs, or analyse clusters of MBs based on population density.

The net result of this will be an up to date equiv of Olivier and Cos (ex NZFS ex SNZ 91) urban areas layer. Ive studied this layer and it semi resembles MBs of old, its closer to existng town boundarys than SNZ's 06 UA boundarys, by a long way, but tends to have been outgrown by urban growth in many places. It also contains more than one shape per town in a manner thats fairly random, and lacks useable attributes. The end goals are a polygon boundary, a point political center and attributes: Name,TA,RC,Area,Pop.

My next question is does anyone have any tips for working with the 2006 meshblock dataset?

This is what i feel i need to do:

1. Remove all attributes, except MB06
2. Multipart to singlepart the entire dateset per unique MB06.
3. Field calc new attr AREA(ha)
4. Join xls data "usually resident population" into new attr POP
5. Field calc new attr DENSITY =POP/ha
6. MB Symbolize/query by DESNITY > N == urban.

However a couple of not "straightforwards".

1. the shape file contains multiple features per MB. Ftools is taking a long time to merge these, 3% per hour, eta 30hrs. QGIS will crash before then im sure of it. But I tested it on 20 features and it has the desired result.

2. According to SNZ [1] there are 41376 MBs in the 06 dataset. The shapefile has 48K odd features, the accompany data spreadsheet has 41384 MBs (pt 1, MB tab), on the part 1 key tab 41370 MBs. None of these agree!

Has anyone done any of this and can help?

Regards

Peter Scott


[1]http://www.stats.govt.nz/surveys_and_methods/methods/classifications-and-standards/geographic-hierarchy.aspx