Auckland’s ‘leafy suburbs’ is a term we’ve heard a lot over the last few years as debate about the Unitary Plan has raged. But what suburbs are the leafiest? Now we have an answer.

Reader Euan has put together some maps looking at the number of public trees within 500m of each residential property and normalised the results to provide a degree of comparison between residences on the isthmus. He hopes to be able to do the North Shore and West Auckland in the future but unfortunately he says there is very little data available for the old Manukau City Council area.

On the analysis, he also notes:

The two major shortcomings associated with this analysis are: A) the public tree data was incomplete; and B) there is an underlying bug with the tool used that has not been fixed. While there’s not much I can do about the incomplete data, there are ways around the software bug. Unfortunately, they are time intensive, and I do not currently have the time to implement the work around. I can always return and fix them later.

In many ways the analysis serves to highlight the change in how we’ve developed cities and streets over the years. If you take a google streetview tour through some of the red areas, like those south of SH20, there are some trees but they’re generally only on private property rather than in the road corridor. I’d also suggest there’s probably a high correlation with the leafiness of an area and the quality of the footpaths. Again those mid-late 20th century developments south of SH20 tend to have narrow footpaths often right up against the road.

Despite the shortcomings, the results are fascinating. As already mentioned there is a serious lack of public trees outside of the streetcar suburbs with the exception being some of the eastern suburbs. One area that bucks this trend though is around Remuera where there is a surprising amount of red, more so than even the CBD. You can also see a fairly dense patch of public trees in Stonefields and I’d expect to see similar results in other new suburbs as the importance of street trees is now more understood.

Leafiness - Isthmus

The second map highlights the top 20% most leafy residences to more clearly show where the leafiest suburbs are

Leafiness - Isthmus - Most Leafy

Lastly this is an earlier map Euan sent me highlighting the data and showing the number of street trees in the CBD

Leafiness - CBD

All up a fascinating piece of work so thanks Euan for putting it together.

Share this


  1. The thing is that the NIMBYs will claim leafy suburbs is probably has not got much to do with street trees but private yards and private trees. Which is stupid.

      1. The part of Freemans Bay around lower Howe St that was redeveloped as “public housing” during the 1950s/60s “slum clearance” has many mature trees away from the street, are they public or private trees? Of course, a “leafy” suburb is defined by the presence of both public and private leafy trees, to suggest that only public trees provide the leafy feel is disingenuous. Epsom Ave is a good example, where many of the street trees are small and immature, and the large mature trees are on private land. Sadly, it is these mature trees that get felled for infill housing…

        1. Yep, unfortunately there are certainly plenty of problems with the tree data I used, both in terms of geographic scope and the tabular attributes of the trees themselves (age, diameter etc…). The lack of robust attributes for each tree means that there’s no way to uniformly differentiate an immature tree from a fully developed one. If that could be done, then you can weight the analysis appropriately.

    1. Hi Errol, I’m the student who quickly put these maps together.

      Cornwall Park is actually the park that tipped me off in regards to the data quality. The entrance area off Manukau Road, and the lane leading up to the summit, have all of the park trees recorded, but nothing more for the rest of the park. This obviously affects the results as some areas have more robust data that others. On that particular cluster, I too found it odd. I’ll have to go back and double check the logs, but I have a suspicion that the street network I used as the basis for this analysis might have contributed to that low/red cluster right by the park.

      Your comment on tree size also touches upon an interesting point. It is actually possible to weight this analysis by some attribute of the trees themselves. So, instead of simply counting the number of trees (as was done here), the calculated value could instead be amount of tree biomass within 500m of each residence. Doing something like this would mean that the quality of the trees would be much more important than the quantity of trees. I’m no tree-expert — so have no idea if biomass can be quantified in such a manner, and whether it would be useful here — but hopefully you get the idea. Weighting the analysis like this, however, would require a far more robust dataset than is available, unfortunately.

      1. Hi Euan, thanks for the maps. I’d love to have a look at the street tree dataset, but I can’t seem to find it on opendata or council’s GIS viewers – best I found was “Notable Trees” and LINZ Topo50 Tree Points. Is it data you had to specifically request from the council? Is it publicly available?

        1. Yeah sadly it will not be found on the Council GIS viewer. The data came from a University of Auckland GIS database that contains ~12TB of data from different government agencies. Unfortunately the data cannot be shared (as far as I am aware, anyway).

        2. Hey Evan.

          Could you share the name/source/etc of the dataset?

          It might be possible to download the dataset direct from the appropriate agency, or request it over email or something.

      2. Still a useful exercise that you have done, thanks.
        I assume the lack of data in Cornwall Park proper is at least partly due to it not being a Council reserve (while the Manukau Road entrance and summit are) – a lot of ‘private’ trees. I assume the same would apply to any DOC land (eg North Head?), schools (does Mt Albert Grammar still have a big farm near it?) etc.

        1. Ah. I wasn’t aware that the Council didn’t entirely own/run Cornwall Park. It is divided into two land parcels, with only one of those having its trees recorded. That’ll explain why.

          The MAGS animal farm is excluded, as is North Head. Can’t do much about MAGS and other private properties at the moment, but DOC might have a file floating around that can be integrated into the Council data. If/when I get around to refining this approach, then that’s definitely something to consider.

  2. Currently packing my tent to move to Albert Park…so leafy! Also I live in an apartment and have some greenery inside, but having access to so many parks is only something that the backyardless truly understand, I suspect. As far as am I concerned if you can walk to a park, a library and a train station…you are doing something right. But then again I abandoned the “leafy” suburbs many blue moons ago so have very little understanding of the NIMBY argument. It is important to have access, not exclusive ownership. And there are four trees across the road from me in the CBD. Now that is access!

  3. The problem is that Auckland Transport don’t plant enough berm trees
    What then happens?
    Empty berm = gets parked on
    Car rips up grass
    Ugly muddy mess

    Cmon AT! Limes, lemons are hardy as hell and could be put in all over the place on every berm.

    1. I’m not sure what they put in to the replace the mature tree outside my neighbors’ (taken down due to the new big water main), but I’m fairly sure they aren’t fruit trees.

    2. roading engineers hate things that aren’t frangible! I’ve been involved in projects where we were asked to remove street trees (within berms and the median) from the design because trees weren’t considered frangible and hence a safety hazard for when a speeding motorist loses control.

      1. As a traffic engineer the obsession with everything being frangible frustrates the hell out of me. In a 50km/h urban environment the risks of serious injury associated with having non frangible trees or furniture are really pretty low and in my mind well off-set by the positive effects of making drivers feel closed in which makes them drive a bit slower. This gives a much better outcome for all users of the road reserve.

        1. I haven’t had the “trees are a hazard” comment raised in all my 10 years of work as a NZ traffic engineer. Then again, almost all my work is in urban (50 km/h or below) streets. It may well still be alive on rural roads, but the worry that trees are being declined for traffic safety reasons is a (historically true) claim that is less and less true. As others have noted, it’s more about money / maintenance.

    3. My understanding is AT pays for the initial purchase cost of street trees but AC Parks pays for the maintenance so both have to agree to new plantings and Parks has no cash so is not keen

      1. Witness the ripping out of plants on Grafton gully cycleway. Someone can’t be bothered with maintenance, hence plants for motorway drivers but not for peds and cyclists. D’oh.

        1. NZTA pays for motorways maintenance. Cycleways are maintained by AC parks, so yes, there are some perverse incentives at work (though admittedly, much of planting along cycleways in the past HAS been way too close).

  4. Slightly counter-intuitive results in a few places I know well. Possibly because many large “private” trees give leafiness to the public areas also. I suspect some of Remmers has the same quirk, big old trees that are technically private. Of course the private ones are less protected.

  5. Some great use of GIS there Euan and a good place to start. Certainly the perception of ‘leafyness’ isn’t only restricted to public trees but also the trees and plants grown on private property which will add to the feel of an area. Another way to approach it might be to use aerial imagery to assess the concentration of green areas, converting a raster aerial image to polygon and then selecting out only those polys in a certain value range to use as a starting dataset to then whittle down to what you’re after. As aerial imagery is often flown in the summer the grass is generally browner so it will make it easier to quantify and not need to be included. You can then eliminate things unwanted values based on land use data perhaps (if you have that available). Then with what you have left run a similar analysis as you’ve done, or use hotspot mapping, or assign greenness concentration per census meshblock and then visualise via standard deviation or whatever (I’d leave that up to the statisticians to advise on that). It’d take some playing about to get optimal results and to not end up with unusably large datasets, but it could help correlate with your current findings.

    1. Remotely sensed data could certainly be integrated into the analysis. Indeed, a few years ago some researchers developed a method to measure the ‘greeness’ of streets using remotely sensed imagery.

      Haven’t looked at it for a while, but I believe their approach does require an explicit origin and destination. I guess it could be modified to work on a buffer based approach, as was used in this short analysis.

      Somewhat related to that, I recall seeing/reading somewhere that Portland recently used LiDAR to do a tree inventory of some sort. If you could leverage LiDAR data to both A) identify the centroids of individual trees; and B) provide some sort of estimation of biomass or the like (see my previous reply to Errol), then you could use these same methods to produce much more accurate results. Although I should say that I haven’t delved into raster analysis for a few years now. So I have no idea how hard/easy it would be to do all of that.

      On using areal analysis such as Meshblocks, I personally actually prefer to avoid them (or CAUs) due to the stats errors that they introduce (the modifiable areal unit problem, if you’re familiar with it). That, and they provide a poor representation of the built environment. On the other hand, using buildings or cadastral parcels (as I used here) provides a more detailed and realistic representation. At least IMO, anyway. In saying all of that, I’m pretty sure I ran some quick counts through LISA analysis at some point. Definitely something to keep in mind for future work.

      1. Point taken re MAUP problem, as I said I’m not a statistician! I’d consider trying to define ‘leafiness’ as something wider than just the presence of trees (other plants have leaves too), which is why using remote sensing would perhaps give you the concept of ‘greenness’ as you reference. That is – just focussing on trees is restricting your analysis a bit, unless of course this is the agreed definition of ‘leafiniess’. If you could define this greenness, and then select all instances of that within a standard buffer distance from a road centreline (e.g. 50m) which would present an adequate representation of the greenness visible from walking or driving down a street – i.e. contributing to the ‘leafy feel’. However you then standardise it to be comparable across the city is then up for debate. As for the tree data, while useful, is limited as you have said. So I’d look to use it as a good reference point to corroborate other data (remote sensing), rather than the core dataset and be disappointed that it doesn’t provide an accurate picture of real life. Anyway, am sure you’ve considered this – and would use better/more accurate datasets if immediately available! Good luck with further analysis.

  6. The ‘leafiness’ is not public trees only. Suburbs are leafy in part because people are able to have space to have their own trees in their own backyard. From an environmental perspective we need to have as many trees as possible, especially those that are able to provide food/shelter for native birds and insects. But to be fair, some of the “leafy suburbs” actually are probably not planted that well for that, tending to be planted more with trees that worked well in England.

    I hope that as we intensify insects like wetas and small reptiles like our native lizards are not lost completely from intensified areas. They will be likely to find it more difficult to find food sources or shelter if there are fewer trees and less green space, so that may happen.

  7. Why does this only look at public street trees? That’s absurd. By that token you’d conclude Titirangi has no trees!

    I don’t think I’ve ever read a more inaccurate assessment of something. The reality is pretty much the opposite of what is written.

    The vast majority of trees are on private land, which is why densification reduces leafy suburbs. Hiding that fact is possibly the motive for putting forward the badly flawed assessment?

    Developers of course hate trees. They always clear fell a site, rather than build around them.

    1. Intensification will not necessarily make a suburb less ‘leafy’. Stupid design will have much more effect.

      I mean, let’s see our typical south-east Auckland suburb

      Not very “leafy”, is it? Note how density is not the problem here, the average lot is probably over 500m² here.

      Now if you look overseas: here’s a random residential area close to where I grew up:

      That area is much more “leafy”, but it also has twice the density compared to the Auckland suburb above. Both images are at the same scale, in case you’re wondering.
      The key thing here (and in tramway era suburbs in Auckland as well) is that the garden has some depth, so there is some room between the house and the back fence to plant trees.
      Unfortunately we don’t have that tradition with planting trees along streets, and older streets over there tend to be too narrow to fit trees in. But newer streets tend to be a bit wider to accommodate street trees (and, of course, more parking).

      And the statement about developers is not true either. Some developers know the values of trees and will do some effort to keep them standing when developing a site.

    2. The “leafiness” is also closely correlated with the age of the suburb. For example, when Ponsonby and Grey Lynn were first developed there was very little tree cover – this is very easy to see when looking at historic images through the Auckland Libraries website. Much of the land on the isthmus and north shore was used for grazing farm animals in the 19th century and trees weren’t particularly useful. Many suburbs which people describe as leafy didn’t build around trees as you are suggesting we should now, they were blank plots and over time residents have planted trees to help personalise their property according to their preference at the time. The same thing will happen with densification, yes we may loose trees in the short term, but these will be replaced in time when new residents move in and grow into mature specimens that future generations will appreciate. I understand that we should be looking to protect exemplary tree specimens but blanket tree protection at the expense of new housing runs counter to the historical pattern of urbanisation in post-1840 Auckland.

    3. More conspiracy Geoff, ‘motives to hide the evils of densification’? Er no, it’s based on a dataset of public trees because that’s what is available.

      However if you would like to inventory all privately owned trees in Auckland so that we all have the right data then please go ahead. However if not, then you can hardly complain when people do analysis with what is available.

      I await your superior methodology and data sources with anticipation. Guest posts always welcome.

    4. Geoff, there are no doubt limitations in this analysis, but to come on here and make a personal attack (‘I don’t think I’ve ever read a more inaccurate assessment of something’) on someone who has volunteered their time to put this together reflects more on you as a person than it does on their work.

    5. This is exploratory work that was setup in about half an hour and left to run overnight. There were never any claims to being an objective truth on the matter or some such. I was just interested to see what patterns emerged with the data that was at hand. As such, I can only work with what’s available to me.

      While there are plenty of trees used for this analysis — 98,391 to be specific — there are still many that are undoubtedly left out. Most of them being on private land, I imagine. Are these trees absent included in other datasets? Probably not. But I haven’t searched extensively, so cannot say for sure at this point. Can I create my own tree registry? In theory, yes. James gave some suggestions above on how this could be done, but automating the identification and extraction of trees from aerial imagery is not an easy task. And one that I probably have to up-skill on in order to implement. As I have more pressing projects at the moment, it’s not really a viable option for now. It could be in the future, however.

  8. A good first effort but what is clearly required is the addition of further data relating to site coverage rules, minimum lot size and other planning tools that more or less guarantee that trees have insufficient space for their roots, let alone the canopy. As we intensify, most zones will have no private trees. In off-shore jurisdictions, street trees compensate for the lack of private trees. Of course that involves Auckland Transp[ort who currently have difficult distinuishing a street tree from a lamp post! Also services need to be put underground to facilitate that conpensatory action. The Australasian maze of aerial power and service lines currently prevent specimen trees being planted.

    1. Which plan are you reading? I see lots of rules requiring low building coverage ratios and lots of landscaping.

Leave a Reply

Your email address will not be published. Required fields are marked *