GIS data questions


Phineas Packard
Hi All,

I have to get some GIS data for a project I am working on but I honestly have no idea where to start. What I need is as follows:

1. I need to get the lat and long of the point in a post/zip code with the highest population density (post codes are from Australia). I have the geographic center but in Australia the postcodes can be huge and the geographic center could be miles and miles away from where anyone lives.

2. I want to then calculate the distance between this point and the street address of another place using the road system. Niavely I assume I can just plug in the lat and long cords of both points into google maps API and pull out resulting distance but I am sure that there are better ways of doing this.

BONUS ROUND: Finally I am thinking about getting really wild and trying to calculate a radius in km of how far one could travel by car, on average, from a lat long point given the associated road systems around a lat and long point.


Less is more. Stay pure. Stay poor.
I have done rudimentary versions of this. In the US we have some good accessible data on populations and census tracts and at times smaller areas. My first hoorah, I had the longitudes and latitudes of specific hospitals and calculated distances (as a bird flies on an ellipsoidal shape) using a bunch of Tans, etc. Latter I found the GeoDist function in SAS. You would be interested in Manhattan Distances I believe. Also, how do you define the most person dense area. Are we talking cities or any place? In US, we have readily available population densities for tracts. However tracts are not all the same size. We have some Jeffersonian style blockings of areas, but the older the area the worst the area shape (kind of in general).

Yeah, I believe you can do everything you want, I just don't know the programs GISarc, or resources. Good luck!!


Super Moderator
Have you looked on the Geoscience Australia web site? They have bucket loads of demographic and other layer available for free - just need to register.


Probably A Mammal
I never did it before, but we talked about this in my GIS training. If you have another layer of data for population residence (parcels of the right type?), or negatively, a layer reflecting non-population centers like parks or industry, then you can take your areal boundary and crop out or keep only the part of it that is relevant. Then you use that remaining space (unit weight unless you have true grid density values) to compute the center of mass of the polygon. Does that make sense? Of course, if you have an actual density layer, then just computing the weighted center of mass should work to get a better centroid of the polygon.

As for driving distances, that's tricky. You can either take distances as the crow flies with a general average that one can drive such and such distance over this much time. Otherwise, you'll need a true road network layer to do drive times (to deal with one-way traffic and such; elevation might matter, too, if you need great accuracy). Manhattan distance would of course work well on such a network as a basic approach, but if you don't have too much data, you can try to use Google API. They have limitations, as it's not a service designed for analysis. It's designed to service web pages with limited traffic and API requests. Still, I made an ArcGIS (or started to) tool for doing just that. I'm sure the API has changed since then, however. If you have a budget, you can also just pay for such GIS services.

If anyone knows of R, Python, or Java libraries for these sorts of computations, that would be great to know! I haven't kept up on QGIS in years, but since they're open source, they might have some tools to assist in this since it integrates with Python.