Archive for the ‘shapefiles’ Category

I’ve long been on the search for a free, national road network GIS dataset for Australia. The search has been surprisingly difficult.

If you expect like me that the government would produce an official, national road network GIS database, then your expections are about the be dashed. In Australia, the maintenance of roads, and therefore their corresponding GIS datasets is a state responsibility. This means that there is no single, national official road network produced by the federal government. Even the data owned by state governments is not generally made freely available. While it is sometimes possible to contact state governments and receive a copy of a statewide road network, if you want to assemble a national road network the process has to be repeated for all states and territories. Furthermore, government derived networks generally have onerous licensing terms that mean they cannot be redistributed.

So what data sources are available? I have experience with four:

1. The canonincal combination of state-government road networks is produced by the Public Sector Mapping Agency (PSMA) Australia. Their Transport and Topography product, released quarterly, has been derived from the relevant bodies from each Australian State and Territory jurisdiction as well as the Commonwealth mapping agency. The major advantage of the Transport & Topography dataset is that it is up to date. PSMA also attempt to maintain consistency in attributes across the state derived networks (with varying degrees of success in my experience). Furthermore, PSMA attempt to correct spatial topology errors and perform automated quality assurance. Transport & Topography also has road names that are compatible with the address names in PSMA’s excellent Geocoded National Address File (G-NAF), which is an additional bonus. The downside is that it is very expensive, and its redistribution terms are restrictive.

2. A time-honoured source of road network data is VMAP0. VMAP0, short for Vector Map level 0, is a global data set, declassified by the US National Geospatial Intelligence Agency. VMAP0 was created from the digitisation of military navigational charts, and is perhaps still the most comprehensive global data set for topographic map data that is freely available. That said, there are two major problems with this dataset. First, because it was created from 1:1,000,000 scale maps, the road network is very sparse, meaning even some major roads are not listed. Second, VMAP0 has not been updated for twenty years! The advantage of this dataset is that it is public domain licensed, meaning you can do what ever you like with it. VMAP0 for Australia is available from the USA National Imagery and Mapping Agency, but requires software that can read its funky format like OGR. I believe ESRI also used to distribute VMAP0 shapefiles branded as the Digital Chart of World.

3. The new kid on the block is OpenStreetMap (OSM). OSM uses data contributed by volunteers (a.k.a. ‘Volunteered Geographical Information’ or VGI) to build up a rich spatial database (think Wikipedia for maps). While they are perhaps best known for delivering Google Maps like tiles on their website, you can also download their raw data. The good people at CloudMade take the raw OSM data, convert it to a number of accessible formats, split it up by country and make it available for download on their website. I am hesitant to use this data at the moment, I as don’t know how about its completeness in Australia, or how to selected out roads and not footpaths, for example. However, this has the potential to be a real competitor with Transport & Topography in terms of its currency and completeness, so I’m keen to learn more from anyone using OSM road network data in GIS analysis.

4. Finally, Geoscience Australia (GA) has a series of vector layers that is uses to produce 1:250,000 topographic maps for the whole continent. These are excellent data, consistent and well-curated, but with a couple of caveats. First, because these layers were created for topographic mapping, they are only at a course level of detail. Not all roads are included, particularly in residential areas, and those that are may have been moved in the cartographic production process to stop them from overlapping. Second, these data were last updated in June 2006, meaning that for some purposes they will be too old. Nevertheless, Australia’s national road system doesn’t change that fast, so for some purposes this will be an ideal dataset. Best of all, in 2009, GA relicensed this previously proprietary dataset to the Creative Commons 3.0 by-attribution license, meaning that you can use it, redistribute it and derive your own products from it, as long as you acknowledge Geoscience Australia as the original custodian. These data are freely available at the national scale in ArcGIS Personal Geodatabase format from GA although if you want pre-made shapefiles you’ll have to download individual map grids or buy their product.

If you know of any other national road networks for Australia, let me know.


In a previous installment I demonstrated how to use ogr2ogr to reproject shapefiles. It turns out that its a bit trickier when you are reprojecting from AGD66 to GDA94 (and vice versa). This is because the ellipsoids they use to model the earth are different, and it turns out the be quite difficult to calculate the exact equivalent locations between them.

Difference between AGD66 and GDA94 at the ANU

Difference between AGD66 and GDA94 at the ANU

The most accurate way to convert between the two coordinate reference systems is to use a pre-calculated distortion grid. Fortunately, some clever surveyors have made one for us, available here (Webcite archive).

To download the distortion grid and put it in the right place for us to use, you can paste the following into your bash shell:

mkdir -p ~/bin
mv "A66 National (13.09.01).gsb" ~/bin/a66_national.gsb

I’ve also archived the grid files using Webcite, if the ICSM link is dead replace please see the archive.

Please note that if you change the filenames to something that contains spaces or funky punctuation ogr2ogr might not find it, and will not perform the grid transformation, leaving you with incorrect coordinates. So I would recommend just running the above code verbatim.

Then, to transform a shapefile from AGD66 to GDA94 type the following:

ogr2ogr -f "ESRI Shapefile" -s_srs "+proj=longlat +ellps=aust_SA +nadgrids=~/bin/a66_national.gsb +wktext" -t_srs EPSG:4283 outputgda94.shp inputagd66.shp

You will need to change the filenames outputgda94.shp and inputagd66.shp to whatever suits you.

Firstly, you might want to view the projection information in the shapefile, ROAD_CENTRELINES.shp in this case.  The perl one-liner stops ogrinfo from printing copious data about each vector feature in the shapefile:

fmark@fmark-laptop:~$ ogrinfo -ro -al ROAD_CENTERLINES.shp | perl -e 'while (<STDIN>) { if (/OGRFeature/) { exit } print; }'
using driver `ESRI Shapefile' successful.

Geometry: Line String
Feature Count: 4596
Extent: (128.999678, -25.998883) - (138.000000, -11.067068)
Layer SRS WKT:
GEOGCS["Longitude / Latitude (GDA 94)",
ROAD_NAME: String (100.0)
ROAD_TYPE: String (15.0)
ROAD_SUFFI: String (15.0)
ROAD_LABEL: String (130.0)
Length: Real (19.11)

We can see from this that the geographic coordinate system of the shapefile is GDA94.

In order to specify the projection we want to use, we need to know its EPSG code.  What’s an EPSG code?  Its a number that uniquely defines some well-known projections (according to Wikipedia, EPSG stands for the European Petroleum Survey Group, so we have big oil to thank for our daily GIS tools, as well as big military). Some useful Australian EPSG codes are:

4326 WGS84
4283 GDA94
4202 AGD66
3577 GDA94 / Australian Albers
3112 GDA94 / Geoscience Australia Lambert
28348 GDA94 / MGA zone 48
28349 GDA94 / MGA zone 49
28350 GDA94 / MGA zone 50
28351 GDA94 / MGA zone 51
28352 GDA94 / MGA zone 52
28353 GDA94 / MGA zone 53
28354 GDA94 / MGA zone 54
28355 GDA94 / MGA zone 55
28356 GDA94 / MGA zone 56
28357 GDA94 / MGA zone 57
28358 GDA94 / MGA zone 58

More can be found at

In this case, I want MGA zone 53, or EPSG:28353.  So to reproject:

ogr2ogr -t_srs 'EPSG:28353' ROAD_CENTERLINES_MGAZ53.shp ROAD_CENTERLINES.shp

Note that the output filename comes before the input filename.  Also, for some reason on ubuntu karmic there was a missing dependency, so I needed to:

sudo apt-get install libproj-dev