It’s nice to know where a lake lies. It’s better if you know how fast rain can get into it
Perhaps in need of a high-value product after its G-NAF (Geococded National Address File) was published for free a Data.gov.au, the PSMA has decided to “fill in” the white space in national mapping.
As Dan Paull, PSMA’s CEO, told Vulture South, a map that showed address points, property boundaries, road centrelines and rivers left out important analytical information.
“The buildings and the infrastructure – that represents a lot of investment”, he said, but there was “no means to conduct analysis at scale, across the whole country”.
Hence the new “Geoscope”: a GIS-ready database that “describes the whole of the built environment – footprints, height, roof materials, solar panels, swimming pools, information about vegetation in the area, land cover”, and even details like surface imperviousness – whether it’s road or rocks, bare earth or a body of water.
Imperviousness, Paull said, is a good example of how much better analysis can be with the Geoscope data set.
Even open source GIS tools have well-established libraries for predicting flood mapping, but they focus on how a catchment fills “from the bottom”, so to speak.
DigitalGlobe co-founder and CTO Walter Scott explained that “water doesn’t start at the lower level, it starts where it falls”.
In an urban environment, he explained extreme weather can lead to flood damage by water that’s coming downhill towards the catchment – because even half a meter of water moving at speed can do considerable damage to whatever lies in its path.
Understanding the surfaces on the path between rainfall and the bottom of the catchment therefore helps predict where, because of the volume of water, a property might be at risk.
“The major risks – fire risk, flood risk, wind risk – for all of these things, to answer the question about likelihood, you need information about the building, and its surroundings,” Paull said.
With that data filled in, “you can look at one building, or every building in the country, and understand the possibility of a tree falling”.
Scott said creating the Geoscope dataset was a combination of machine learning and crowd-sourcing.
Machine learning helped “identify building footprints, identify the land cover and the delineation of trees.”
Roof materials are machine-classified according to data collected by shortwave infrared satellite sensors, while multiple satellite images of buildings from different angles helped measure building height and roof pitch.
Crowd sourcing, Scott said, helped identify things like swimming pools and whether there are solar panels on roofs.
The first release of Geoscope covers Sydney, but by 2018, Paull said, the product will cover the whole eight million square kilometres of the Australian continent.
As with other high-value PSMA data sets, Geoscope will be offered through resellers, with end user licensing depending on what data a customer wants and services shipped with the data. ®