Much of CSIRO’s work has application in many countries – our contribution to the development of wireless networking (‘WiFi‘) is well known, and the technology is ubiquitous around the world. But our expertise is varied and the focus of our Data61 unit – on data innovation and digital enterprise – means much of our global impact is focused on contemporary, digital solutions to age-old problems, such as natural disasters.

Few natural threats are as unsettling as flooding – the scale of these disasters are significant, and can cause long-lasting damage to infrastructure, property and in many cases, result in the loss of life. Advances in computing power, algorithms and domestic expertise on flooding make Data61’s research on predicting the outcomes of these disasters valuable in many global regions.

The Chilean project

The Copiapó River basin in the Atacama region, Chile, normally has a dry climate, but is vulnerable to natural disasters. Infrequent but extreme rainfall results in flash flooding – in March 2015, disastrous floods resulted in a series of mudflows that caused 26 deaths in the region and significant property damage.

“Catastrophic flooding events in 2015 and 2017 in Chile are a clear illustration of the significance of infrastructure decisions in regions prone to this particular mode of disaster”, says Dr Raymond Cohen, Senior Research Scientist on the Swift team at CSIRO’s Data61. “Copiapó city is downstream of large catchments, and it’s a very dry region. Water runs off much of the terrain – there aren’t any areas that could absorb water, such as forests. There is also no storm water drainage in the town. It’s a mix of factors that results in a very serious impact of flooding”.

The challenge this region faces is neatly aligned with the Swift team’s work – creating simulations of urban floods, and changing variables to inform decisions around infrastructure. This includes modelling how effective mitigation measures will be, such as spillways and valves.

“We currently have a live project with the Dirección de General de Aguas (DGA) in Chile”, says Dr Mahesh Prakash, senior principal research scientist and group leader at Data61. “Our prototype focuses on Copiapó, and the DGA is interested in licensing this software for the whole of Chile. We’ve also had interest come through from China and India”.

Local expertise

Much of our global expertise stems from proving the technology in domestic contexts. “We looked at records of Tropical Cyclone Alby in 1978. Flood maps of Bunbury in Western Australia showed flood heights at specific locations”, says Dr Cohen. “We put that information into our software, ran the simulation and compared Swift’s output to the actual records of flood height. This is a process we use to verify the operation of the model. We’ve also done work on the City of Port Phillip, Townsville, Toowoomba, the Shire of Murray, the City of Geelong and the Kakadu National Park”.

Old cartographers flood map of Bunbury in Western Australia showing flood heights at specific locations; overlaid with historical images of part of the town under flood water.

There are challenges in modelling urban floods. Variations in surface flow are impacted by urban drainage networks,which change regularly, which in turn has a large impact in the resultant effects of inundation and flooding. There are difficulties in sourcing geospatial and weather data, too.

“The value in getting this right is significant – understanding the nuances of creating these models can help planners prepare for disastrous events well in advance”, says Dr Prakash. “A core philosophy in our work is building flexibility into our products, such that they can be used across the world, for a range of different applications and different timescales, and by a variety of users”.