Labs Fire Predictive Services Hackathon

Recently, Jerome and Joe participated in the inaugural Victorian Department of Environment, Land, Water and Planning (DELWP) Bushfire Predictive Services two day hackathon.

20161011 165750

The event, held in a co-working environment in the Melbourne CBD, brought together bushfire modelling and analysis, weather, and technical experts from DELWP, the Bureau of Meteorology, CFA, the University of Melbourne, Amazon Web Services and Geoplex to tackle some ‘meaty’ problems around bushfire modelling, prediction and visualisation.

Five teams tackled problems ranging from the most efficient way to scale AWS infrastructure to support statewide prediction of potential hot spots on bad fire condition days, to investigating approaches to visualising the inherent uncertainty in fire modelling, to building a user focussed web application for communicating how long a fire may take to impact a certain location.

Joe’s team tackled the problem of visualising uncertainty in fire models, by developing a prototype interactive web mapping application that displayed an aggregated output of locations predicted to be ‘impacted’ by a collection of fire models (differing by ignition point location), which were symbolised to represent four classes of ‘likelihood’. The mapping component was also linked to dynamic charts which showed the potential area of each ‘likelihood’ classification. This prototype was whipped up using ReactJS, OpenLayers3 and ChartJS with PostGIS and GDAL/OGR used for processing the data.

Jerome’s team was focussed on creating a simple and effective web application to communicate ‘time to impact’ for any location based on the forecast weather conditions. One of the goals of this prototype was to try to tackle community readiness in the event of a bushfire by providing a simple approach to visualising how long it may take to impact a location based on a statewide fire prediction model. Jerome’s team built their prototype using ReactJS, Leaflet with an AWS beanstalk, Ruby on Rails and PostGIS backend.

Overall, the two day event was fantastic and really provided excellent value to DELWP as it allowed them to test and prototype some ideas without committing to a high cost upfront. It also goes to show how a multi-disciplinary and cross-organisation approach to tackling problems with a mix of government and commercial resources can work together to achieve some amazing outcomes.

Well done to all the hack teams involved and to DELWP predictive services for putting on an excellent hackathon!