2018-05-29T08:31:00New web application improves fire danger predictionDigitising complex fire management calculations and supporting land managers’ decisions.Fire danger prediction and decision support system.Seven Day Fire is an initiative borne from the ground up. Land managers in the Queensland Parks and Wildlife Service identified the need to use digital technology to aid the complex calculations required to predict bush fire danger over the next seven days. Traditional fire danger predictions are time consuming and complex. This often results in decisions being made on poorly analysed data or gut feelings.
Seven day fire digitises the process and makes it available on mobile devices.
Phase one (current) involves software engineering and testing within Government however if successful, this tool could be made available for all land managers across the country, as a Digital First initiative of the Queensland Government.
Application includes state land managers, rural fires staff and volunteers as well as the agricultural sector.
Improved accountability and integrity in decision making in fire management operations will reduce costs to the state of instances of poorly planned fires escaping and threatening infrastructure and lives.Emergency services and safetyEnvironment, land and waterOtherGovernment employeesPeopleCollaboration,Connectivity,TrustHarness skills and experienceSolve the right problem,Digital by default,Prefer open over closed,Harness skills and experience – from inside and out,Leave no one behind,Experiment, learn and improveCurrentNational Parks, Sport and RacingAgriculture and Fisheries,Queensland Fire and Emergency ServicesGovhack initiative
2018-06-05T08:35:00Storm tide forecasting with machine learningForecasting sea level up to 24 hours in advance during extreme weather eventsA Proof of Concept project that trialled machine learning to predict sea level up to 24 hours in advance during extreme weather events, the PoC informed DES about machine learning use to improve science delivery.As part of the Department of Environment and Science (DES) Accelerating Science Delivery Innovation Project, the Coastal Impact Unit partnered with Deloitte to conduct a machine learning Proof of Concept (Poc) to predict sea level up to 24 hours in advance. The objectives of this PoC included determining if machine learning can be used to improve quality outputs to support better science decision making, to understand the technologies available to support machine learning and to determine the benefits of using machine learning.
The current process predicts tide level and not storm surge levels. The PoC aimed to forecast the residual (storm tide level) up to 24 hours in advance. The resulting information can be used by government, the private sector and the community to support activities such as evacuation and recovery decisions during extreme weather events by local government, incorporated into policy (QFES Tropical Cyclone Storm Tide Warning - Response System Handbook) and forward planning of marine based recreational and commercial activities.
The team used an agile procurement model similar to that used by the Testing within Government (TWIG) program for a quick and targeted engagement, including a problem statement as well as discovery and pitch sessions. Delivery used an agile approach with showcases providing a good avenue to engage with other science areas in DES. Co-location and daily meetings were successfully used to ensure knowledge transfer.
The machine learning models delivered by the vendor addressed the original problem, however more work would be required to progress this to a usable forecast system. The AWS machine learning environment was used for the PoC and provisioned by the High Performance Computing support team in DNRME. This was achieved in a short timeframe due to the support team's ability to directly manage the virtual environments in the cloud.
The learnings from this PoC can be used to build machine learning capacity and more design efficient data management systems to enable Storm Tide Prediction, and solve other problems with machine learning. If machine learning moves to operational use for Storm Tide Prediction, the ultimate benefits to the State and Queenslanders will be improve public safety, minimise injury, property damage and economic impact through better advice during weather events and using the information for better urban planning.Environment, land and waterGovernment employeesPeopleExperiment, learn and improveHarness skills and experienceCompleteScience, Information Technology and InnovationNatural Resources and MinesDeloitte