Go to top of page

Science and Innovation

Goal: World-class research and development enabling the Bureau to deliver
better climate, oceans, water, weather and Earth system prediction information and insights to its customers, and create shared value with our partners.

The Science and Innovation Group harnesses Earth science undertaken by an increasingly diverse and inclusive workforce to provide innovative systems and services that deliver great impact and value for the Bureau’s customers and partners. These systems and services are enhanced by mutually beneficial national and international relationships that allow the Bureau and our partners to achieve more together than they can alone.

For 2020–21, the Group consisted of two programs with the following responsibilities:

Program

Responsibilities

Research

Earth systems modelling
Weather and environmental prediction
Climate research
Innovation

Research to Operations

Transfer of research outputs to operations
Validation and verification of models and model output
Data assimilation and post-processing

Throughout 2020–21, the Science and Innovation Group focused on the delivery of five outcomes that support the delivery of the Bureau’s Strategy and purpose. The Group’s achievement in delivering each of these outcomes is discussed below.

Outcome 1: Customised impact-based forecasts and warnings when and where it counts: more localised, timely and better information for cities and regional areas.

Achieving the outcome

Developing future forecasting systems to predict tornadoes

Higher-resolution models are required to deliver more localised weather and climate information for cities, by more accurately forecasting hazards on neighbourhood scales. During the year Bureau researchers conducted cutting-edge numerical simulations to test the Australian Community Climate and Earth System Simulator (ACCESS) model’s capability to forecast tornadoes.

Analysis of ensemble simulations at a 100 m grid resolution has shown that future operational numerical weather prediction systems could provide a significant increase in tornado warning lead times.

Developing better numerical weather prediction systems

Bureau researchers worked alongside researchers at the UK Met Office and across the Unified Model Partnership to test and evaluate new versions of the regional modelling system used for weather and climate prediction. The prototype configurations have shown a range of improvements, including improved representation of atmospheric processes, in particular clouds.

Improving our wind forecasts

Significant improvements were made to the Bureau’s wind forecast process with two new wind grids added, one presenting the maximum wind over the last hour and one providing an instantaneous presentation of wind on the hour. These changes support the production of wind forecasts that are more consistent across all lead days and all synoptic situations, providing clarity in service definition and supporting improved fire danger rating forecasts.

Improving volcanic ash dispersion modelling

The Dispersion Ensemble Prediction System (DEPS) generates forecasts of the likelihood of volcanic ash dispersion and produces crucial information to help Bureau experts provide guidance to the aviation industry. An upgrade to DEPS progressed significantly during the year and was successfully tested in June to demonstrate it is ready for transition into operations.

Infographic showing the number of peer-reviewed journal articles published by the Bureau each year. In 2020-21, 155 peer-reviewed scientific journal articles were published.

Highlights and significant events

Improving thunderstorm prediction

In March, major improvements were made to the Bureau’s thunderstorm prediction system – Calibrated Thunder – which now delivers improved probabilistic thunderstorm predictions across all of Australia. The system is based on the Bureau’s global ensemble model ACCESS-GE3, calibrated against lightning observations from the Weatherzone Total Lightning Network. The upgrade has also extended thunderstorm probability forecasts out to nine days in the future and extended forecasts to provide an additional 200 km of coverage offshore.

Outputs from the system flow into the Graphical Forecast Editor and into Visual Weather,
the Bureau’s operational data visualisation system. This information is provided to Bureau meteorologists across the country for incorporation into forecasts and warnings. Further upgrades to the system are planned for 2021–22 under the Public Services Transformation Program and will focus on strengthening predictive skill and consistency to support improved forecast automation.

Forecast from the updated Calibrated Thunder system showing the calibrated probability of lightning over a 24-hour period (lower right panel) alongside key components used to statistically construct the forecast.
Forecast from the updated Calibrated Thunder system showing the calibrated probability of lightning over a 24-hour period (lower right panel) alongside key components used to statistically construct the forecast.

Improving weather prediction through high-resolution ensemble modelling systems

High-resolution weather models have the advantage of better representing small-scale processes in the atmosphere, including physical processes within severe convective storms, such as the updrafts and downdrafts, and the production of rain and hail. Because these models closely mimic the key processes of storms, they can predict the location and intensity of thunderstorms and provide unparalleled guidance for severe weather warnings of heavy rainfall, wind gusts and hail.

While high-resolution models present many advantages, the smaller grid sizes create a challenge in terms of predictability. Small-scale forecast errors grow quickly and because convective systems evolve rapidly, they are strongly affected by uncertainties.

In August, the Australian Community Climate and Earth System Simulator for cities (ACCESS-C) model was successfully upgraded to version ACCESS-C3 which includes hourly assimilation of observed data and greater use of locally produced observations. As a result, ACCESS-C3 delivers more accurate forecasts and can provide more rapid forecast updates (down to an hour) for Australia’s capital cities.

In December, the Bureau’s first convection-permitting ensemble model – ACCESS-CE3 – was introduced into operations for Australia’s eastern states, enabling the prediction of an ensemble of equally likely weather conditions for each forecast period for major city regions (as opposed to a single prediction). The degree of spread of this ensemble provides an estimate of the forecast uncertainty, which informs the level of confidence to place in forecasts of weather events.

One-hour rainfall accumulations (mm) for each of the convective-scale ensemble members for the New South Wales flood event of March 2021.
One-hour rainfall accumulations (mm) for each of the convective-scale ensemble members for the New South Wales flood event of March 2021.

Next steps

Key activities to be delivered in 2021–22 to help achieve Outcome 1 include:

  • replacing the main forecast guidance system, Gridded Operational Consensus Forecast, with Integrated Model Post-Processing and Verification (IMPROVER) to leverage ground-breaking developments in post-processing
  • commencing the migration of the Bureau’s operational numerical environmental weather prediction models to the Bureau’s new supercomputer and data centre
  • conducting research and development on current and next-generation numerical environmental weather prediction models to deliver more localised, timely and better information for cities and regional areas.

Outcome 2: Reliable and trusted forecasts: enhanced assimilation of observations for more accurate predictions.

Achieving the outcome

Improving extreme weather awareness with the National Analysis System

The National Analysis System was developed to produce hourly estimates of key weather variables across Australia on a 2.2 km grid, and works by consolidating a wide range of high-frequency weather observations using data assimilation with the Bureau’s numerical environmental weather prediction system.

The system was set-up in ‘real-time’ mode for testing, and successfully produced frequent and continuous outputs of key weather variables for extreme events such as storms and hail between November and May. This demonstrated its suitability as an operational system on the Bureau’s new high-performance computing infrastructure, where it will provide more timely awareness of evolving weather situations.

Improving quality control processes for observations

The Bureau introduced the Monitoring and Updating of Station Lists (MUSLi) system to enable regular statistical monitoring of non-satellite observations used in the city-scale forecasting model ACCESS-C. Statistics, plots and maps are generated by MUSLi to provide information about the quality of surface observations from Australian network stations. This enables poor-quality observations to be identified and removed from forecasting models, leading to greater forecast accuracy.

Using multiple data sources to provide better weather forecasts

The Gridded Operational Consensus Forecast (GOCF) calibrates and combines local and international numerical weather prediction model outputs to increase the skill and value of weather forecasts, and underpins the Bureau’s efforts to automate weather forecasting. In 2020–21, GOCF wind speed forecasts were dramatically improved in both space and time, particularly in populated coastal areas and around rough and elevated topography, such as the Great Dividing Range. Rainfall, humidity and temperature outputs were also improved in a major upgrade of the GOCF, contributing to higher quality and more accurate forecasts, including for forest fire danger indices.

Highlights and significant events

Maintaining observations in the wild Southern Ocean

Ocean buoys provide the Bureau with valuable data to improve forecasts and understand oceanographic characteristics but require careful maintenance to ensure they continue to operate at high capacity. The annual maintenance program includes a trip to one of our most remote observation sites in a wild part of the Southern Ocean in March.

The critical ocean buoy, known as the Southern Ocean Flux Station, is part of an international network of ocean observation sites and is anchored to the ocean floor 4.5 km below at 47 ° south, between Tasmania and Antarctica – about 36 hours from Hobart by ship. The location was selected primarily for its oceanographic characteristics, but also provides a rare set of atmospheric observations in a data-sparse area known as the ‘roaring forties’ – unrelenting gales, large waves and severe storms.

Map showing the location of the ocean buoy (triangle) southwest of Tasmania. Colour and arrows indicate prevailing wind speed and direction.
Map showing the location of the ocean buoy (triangle) southwest of Tasmania. Colour and arrows indicate prevailing wind speed and direction.

The buoy provides important observations such as near surface winds, air temperature, humidity and pressure, rain, sunlight and infrared radiation, and sea surface temperature, which are used to estimate the transfer of energy between the ocean and the atmosphere as well as carbon-dioxide fluxes and a host of bio-geo-chemical ocean measurements.

Members of the maintenance team (foreground) in front of the Southern Ocean buoy (background). Photo courtesy of Ben Arthur (CSIRO).
Members of the maintenance team (foreground) in front of the Southern Ocean buoy (background). Photo courtesy of Ben Arthur (CSIRO).

Next steps

Key activities to be delivered in 2021–22 to help achieve Outcome 2 include:

  • integrating advanced observations capabilities with operational systems for improved situational awareness and nowcasting of severe weather
  • developing and using tools to quantify the value and quality of observed data to numerical environmental weather prediction systems and automated operational forecast processes
  • developing and utilising enhanced verification to ensure the quality of additional advanced observations.

Outcome 3: An Earth system numerical prediction capability: fully integrated atmosphere, ocean, sea-ice and hydrology models.

Achieving the outcome

Improving the representation of clouds and rain in ACCESS

While higher-resolution models improve many aspects of rainfall prediction, there are numerous deficiencies that impact forecast accuracy. During the year, Bureau researchers added an additional cloud microphysical process to the ACCESS forecasting model to represent the heterogeneous freezing rain process. This development produces a more realistic representation of convective clouds and improves simulated rain rates in the tropics. This model development will be included in the UK Met Office’s next regional model release and will be used in weather forecast systems by Unified Model partners around the world.

Improving forecasts of ocean conditions

In May, the Bureau’s Ocean Model Analysis and Prediction System – OceanMAPS – was successfully upgraded to version 3.4 featuring an improved scheme for assimilating observed data. The upgrade reduced forecast bias for ocean variables such as sea surface temperature, salinity and wind to support better decisions around maritime-related activities.

Providing high-resolution regional ocean forecast systems for Defence

In June, the Bureau delivered the first operational regional ocean forecast system for Defence. This system has an increased horizontal resolution when compared to the global ocean forecast system (OceanMAPS) and includes tidal forcing, which enables the representation of internal and external tidal modes that are important for oceanic processes such as ocean mixing. The increase in resolution and added physics ensure that the forecast system will give an improved representation of the apparent oceanic variability. This forecast system represents an important missing link between the global ocean forecast product and the finer coastal scale ocean models developed by partner agencies. A second regional ocean forecast system has also been developed with a go-live date in August 2021.

Developing a new wave forecast system

During the year, the Bureau developed a new wave forecast system to replace the existing global and national wave forecast models. The new wave model (AUSWAVE-G3) features ~12 km spatial resolution globally with refinement around sub-grid scale features at ~6 km resolution, and uses guidance from the Bureau’s numerical environmental weather and ocean prediction models.

Wave height forecasts for 7 days ahead produced by AUSWAVE-G3 are of a similar level of accuracy compared to forecasts for 4 days ahead produced by earlier models. The multi-resolution global wave model provides seamless wave forecasts up to 7 days ahead at higher resolution around Australia; an improvement from 3 days ahead previously available.

Highlights and significant events

Developing the new Australian Water Outlook

The Bureau successfully developed and tested a new operational Australian Water Outlook service, including a suite of services and nationally consistent outputs such as historical and near-real-time hydrological information (from 1911), short-term and seasonal hydrological forecasts (for the next 9 days and 3 months ahead), as well as projections of hydrological change for a series of 30-year time periods out to the end of the century.

The system uses a range of climate variables from observations and modern operational climate models, coupled with cutting-edge analysis techniques and assimilation of near-real-time satellite soil moisture data. With these data inputs, the Australian Water Resource Assessment Model (AWRA-L) provides a consistent set of hydrological outputs – such as soil moisture, evapotranspiration and runoff – on a ~5 km grid across Australia.

The Australian Water Outlook is the culmination of collaborative scientific and operational systems development by the Bureau and our research partners (CSIRO, ANU, University of Melbourne, UNSW, Monash University, Griffiths University and ARC Centre of Excellence for Climate Extremes).

In June, the AWRA-L seasonal forecasting suite was successfully deployed to the Bureau’s high-performance computing production environment and will enhance Bureau’s capability to provide landscape water balance information for Australia.

Projected rainfall changes for 2050.
Projected rainfall changes for 2050.

Next steps

Key activities to be delivered in 2021–22 to help achieve Outcome 3 include:

  • conducting research and development on upgraded seasonal, ocean and wave predictions
  • prototyping experimental numerical environmental weather and ocean prediction models, and Next Generation Modelling System components
  • undertaking high-resolution water and water resources modelling for disaster risk reduction, seasonal planning and operations and long-term infrastructure and investment planning.

Outcome 4: Seamless weather and climate insights: historical observations and predictions, from minutes to decades.

Achieving the outcome

Predicting rainfall bursts

As part of the Northern Australia Climate Program, a new prototype forecast product that shows the potential for 3-day rain events (known as rainfall bursts) to occur was developed in a collaboration with the University of Southern Queensland.

Rainfall bursts over Australian tropical and semi-arid regions are typically associated with broad-scale convection during the wet season and can impact business decisions in the grazier industry. After consulting with customers, four different 3-day rainfall thresholds were identified to represent soil type and property decisions.

The prototype rainfall burst forecast product comes from the Bureau’s seasonal forecasting model ACCESS-S1 and was released on the Bureau’s Forecast Visualisation Tool in November for testing by program partners in the grazing industry. Following positive feedback the rainfall burst product will be made available to the public from late 2021.

Map showing the forecast probability that a rainfall burst, defined as at least 20 mm in 3 days, will occur in the week of 24 February to 2 March.
The forecast probability that a rainfall burst, defined as at least 20 mm in 3 days, will occur in the week of 24 February to 2 March.

Delivering an improved understand of key natural hazards

In June, the Bushfire and Natural Hazards Cooperative Research Centre (CRC) completed its work program, helping the Bureau, emergency management agencies and other research institutions within the CRC to better understand and predict a range of natural hazards.

Bureau scientists developed an improved understanding of tropical cyclones, severe thunderstorm events, bushfires and east coast lows. Improved hazard impact forecasts were investigated, as were methods of using weather models to provide soil and fuel moisture information to calculate fire danger. The Bureau will continue to build on the important research effort undertaken through the CRC over a number of years.

Highlights and significant events

New multi-week and seasonal forecasts of extremes

Climate variability and extremes represent a key risk for agriculture in Australia. Through the Australian Government’s Rural R&D for Profit program’s ‘Forewarned is forearmed’ project, the Bureau developed the first forecasts of extreme climate events for weeks to seasons ahead.

As part of the project, Bureau researchers produced a large set of prototype forecast products for trial by the agriculture sector. Industry reference groups from the dairy, red meat, grains, sugar and wine sectors played a crucial role in the development of these new products, as well as providing important feedback on user needs for seasonal forecasting and impacts on farm decision-making.

A strong feature of the project was the direct link between research and industry reference groups. This approach enabled efficient feedback between the different components, facilitating a faster and more effective path to delivering practical outcomes to customers.

Informed by industry engagement, a subset of five forecast products have been selected to become operational and will be delivered to the public on the Bureau’s website through a staged approach later in 2021.

The rainfall decile bars product uses location specific bars to indicate the shift in probabilities compared to usual across the deciles. This example shows rainfall forecasts for Dubbo for upcoming months and seasons.
The rainfall decile bars product uses locationspecific bars to indicate the shift in probabilities compared to usual across the deciles. This example shows rainfall forecasts for Dubbo for upcoming months and seasons.

Advancing climate change science

After a six-year work program, the National Environmental Science Program Earth Systems and Climate Change Hub concluded in June. The goal of the hub was to develop actionable and foundational science to support a more resilient Australia. Hub partners included the Bureau, CSIRO the Australian National University, University of Melbourne, Monash University, University of Tasmania and University of New South Wales. The Bureau was involved in many aspects of climate change science undertaken by the hub, including climate projections, climate change attribution and impact analysis.

Science from the hub featured heavily in examination of recent high-impact extreme events such as Black Summer 2019–20, while also informing longer-term resilience planning such as the production of climate change projections to inform adaptation planning in the Gondwana Rainforests of Australia World Heritage Area. This research projected a continued increase in temperatures but a decrease in humidity, as well as changes to the cloud height, which could impact the tolerance of rainforest species.

 the Queensland Future Climate (QFC) medium emissions scenario, the QFC high emissions scenario and the New South Wales and ACT Regional Climate Modelling high emissions scenario. The magenta cross indicates the location of one of the World Heritage Area Gondwana Rainforests, the Tweed Caldera.
Ensemble mean change (metres) of cloud base height for the month of September, 2020 to 2040 relative to 1990 to 2010 for three ensembles: the Queensland Future Climate (QFC) medium emissions scenario, the QFC high emissions scenario and the New South Wales and ACT Regional Climate Modelling high emissions scenario. The magenta cross indicates the location of one of the World Heritage Area Gondwana Rainforests, the Tweed Caldera.

Sharing weather and climate knowledge in outback Western Australia

In May, the Bureau participated in a ‘climate roadshow’ to the Gascoyne and Pilbara shires of outback Western Australia as part of the Northern Australia Climate Program. The trip focused on helping regional producers to find, understand and use relevant forecasting information and comprised 10 station visits over eight days, covering more than 2000 km.

Alongside the Bureau were expert climate scientists and ‘Climate Mates’ from the University of Southern Queensland and Rangelands Natural Resource Management. Through a series of workshops, Bureau experts discussed important climate drivers and their impacts and participants provided practical advice and feedback for improving the Bureau’s delivery of forecast information.

Dr Andrew Marshall discusses the El Niño–Southern Oscillation at De Grey Station in the Pilbara.
Dr Andrew Marshall discusses the El Niño–Southern Oscillation at De Grey Station in the Pilbara.

Demonstrating multi-year El Niño forecast skill

Bureau researchers demonstrated that useful skill exists in predicting the El Niño–Southern Oscillation (ENSO) up to at least 18 months in advance. The research, conducted under the Northern Australia Climate Program and in collaboration with researchers at the University of Southern Queensland and the European Centre for Medium-range Weather Forecasts (ECMWF), analysed a hindcast set for years 1901 to 2010 with the ECMWF seasonal prediction system from 1 May and 1 November.

A key finding of this research (shown in the figure) is that forecasts can usefully predict if a current El Niño will terminate in the following year, or continue. Back-to-back El Niño years typically lead to a period of prolonged drought, so the ability to predict El Niño beyond one year would be very valuable to the Bureau’s customers. As a result of these findings, the Bureau will investigate extending its El Niño outlooks up to 2 years into the future, compared to the current 6 months.

This figure summarises multiple 2 year forecast (solid line) and observed (OBS, dashed line) evolutions of Pacific Ocean sea surface temperatures (SST) starting near the peak of an El Niño event in November. The +1 and +2 in superscript indicates years since beginning the forecast. Because the solid and dashed lines trend similarly, this demonstrates the forecasts are accurately predicting if the following year will see another El Niño event or not.
This figure summarises multiple 2 year forecast (solid line) and observed (OBS, dashed line) evolutions of Pacific Ocean sea surface temperatures (SST) starting near the peak of an El Niño event in November. The +1 and +2 in superscript indicates years since beginning the forecast. Because the solid and dashed lines trend similarly, this demonstrates the forecasts are accurately predicting if the following year will see another El Niño event or not.

Next steps

Key activities to be delivered in 2021–22 to help achieve Outcome 4 include:

  • delivering advice to customers on the impacts of climate variability and change, and developing user-focused products for sub-seasonal, seasonal and climate change timescales
  • developing seamless weather and climate forecasting spanning the gap between days and weeks through seasons, near seamless national water prediction, and short-term to seasonal water forecasting
  • producing national climate projections for Australia, delivered in partnership with other agencies, and the development of methods underpinning temperature and attribution service.

Outcome 5: Innovation excellence that exceeds community and customer expectations, in a safe and diverse environment.

Achieving the outcome

Harnessing innovation to improve Bureau services

Implementation of the Bureau’s new Innovation Framework, and accompanying three-year roadmap of actions, commenced during the year. All actions have progressed as planned, with seven actions becoming ongoing activities ahead of schedule. Activities included:

  • a strategic foresight investigation on artificial intelligence and machine learning, developed with the Data Science Community of Practice
  • a machine learning event series which highlighted several potential uses of the technology to improve Bureau services
  • significant in-roads into developing an enterprise approach to intellectual property management that supports the Bureau’s innovation objectives
  • significant progress into investigating options for a development ‘sandpit’ for experimentation and testing purposes, which would be suited to testing ideas with customers
  • enhanced mobility of staff to work in other workplaces to develop skills, relationships and experience relevant to the Bureau’s mission
  • engagement with Communities of Practice with subject matter expertise in data science and project management.

Exchanging expertise and experience across the Australian Public Service (APS)

During the year, the Science and Innovation Group participated in a pilot staff exchange with the Australian Bureau of Statistics (ABS) as part of the APS Data Professions stream. The pilot helped to build the skill sets of staff and create a network of data professionals across the APS. Under the pilot, staff seconded from the ABS worked on new innovations to apply machine learning to the prediction of severe weather events.

Science Advisory Committee established

The Bureau’s Science Advisory Committee (BoMSAC) was established in readiness for its first meeting in November 2021. The Committee was established to share knowledge and experience and provide advice on the implementation of the Bureau’s Research and Development Plan and review its performance.

BoMSAC will also review and monitor the alignment between the Bureau’s planned research and developments within the global and Australian science community and advise on new opportunities for collaboration with complementary organisations for the benefit of Australia and the global meteorological science community.

Creating opportunities for people with disabilities

The Bureau’s Research Program was pleased to host two interns from the Australian Network on Disability’s ‘Stepping Into’ internship program, providing the Bureau with access to graduate talent while providing the interns with useful career experience. The interns worked on quality-controlling a large set of seasonal forecasts and supporting data assimilation for the Bureau’s city-scale forecasting model ACCESS-C. The interns brought enthusiasm and capability to their respective teams, offering useful first-hand insight in developing inclusive work practices.

Next steps

Key activities to be delivered in 2021–22 to help achieve Outcome 5 include:

  • further implementing the Innovation Roadmap, including implementing a process for evaluating ideas through an innovation pipeline
  • holding the first meeting of the Bureau of Meteorology Science Advisory Committee and leveraging the Group’s knowledge and experience
  • supporting the Group’s implementation of enterprise diversity action plans, the achievement of diversity targets and further embedding work, health and safety.