Profile with Dr Matt Beaty on the Australian Climate Service
‘Building the scaffold in your brain where different techniques sit can really help you.’
Meet Dr Matt Beaty, a Spatial Data Scientist currently on secondment at the Bureau of Meteorology working on the Australian Climate Service (ACS). The ACS is a partnership made up of world leading science, information and expertise from the Bureau of Meteorology, Geoscience Australia, The Commonwealth Scientific and Industrial Research Organisation (CSIRO) and Australian Bureau of Statistics. It brings the Commonwealth’s extensive climate and natural hazard information into a single national view. This will enable the Australian Government to support better planning and preparedness for natural disasters and enable better response and recovery to disasters when they strike.
Matt and I met to talk about his work on the ACS, his experience as a data scientist in the APS and how to build your skills while working on big projects.
Can you describe some of the data gaps and challenges that led to the establishment of the ACS?
The establishment of the ACS came directly out of the Royal Commission into National Natural Disaster Arrangements which was prompted by the Black Summer bushfires. The Commonwealth Government response to these disasters included the launch of the National Recovery and Resilience Agency which has absorbed the existing National Bushfire Recovery Agency and the National Drought and North Queensland Flood Response and Recovery Agency. This new agency, along with Emergency Management Australia is supported with information from the Australian Climate Service.
What are some of the data challenges you are working to address through your role in the ACS work?
A great deal of data sits in different Commonwealth agencies, with data sharing happening largely in the context of projects. Through ACS, a lot of effort is going into improving data accessibility, metadata, harmonisation, and spatially enabling data at scales that are relevant to the questions that are being asked. ACS will also have analytical capability, where stakeholders can turn to for help with particular data needs or problems. As such, challenges aren’t just about access to data it is also about access to expertise and analysis.
You started your career as a scientist at the CSIRO and have continued work as a data scientist in government since. What do you love most about this kind of work?
I’m currently at the Bureau of Meteorology (BoM) on a secondment from the Department of Health. I’ve always been an applied scientist, I like the aspects of trying to figure out how to solve problems with data. My actual training is in geography and geospatial techniques, and I’ve been quite fortunate to work with multidisciplinary teams throughout my career. Being a data scientist allows me to understand different areas that people are coming from and I can come up with different approaches methods to solve problems.
What was your original training in (e.g. undergraduate degree) and how have you built your analytical skills since? What tips do you have for people looking to keep their skills up to date?
Personally, I haven’t found courses as useful as I find on the job training and experiences. However, there are times where undertaking a particular course would be of benefit. I also think that professional networks are where you can learn something useful such as attending data related conferences. For me, I enjoy attending geospatial conferences where I can go and learn about the latest innovations and bring it back to my team.
I also think that secondments and embedding in teams outside of your usual area are valuable. It is especially important to make such opportunities available to junior staff, such as graduates, and buddying them up with a data scientist to accelerate their practical experience.
Lastly, it is worthwhile to incorporate casual reading about what other organisations are doing and using, such as what technologies are out there. For example, if you hear about machine learning and you don’t know what that is, then take a closer look and learn about it. You don't necessarily need to do a deep dive but building the scaffold in your brain where different techniques sit can really help you.
What key advice do you have for data analysts in the APS working on big projects?
It is important to manage expectations from the project teams on what you’re able to do and how much time it will take. You need to pick the right opportunities but also manage the balance between spruiking the work that you can do and ensuring everyone is clear and aware of the limits.
I also think you need to be able to explain some fundamental aspects of the data and methods you use for a given problem. It is important if you’re a specialist and you’re working on a project to really try to spend time talking to people that are submitting the questions, really understand what the project is about and what the project team is hoping you can do.
Using a geospatial example, if you may be asked to use a particular geography, for example postcode, Statistical Areas 2, or Local Government Areas. Those choices matter, not just because of how the map looks, but when you use a particular geography you’re also saying something about the scale of the phenomena that you’re interested in.
Finally, try to be involved with the broader project team. Even if you are only an observer in meetings, this will assist you in being across the communication and to ensure you don’t do your own thing without really understanding the goals of the project.
What's next for your work in the ACS? What will be the anticipated outcomes of your research for the ACS customer agencies?
One area that will continue to take up much of my time involves working with linked data including the ABS MADIP and BLADE datasets. Specifically, I am working on developing new variables that address particular data gaps (e.g., up-to-date vulnerability measures), linking person level data to different hazards (e.g., heatwaves, floods, bushfires) and conducting some analytical projects to address specific data issues (e.g., natural experiments to look at impacts of past events).
You can learn more about the ACS here