Skip to main content
Time to race start
00 days
:
00 hrs
:
00 min
:
00 sec

The other race: to understand the weather

Home 2025 The other race: to understand the weather

Article Image

Every Boxing Day, as the fleet lines up on Sydney Harbour for the Rolex Sydney Hobart start, and thousands of eyes turn south toward Bass Strait, another race is already well underway — the race to understand the weather.

news
—

For the Bureau of Meteorology (BOM), forecasting the conditions for the Rolex Sydney Hobart is not about predicting a single outcome, but about mapping a range of possibilities, narrowing uncertainty and identifying risks long before the starting gun fires. 

The process begins well before the yachts leave their moorings. Around nine to 10 days before the race, forecasters begin building a long-range outlook, drawing on multiple global weather models to understand what large-scale systems might shape the race. 

“It’s about identifying the broad pattern first,” said Edward Townsend-Medlock, a meteorologist involved in the BOMs long-range outlooks. “At that range, we’re not forecasting exact wind speeds at a specific location, but looking at which major weather features are most likely to be present on race day and into day two.” 

Two types of models underpin that early analysis. Deterministic forecasts — often described as a ‘most likely’ — are highly reliable in the short term and can capture broad-scale features out to about 10 days. Beyond that, their accuracy drops, particularly when it comes to timing. 

Alongside them sit ensemble forecasts, which run hundreds of lower-resolution simulations, each with slightly different starting conditions. The result is a spread of possible scenarios rather than a single outcome. 

“Ensembles are especially valuable because they highlight the outliers,” Townsend-Medlock said. “Those less likely but high-impact scenarios can be critically important when you’re dealing with severe weather.” 

For the Sydney Hobart, deterministic models such as the European ECMWF, Australia’s ACCESS model and the US-based GFS are used in combination with ensemble data and historical experience. Past races with similar atmospheric setups provide valuable context — not as templates, but as guides to how conditions might evolve. 

As race day approaches, the forecasting effort intensifies and expands. Five to seven days out, the work is typically handled by one or two specialist teams, including national production and tropical cyclone forecasters. But inside four days, the circle widens. 

Marine forecasters step in with greater detail on wind, seas and swell. Thunderstorm specialists assess convective risk. Fire weather teams may become involved if heat, wind and dryness combine over land. 

In the final two days, if hazardous conditions are expected, coordination ramps up again. Rainfall and water teams may assess flood risk, while tropical cyclone staff increase coverage if a system threatens coastal waters. 

“The closer you get, the more specialised the focus becomes,” Townsend-Medlock said. “It’s a layered process, with more expertise added as the forecast sharpens.” 

Forecasting, however, is never just about prediction — it is also about accountability. Every forecast is later checked against what actually happened, in a process known as forecast verification. 

Observations from weather stations, ocean buoys, satellites and other data sources are compared with predictions to measure accuracy. The results feed back into model development and statistical techniques designed to improve future forecasts. 

“It’s a core part of modern meteorology,” Townsend-Medlock said. “Verification helps us understand where forecasts perform well and where they can be improved. It’s also an active area of research.” 

The evolution of forecasting over recent decades has been driven largely by technology — particularly satellites and computing power. 

Today’s forecasts rely on two computationally demanding steps: ingesting vast quantities of observational data to define the current state of the atmosphere and then calculating how that state will evolve over the coming days. 

Satellites, first introduced in the 1960s and 1970s, now provide most of that data. Improvements in resolution and observation frequency have dramatically increased the accuracy with which forecasters can initialise models, especially over oceans where traditional observations are sparse. 

On the computing side, modern global forecasts are run on supercomputers, but with strict time limits. To be operationally useful, a forecast must typically be completed within about six hours, constraining how fine-grained the model can be. 

Even so, today’s deterministic global models run at resolutions of roughly nine to 12 kilometres — a dramatic improvement on earlier generations. Much of the gain in forecast skill has come from increased computing power, allowing atmospheric processes to be represented more realistically. 

For sailors, the forecast is not a script, but a strategic tool. For forecasters, it is a living process — refined, tested and re-examined right up to the moment the fleet disappears down the coast. 

As the Sydney Hobart has shown time and again, the weather will always have the final say. The job of forecasting is simply to ensure there are as few surprises as possible. 

Steve Dettre/RSHYR media