Business and the Bureau: working together to unlock the value of advances in forecasting
22 June 2015
New developments in weather forecasting are changing the game for businesses seeking to manage their exposure to the elements. For many, working in partnership with the Bureau of Meteorology offers increasing opportunity to optimise the efficiency of their operations and protect the safety of their workforce.
Forecasting has evolved from a simple ‘single-answer’ prediction (‘it will rain today’), to a more nuanced chance of that weather occurring (‘a 70 per cent chance of rain’). This ‘probabilistic’ (probability-based) capability makes available a new level of information for managing the impact of forecast conditions. It means businesses can now work with the Bureau in new ways as they re-engineer their processes to make the most of the opportunities this offers.
Advances in science means we can now better define weather uncertainty—and work with businesses to help them translate that into better decision-making. This puts more control in the hands of business operators, enabling better risk-management and increased productivity.
The new paradigm—probabilistic forecasting
The standard method of conveying weather information—the traditional, single-answer (or ‘deterministic’) forecast has been superseded. Meteorology has advanced, and new and better information is now available. The forecast no longer expresses a single predicted outcome—instead it attempts to capture the full range of the conditions that might eventuate in the following days.
A traditional weather forecast might take the form of a number (‘the temperature is forecast to reach 28 today’), or it might be a description (‘showers forecast tomorrow’). In both cases, the information has limitations—it doesn’t describe the spectrum of possible outcomes. A single-answer temperature is most useful if the chance of it occurring is 95 per cent, but what if the chance is only 50 per cent? The level of uncertainty about the outcome isn’t conveyed, nor what else could eventuate if the expected situation doesn’t.
We are now using ‘uncertainty’ information to give a fuller picture. Events such as showers and thunderstorms are now given a chance of occurrence in forecasts (Fig. 1). Likely rainfall ranges are also included.
Figure 1. In addition to the deterministic ‘shower or two’, the rainfall forecast includes the percentage chance of rain and likely amount.
Deciphering the data
Some weather phenomena are particularly challenging to forecast. There is greater uncertainty, so conveying this uncertainty becomes even more important. The importance increases for higher-impact weather events.
Consider a tropical cyclone off the northwest coast of Australia. This tropical cyclone has a 50 per cent chance of tracking west from its current position, a 30 per cent chance of tracking southwest, and a 20 per cent chance of tracking northwest. In the ‘single-answer’ world, a forecast would explain that the system was expected to track westwards. However, this forecast would not prepare coastal communities and offshore installations for the approximately one-in-three chance of the cyclone moving towards them.
Our tropical cyclone warnings use a ‘cone of uncertainty’ to present this information—illustrating the expected variation in potential tracks (Fig. 2). The wider the cone, the more uncertain the situation and the greater the potential for variability. If the cone is narrower, there is greater confidence that the system will move nearer the ‘most likely’ path.
Figure 2. Tropical cyclone track showing uncertainty information. The grey-shaded ‘cone of uncertainty’ indicates the spread of likely tracks.
Across the world, weather agencies run high-resolution numerical weather models to help predict the possible track of a cyclone. It is now possible to run these models many times with slightly varying initial conditions, to get a good representative spread of possible outcomes. In tropical cyclone events we can now view upwards of 50 potential tracks at once.
With such a wealth of information available, access to targeted data—and support on its interpretation and application—are vital. Working closely with industry clients, we can identify the most important data and present them clearly and simply. This ensures businesses are able to develop their most effective response strategies.
A world of opportunity
Responding to this increased level of information may initially seem more complex than with a traditional forecast; however, once understood, it provides greater opportunities and increases decision-making confidence.
If a forecast shows a cyclone passing well to the north of a particular offshore installation, the manager has nothing on which to base their assessment of risk to the facility. If uncertainty information is added the manager can see the expected likelihood of impact on the facility—even though it is not near the ‘most likely’ track. He or she can determine an acceptable level of risk and ‘false alarm’ ratio. For example, if a 5 per cent chance of impact is forecast for the location, the risk may be considered unacceptably high—especially if such a level of risk is encountered several times a year, leading to an impact under those conditions once every few years. On the other hand, responding to that risk, which could involve shutting down operations and evacuating staff, can be enormously expensive, with its own inherent risks—and a 5 per cent risk would be a false alarm on the majority of occasions. Risk-management decisions will obviously be highly business-specific. The important point is that those making them now have a significantly higher level of knowledge to inform such key decisions.
Traditional tropical cyclone response plans, based on a ‘single-track’ forecast, generally follow this pattern: ‘if the system is forecast at strength x within distance y of location at time z—then respond’. Probabilistic forecasting adds a new element—and a new dimension: ‘the system has a w per cent chance of reaching strength x, passing within distance y of a location at time z’. This puts more power in the hands of site managers, allowing greater control over setting a threshold of acceptable risk, and therefore greater control over the safety of employees, and the costs of managing the risk.
Broadly speaking, for anyone in a weather-sensitive industry, the more you know about what to expect from the weather conditions, the smarter your response can be and the better you’ll be able to manage your business. Probabilistic forecasts are just the start. We are working towards providing even more information of high value to industry. For example, the probability of exceeding given wind or wave thresholds would assist businesses in dealing more effectively with impacts from a wider variety of weather events. And in the meantime, business and the Bureau must continue to work together to unlock the full economic value of recent advances in the science of meteorology.
Now is a good time for weather-sensitive industries to be having a conversation with the Bureau—how could you better unlock the value of advances in forecasting? Businesses may contact the Bureau at email@example.com.