Process of forecasting
Meat & Livestock Australia October 24, 2023
The process of forecasting the short-term weather and the longer-term climate is both an art and a science. In this section, we’ll introduce you to the concept of forecasting and help you understand some of the factors that affect it, such as climate drivers.
The climate decision gamble – is it like chess, poker or the pokies?
Australia has a highly variable climate which makes farming decisions difficult. If we knew with certainty that our desired rainfalls and temperatures would occur when we needed them, we could maximise our opportunities and minimise our risks. Often, we think of climate forecasts as having a random chance of happening, like when we play the pokies. We feed our money into the machine and have no idea whether we’ll win or lose.
Farming is a risky enterprise, but unlike the pokies the decisions involve degrees of prediction and judgement. So, farming is more like a chess or poker game. In chess the only hidden information is the strategy and supporting tactical moves in the players’ heads however, both players see all the chess pieces and their positions on the board. There is no rolling of the dice to make one of the pieces disappear from the board. Losing a chess game is based on poor tactical moves, not bad luck. Poker on the other hand, is a game of incomplete information – this is decision making under uncertainty. Losing a hand of poker may involve bad luck, but the game involves a high level of skill, as evidenced by the world champion poker players. When we approach our farming decisions, we should be thinking like these pokers players.
The first step in making a good farm business decision is to acknowledge that we are dealing with a level of uncertainty. Champion poker players are continually considering probabilities and recognise each decision has a range of possible outcomes; some of which are more likely than others and some carrying more consequence than others. The farming game involves a bit of chess and a bit of poker. We need to gather the best information available to us at the time and think through the possible outcomes.
Key message
- Australia has a highly variable climate
- Making climate-risk decisions in farming is like playing poker – it is decision-making under uncertainty
Weather versus seasonal climate forecasts
While weather forecasts and climate forecasts are both types of predictions related to atmospheric conditions, they differ in several ways.
Weather forecasts are short-term predictions of atmospheric conditions, usually covering a period of a few hours to several days. The Bureau uses a weather prediction model that runs out to seven days and has a resolution of one to two kilometres. These forecasts provide information about temperature, rainfall, wind, humidity and other weather parameters. Weather forecasts are based on current weather conditions, as well as computer models that stimulate how these conditions are likely to change over time. Weather forecasts are constantly updated as new information becomes available, allowing us to make decisions about our daily farming activities.
In contrast, seasonal climate forecasts are longer-term predictions of the average weather conditions for a particular region over a period of weeks to several months. The Bureau uses a physics-based dynamical climate model that runs out to six months and has a 25km resolution. These forecasts provide information about the likely trends in temperature, rainfall, and other weather parameters, based on historical data and computer models that stimulate how the climate system responds to changes in atmospheric conditions. We can use climate forecasts to plan for longer-term activities (such as which crops to plant) and to prepare for potential disasters.
Another key difference between weather and seasonal climate forecasts is their level of uncertainty. Weather forecasts are typically more precise than climate forecasts because they are based on current conditions and predicting the weather in the coming days. By contrast, climate forecasts are based on long-term trends and can be affected by a wide range of factors, such as changes in ocean temperatures, and atmospheric pressure. As a result, climate forecasts are generally less certain than weather forecasts, and are often presented as a range of possible outcomes rather than a single prediction.
The Bureau has recently introduced multi-week forecasts, which are more like short term climate forecasts than long term weather forecasts. They can predict the weather more accurately due to advances in modelling incorporating overarching oceanic climate forces more so than small changes in the weather. The five tools we introduce in the next section uses this advanced methodology.
In summary, weather forecasts are short-term predictions of atmospheric conditions providing information about daily weather patterns that can be used for planning operational activities. On the other hand, climate forecasts are long-term predictions of average weather patterns over a longer period, used for strategic planning.
Key message
- Weather forecasts are short-term predictions, usually covering a period of a few hours to several days.
- Seasonal climate forecasts are longer-term predictions, covering a period of weeks to several months.
Forecast versus hindcast
A forecast is a prediction of weather conditions, for a specific time and place, in the near future. It is usually based on current weather observations and the use of computer models to simulate how the atmosphere is expected to evolve over the next few hours to several days. Weather forecasts are important for planning daily farming activities and making operational decisions about what to do. They are updated as new meteorological data becomes available and they become more accurate as the timeframe for the forecast becomes shorter.
Hindcast, on the other hand, considers what the weather was like in the past, usually for a period ranging from a few days to several decades. Hindcasts are used to evaluate the accuracy and reliability of weather and climate models by comparing their simulations to actual weather observations from the past. Hindcasting involves running a computer model backwards in time, using data (such as historical weather observations, atmospheric conditions, and ocean temperatures) to simulate what the weather would have been like for a particular time and place. As a result, hindcasts are useful for improving weather and climate models, verifying past weather events, and understanding how the climate has changed over time.
Key message
- A forecast is a prediction of weather conditions, for a specific time and place, in the near future.
- Hindcast considers what the weather was like in the past, usually for a period ranging from a few days to several decades.