Guidance on the use and interpretation of our long-range predictions
Long-range predictions are unlike weather forecasts for the next few days. The nature of our atmosphere means it is not possible to predict the weather on a particular day months to years ahead. At this range we have to acknowledge that many outcomes remain possible, even though only one can eventually happen. Over the course of a whole season, year or decade, however, factors in the global weather system may act to make some outcomes more likely than others. It is because of this we can make a prediction , but we still need to show that a spread of outcomes is possible. To do this, we present the likelihood of each outcome by using a range or percentages. As a result, the forecast is useful to assess likelihood and risk, but not for warning of definite events.
The characteristics of long-range predictions affect the way we determine their accuracy. It is not possible to say that a single prediction is 'right' or 'wrong', since a wide range of outcomes are indicated (although some have greater likelihood than others). Instead, accuracy is measured using a number of predictions to find how often the likeliest outcomes match what actually happened. Long-range forecasting systems typically use a large set of predictions of past cases to provide this assessment of skill.
As there is always a chance of one of the less-likely outcomes in the forecast happening, decisions based on a single forecast could be wrong. The full benefit of any long-range prediction is realised if they are used to guide decision-making over a longer period of time.
Long-range predictions are for the average conditions over the specified period. They don't indicate that these conditions will prevail continuously, as the period is likely to contain a range of types of weather. The prediction also refers to conditions averaged over the specified region but it is possible particular locations will experience different outcomes, especially in regions with highly variable terrain.
Category predictions give the chances of experiencing each possible type of conditions (for example 'dry', 'near-average' and 'wet' for rainfall). We use three or five categories in our forecasts. Taking the example of a three-category forecast, each category is defined as having occurred exactly a third of the time in the past. Thus, when the prediction shows a category with a percentage greater than one-third (33%), those conditions are more likely than normal to occur. Conversely, where a category has a percentage of less than 33%, these conditions are less likely than normal.
A potential pitfall in interpreting these predictions is to give undue attention to the most likely category. For example, if the percentages are 50% for wet, 30% for near average and 20% for dry, this shows a higher-than-average chance that conditions will be wet and a lower-than-average chance that conditions will be dry. It would be incorrect to describe this example as a forecast for wet conditions, however, as there still remains a considerable chance of conditions other than wet (50% = 30% for average plus 20% for dry).
Care is also needed in assessing the risks of possible extreme weather associated with the broad categories used in the forecast. A high percentage for dry conditions, for example, would indeed indicate a higher-than-normal chance of drought. But if the chance of drought in a normal year is small, it may still be fairly unlikely. In this situation it would be wrong to make full-scale preparations for drought as though it were almost a certainty.