Seasonal outlooks are often made with reference to above or below median rainfall - i.e., what is the chance of rainfall being above normal or below normal. Many users are more interested in specific rainfall amounts (e.g., 200 mm for the season), or are prepared to act at specific probabilities (e.g., if there is a 75% chance it will be dry).
The new maps result from statistical techniques which transform the outputs from the official seasonal outlook (Drosdowsky and Chambers; 1998, 2001) into information in two forms: the chances that the three-month rainfall will exceed particular thresholds (e.g., 200 mm), and the three-month rainfall amounts that have a specific chance of occuring (e.g., 25%).
Verification has shown that these statistical techniques do not diminish the accuracy (i.e., how well the outlooks performed for above or below median rainfall) or the reliability (how the probabilities produced by the model compared to the frequency of the outcomes) of the forecasts. To get the best results out of any seasonal outlook system, it is important to consider the accuracy of the model at the time of year that is of interest.
The seasonal outlook data is prepared on a 1°x1° grid, or roughly 100 km by 100 km.
It has long been known that Australia's climate is driven, on seasonal time scales, by the heat stored in the global oceans. The Bureau's official seasonal climate outlook model uses the relationship between the major patterns of variability in the sea surface temperatures that represent both the Pacific (e.g., El Niño and La Niña) and Indian (e.g., the Indian Ocean Dipole) Oceans over the years 1950-1999, and relates these to the major patterns of variability in Australia's rainfall over the same period. When the current states of the ocean patterns are fed into the model, it uses these past relationships to make a forecast for the liklihood of rainfall for the coming season.
The model maps and text are typically updated in the last week of each month. The outlook that is issued is for the following three months. For example, the outlook for winter (June to August) will be issued at the end of May.
Outlooks for specific areas and locations are presented in three formats, as well as the standard above and below median outlooks.
The Outlook scenarios convert the seasonal outlook into the rainfall amounts which have a 75%, 50% and 25% chance of occurring. In other words, on the map below, the location highlighted with the black dot has a 75% chance of at least 100 mm (and possibly up to 200 mm) falling.
Values may at first appear to be conservative for high probabilities such as 75%, however this is to be expected. If the climate is not being pushed strongly one way or another (i.e., if there is no El Niño or La Niña event), then the values with a 75% chance of occurring will be close to the 25th percentile, which by definition means that 75% of past years will have had more rainfall for the season, and only 25% less rainfall. If this were not the case, then the model would be unreliable; the chance of a certain rainfall amount occurring would not be accurately reflected by the output of the model.
The Chance of at least rainfall outlooks present the seasonal outlooks model results in a different way. With these maps, 12 separate rainfall totals for the coming season are listed, and the chance of receiving that amount can be read off the map. For instance, in the following example, the highlighted location has at least a 65% to 75% chance of receiving at least 150 mm in the next season. This matches with the Outlook scenarios maps, which suggested a 75% chance of at least 100 mm, possibly as high as 200 mm.
The Table versions of the seasonal outlook provide detailed information for over 260 specific locations and towns across Australia.
The tables list both amounts that relate to the specific probabilities (i.e., the rainfall amounts that have a 75%, 50% or 25% chance of occurring over the coming season), and the 12 rainfall amounts and their associated chance of occurring. For example, the following table lists just the Australian capital cities and the associated chance (as a percentage) that they will receive certain amounts of rainfall. It is immediately apparent that Darwin has a very low chance of receiving substantial rain, while Perth should be the wettest capital for this particular season (July to September).
Stations used in the tables were originally selected in accordance with three main principles.
The development of new high resolution datasets means that we are now able to add locations not previously available. Proposals for new locations, including latitude and longitude, may be sent via the Water and the Land feedback page.
|Location||10 mm||25 mm||50 mm||100 mm||150 mm||200 mm||250 mm||300 mm||400 mm||500 mm||600 mm||700 mm|
Drosdowsky, W., and Chambers, L.E., (1998), Near global sea surface temperature anomalies as predictors of Australian seasonal rainfall. BMRC Research Report No. 65.
Drosdowsky, W., Chambers, L.E. 2001. Near global sea surface temperature anomalies as predictors of Australian seasonal rainfall, J. Climate 14:1677-1687.