Long Range Forecasting
When predicting the weather, forecasting 10 days in to the future is considered a long range forecast. When predicting the climate, long range forecasts are generally on a seasonal timescale or longer. Forecasting the climate months and even years ahead is a very complex process.
The success of these seasonal forecasts depends on a detailed knowledge of how the atmosphere and ocean interact. As our understanding of the relevant processes evolved, increasingly complex models have been produced to use the improved measurements and provide improved seasonal forecasts.
Predicting the behaviour of the ENSO over the months and years ahead offers the best prospect of seasonal forecasting for many parts of the world. In other parts, such as Europe and West Africa, regional sea surface temperatures seem to be the important factor.
The ability to have accurate long range forecasts for various climate parameters e.g. about whether rainfall or temperature will be above or below average, and by how much, would have enormous potential benefits. It is these forecasts that could provide the information needed for longer-term decisions and early warnings of potential hazards. However the credibility of any forecast depends on their track record to accurately predict the climate. Climate predictions will only be accepted if they are more accurate or provide more information than the climatological experience that is currently used to make decisions.
The simplest way to forecast deviations from the normal climate, months to years ahead, is to work out rules, using statistics, linking future patterns to current climate features.
Large-scale, slowly varying climatic anomalies, such as in sea surface temperature that can persist for many months, may force changes in atmospheric circulation patterns and hence departures from normal for local climate cycles.
At first, this approach was not a great success, but now the growing understanding of what drives ENSO and other forcing patterns has made the method more reliable. Empirical methods using, for instance, sea surface temperatures assume that local climate will be affected in roughly the same way each time there is a similar large-scale forcing. Their advantage is that they are relatively easy to apply, as they rely entirely on climate statistics and use modest computer resources. There are, however, limitations. The statistical models generally attempt to predict complex interactions without any specific links to the underlying physical and dynamical processes. This means that they work best when large-scale developments are well and truly under way, but they have difficulty anticipating shifts from, say, warming to cooling or vice versa. As a consequence, they often miss sudden developments.
A more physically-based approach to seasonal forecasts uses computerized general circulation models (GCMs). In one form of this approach, the first step is to predict the development of sea surface temperatures in the tropical Pacific. The predictions may be based on a regional model that considers developments in the tropical ocean in isolation.
Once this model has made forecasts of how the Pacific may behave up to a year ahead, the forecast sea surface temperature patterns are used to drive an atmospheric general circulation model to predict how the weather around the globe will respond. These predictions of seasonal weather have produced promising results, especially in the tropics. The improvement of these forecasts has been built on both advances in the models and better observations from the equatorial Pacific Ocean. Significant developments are now being made to develop fully coupled systems in which the ocean, atmosphere and land surface components of the model continually interact with each other to produce a forecast up to several months ahead.
Making the forecasts relevant
A challenge to forecasters is to ensure that the forecasts are both timely and understandable to the potential users. Since the most successful forecast methods to date have been mostly associated with ENSO events, it is the lesser developed tropical countries that stand to gain the most from these developments. It is important, then, that the predictions must be presented in forms that farmers or fishermen can use; useful predictions cannot rely on computer graphics or statistical arguments, but on statements that can be broadcast over the radio or published in newspapers. In addition, they must be available in time for people to make decisions about what to sow, and in a form that is relevant to making such an important decision. This involves understanding local agricultural practice and variations in climate, and integrating both into a methodology for better decision-making.
Who provides the long-range Forecasts?
The process of computing long-range forecasts (forecast ranging from 30 days up to two years) on the global scale requires huge amounts of computer power along with a very specialized knowledge. For this reason, there are only a few centres around the world that are producing global climate long-range forecasts. The service provided by these centres, known as Global Producing Centres for Long Range Forecasts (GPCs) sets the frame, or context essential for predicting climate and weather on regional and local scales and is used by regional and local forecasting centres.
Long range forecasts on the regional scale are produced by both Regional Climate Centres (RCCs) as well as Regional Climate Outlook Forums (RCOFs). Similar to GPCs, RCCs and RCOFs use the data supplied by the NMHSs to develop detailed forecasts relevant to their region. These models and forecasts are then used by the NMHSs to produce better and more accurate national and local forecasts.
[More in depth information] Regional Climate Centre forecasting products can be found here under the heading “Designated RCCs and Pilots”
[More in depth information] Regional Climate Outlook Forum forecasting products can be found here under each regions webpage.
|© World Meteorological Organization, 7bis, avenue de la Paix, Case postale No. 2300, CH-1211 Geneva 2, Switzerland - Tel.: +41(0)22 730 81 11 - Fax: +41(0)22 730 81 81|