The European Centre for Medium-Range Weather Forecasts is an international organization supported by dozens of countries. Its headquarters are based in Reading, UK, and its high-performance computer is in the top 40 of the Top500 list, and one of the largest computing centers in Europe.
It is astonishing when looking back over the past 37 years since the European Centre for Medium-Range Weather Forecasts (ECMWF) was founded in 1975, to see the progress in the science and practice of numerical weather prediction (NWP). In 1975, global NWP models were in their infancy and weather forecast ability was limited to about three days ahead at most.
A key reason why ECMWF was established was to enable global NWP to advance more rapidly by creating a European collective effort. By any stretch of the imagination this period has been one of huge progress in this enterprise and today we routinely expect weather forecasts to have the ability to look ahead to the second week. Scientific developments, enhanced observational coverage, and increased computational capability have all played a critical role.
What does the future hold and what could we expect weather forecasts to be like in, say, 2030? It is notoriously difficult to foresee the scientific and technological future not least because advances in technology over the next 18 years, if the recent past is anything to go by, are essentially unimaginable. But, perhaps other current trends are more straightforward to extrapolate.
The implied horizontal mesh size of the ECMWF global forecast model (today 16 km - 9.9 miles - for the high-resolution model) has been reducing at a reasonably steady exponential rate for several decades. The objective capability of the NWP forecasts have been increasing at a constant rate (for a useful forecast) at about one day lead time per decade. It may be dangerous to extrapolate these two trends forward but if we do, then by 2030 the model's ability should have extended by about two days further into the future, and horizontal mesh sizes may be in the region of a few kilometres.
Another natural question to ask is what will the global models of the future be able to predict? It is interesting to remind ourselves that since 1992, ECMWF has not only been predicting the weather but also the ocean waves. Of course there is an intimate connection between the near-surface winds and the waves, but also the need for mariners to have good forecasts.
More recently, because we have been extending our forecasts to monthly and seasonal timescales, the ECMWF forecasting model now includes a model of the global oceans coupled to the atmospheric model.
Also, ECMWF have been developing two other areas where related aspects of the natural environment can be predicted using our forecasting systems and data. The first is the MACC project, which predicts atmospheric composition, including greenhouse gases, aerosols, fires, and air quality. The second is ECMWF's first third party activity - the European Flood Awareness System - where the assessment and prediction of catchment-scale hydrology is being explored.
These are, from a scientific, technical, and user viewpoint, very exciting initiatives. One can speculate that the NWP system of the future may be closer to a numerical environmental prediction system. These developments are happening because the science is advancing in these areas, but also because new observations of these properties are available from satellites and elsewhere.
Of course, the science needed is multi-disciplinary with physics, chemistry, and biology all playing an increasing and important role. Techniques like data assimilation, which have their origins in meteorology, can and, are being extended into many other branches of environmental science. There are many uncertainties about the future, but the opportunities to advance the science of NWP and improve forecast skill are there for ECMWF to grasp in its goal to continue to be the acknowledged world-leader in global medium-range prediction.
A version of this story first appeared in the ECMWF Newsletter No. 131 - Spring 2012.
