«Medium-Range Weather Prediction Austin Woods Medium-Range Weather Prediction The European Approach The story of the European Centre for Medium-Range ...»
Relations between the Council and the Director have been good. The Council has been patient when the going was tough, especially when results were slow in coming from the research programme. It has been clear that Council delegates have been proud of the Centre’s achievements. The Centre is generally seen as an indispensable part of the meteorological scene in Europe. By now, some thousands of European meteorologists have worked at the Centre, or attended training courses or seminars, or visited.
Many more have used the Centre’s data in their research, or have used its computing facilities.
It has been remarked that public awareness of the Centre is greater outside Europe than at home, even though it was first set up under COST, an institution of the EEC, now the European Union (EU). In spite of its origins, the Centre has at the time of writing little regular contact with the EU, apart of course from its research programmes. It would be mutually advantageous to
238 Chapter 20develop such contact. Both Director and Council now agree that the profile of the Centre should be raised. The funds for the Centre come from the taxpayers of Europe; there should be awareness of how these funds are being expended and the results that are achieved.
The Centre will continue to be active in developing European and worldwide collaboration in the atmospheric and related sciences. Some funding for many important aspects of the Centre’s work comes from space agencies and the European Union research funds, as well as from national sources.
The Centre has supported many field experiments. It has met space agency requirements for engineering, calibration and validation of data coming from new satellite instruments.
The research community worldwide has been using the Centre’s output freely, easily and extensively. However, the extent of operational use of the Centre’s output in the Member States has not been as great as hoped by some. Some think that the Centre’s data policy in this respect has been somewhat too restrictive, although understandable perhaps when seen in the context of guaranteeing the benefits of membership to those who fund the Centre. The Advisory Committee on Data Policy will continue its efforts to encourage more use of the forecasts.
The Centre is now, in 2005, at a turning point in its history. Although there has been a continuous exchange of personnel between the Centre and its Member States, a small team of scientists and managers stayed on from the 1970s. These include David Burridge, Tony Hollingsworth, Adrian Simmons and Horst Böttger, who played significant roles in the build-up phase and during its first three decades. These have left or will leave soon.
Turnover of staff has nevertheless been one of the Centre’s strengths.
With the amended Convention, noted in Chapter 5, the Centre’s activities will expand. The amended Convention will play a part in ensuring the continued success of the Centre. New States will join, contributing additional fresh scientific talent to the Centre’s team, as well as easing the financial burden on the existing States. New Member States also will have an influence on the direction the Centre will take in the future. The basics however remain. Undoubtedly the focus of the Centre will remain on improving the quality of the medium-range forecasts.
While it is easier to maintain a feeling of enthusiasm in creation than in consolidation, the Centre’s atmosphere and working environment has remained exciting and challenging. New complexities continue to emerge for the development of medium-range prediction. In addition, new demands for environmental monitoring and longer-range atmospheric and oceanic prediction will continue to arise.
And the outlook is... 239 However, the financial outlook is cause for concern. The price of the ever more powerful computers needed to address these important issues is anticipated to rise, and maybe rather steeply. Even ten years ago, the expenditure on pensions was nominal. Now, as more staff retire, pension costs are becoming significant. Council decided on a new, and in the short term more expensive, pension system in 2002. Also, buildings are getting older and need refurbishment, and a new building programme began in 2004. Council in the coming years will have to be able to find significant amounts of new funding just to maintain the present level of the Centre.
The demand for weather information will increase and the need to reduce the risk of weather-related damages will grow. Probabilistic information from Ensemble Prediction Systems has already been used to extract quantitative early warning signals of high-impact weather. Also dynamical and statistical techniques have been applied to obtain weather and weather risk information at the smaller scale and for single locations.
Europe currently lacks operational capabilities to provide adequate warnings of widespread severe weather in the coming season. Fifteen thousand excess deaths were recorded in the heat wave of summer 2003.
Although forecasts of this heat wave in the days leading up to it were good, medium-range warnings for such natural disasters three to seven days in advance and short-range forecasts up to three days ahead need to be further improved. In the coming years, new developments are expected to increase the synergy between the ECMWF global deterministic and probabilistic forecasting systems and the regional, higher-resolution forecasting and application systems run operationally at national and regional levels. New applications will be developed to increase further the use of ECMWF forecasts in different sectors, including health management, agriculture, energy, hydrology and water management.
As the Centre begins to consider its strategy for the coming decade, severe weather prediction is already stressed. Development work resulted in
new forecast products for severe weather prediction based on post-processing forecasts from the Ensemble Prediction System (EPS), including:
• an Extreme Forecast Index,
• tropical cyclone tracks and strike probabilities,
• tropical cyclone frequency in seasons,
• wind gusts and heavy precipitation probabilities, and
• maximum wave and freak wave forecasts.
National Meteorological Services of the Member States, Co-operating States and WMO use the Centre’s products widely for their official duties,
including issuing early warnings and alerts for civil protection, such as:
240 Chapter 20
• storm surge prediction,
• flood forecasting,
• wave forecasting, and
• air trajectory computation, for predicting transport of pollution.
Most severe weather events are limited to geographically small areas, or are caused by small-scale features embedded in larger-scale weather systems. It is clear that resolution is important in predicting their occurrence and intensity. Our ability to forecast severe weather is partly limited by the inherent unpredictability of the phenomena in question, and partly by the skill in predicting the large-scale patterns with which they are associated.
The scales of atmospheric weather systems that can reasonably be described by a numerical model are in fact many times larger than the nominal grid separation. A 100 km grid is capable of describing accurately the dynamical circulation of a weather system whose scale is about 800 to 1,000 km or larger. Smaller-scale phenomena fall, so to speak, between and through the grid-points. Thus, typical climate models with 300 to 400 km grids cannot represent many sub-synoptic scale systems at all, and only poorly represent many features such as storm tracks. Resolutions of 15 to 30 km will improve the description of important structures within active synoptic weather systems. In particular, they will capture better the true intensity of the highly energetic systems associated with severe weather events.
At the time of writing, the global analysis and prediction system at ECMWF has the highest resolution of any such system in operational use.
However, other major forecasting centres plan to run similar or higher resolution models in the next few years: the Met Office in the UK and the Canadian Meteorological Center with 40 km, Japan Meteorological Agency with 20 km, and the German Weather Service (DWD) with 20 km. ECMWF resolution will reach 25 km in 2005. If the Centre is to conserve its leading position, major efforts will be required.
There is international interest in the possibility of designing global climate models with resolutions of the order of 1 to 3 km. Very limited experimentation has been performed. There are many research issues to be addressed. The aim would be to reduce the number of sub-grid physical processes that need to be parametrized, so decreasing the uncertainties and errors in the models. The computational costs are truly vast. At the time of writing, a global version of the Integrated Forecast System (IFS) with a 2.5 km grid would require a full day of computation on the current IBM mainframe to provide a one-day forecast! It will probably be decades before operational global NWP can consider such an approach. The experiences of And the outlook is... 241 both climate research into high-resolution modelling, and regional NWP, which will soon be using such resolutions over small areas, will influence grid refinement in global NWP.
Substantial improvement in the quality of analyses and short-range forecasts has been achieved by improved modelling and data assimilation techniques, and from the improved observations, especially the spacebased component of the global observing system. It is now generally acknowledged that the long-term effort to develop, build and refine the 4D-Var system has been a good investment. The system is built on a firm theoretical basis. Extensions will be implemented to meet future requirements. The 4D-Var technique gives the flexibility required to deal with a wide variety of observations. It can extract information from data that are only indirectly linked with the model or analysis variables. Data assimilation system of the future will need to take full advantage of the information obtained from diversifying space-based observing system technologies, in terms of meteorological quantities, greenhouse gases, aerosols and airborne chemicals.
We have noted that the ECMWF variational data assimilation system is an ideal tool for determining the uncertainty in the atmospheric analysis. It is ideal also for directing the deployment of “targeted observations”, for example sending unmanned aircraft to collect data form areas crucial for the future development of a storm. Operational targeting has been implemented for several years in winter over the Pacific, and even longer over the Caribbean basin during the hurricane season. Over the Atlantic, the potential to target storm tracks was assessed in 1997 during the FASTEX field experiment, and in 2003 during the north Atlantic THORPEX Regional Campaign. The European Composite Observing System Programme of EUMETNET was at the time of writing developing a concept of operational targeting for the Atlantic Basin. ECMWF is well placed to contribute to these activities.
While the analyses produced through data assimilation serve their primary function as initial conditions for deterministic forecasts, they also provide a long-term record of the atmosphere and climate. Furthermore, environmental monitoring for the global Earth system is becoming increasingly important. Thus, different requirements are imposed on the design of the future data assimilation system, partly overlapping and partly conflicting: higher resolution, reliable estimation of analysis uncertainty, longer assimilation windows, increased number of analysed fields, and coupling to ocean and land-surface analyses.
Parametrized physics will remain an important aspect of the Centre’s IFS for the foreseeable future. With increased resolution, orography will be
242 Chapter 20better represented. While some mesoscale convective systems will be resolved, parametrization will still be necessary. In fact, the requirements for parametrization will be even more demanding as there is a gradual transition from “parametrized” to “resolved” processes. The behaviour of parametrization at a variety of resolutions is of particular importance to the ECMWF environment. The model is applied with a wide range of resolutions, from seasonal forecasting at low resolution to the deterministic forecast and “outer loop” data assimilation at the highest resolution.
New research elsewhere will be followed closely; promising developments worldwide will be evaluated in the context of the Centre’s requirements. Experience has shown that the link between research and implementation in large-scale models is by no means trivial. For example, studies of entrainment in shallow convection, and of diffusion in stable boundary layers, have suggested rather different parameter settings to those used in large-scale models. Often some aspects of model performance deteriorate, due to compensating errors, after making an improvement to a part of the model.
There is an increased demand for good quality precipitation forecasts, for example for predicting severe weather, and for hydroelectricity generation.
Further development of the cloud and convection schemes and optimisation of these schemes in their interaction with the model dynamics will be needed to meet this demand. The use of sub-grid variability of moisture as a new model variable is central in this line of research. Work on the moist physics will go hand in hand with work on assimilation of precipitation and clouds.
Over the years, the number of applications of the ECMWF system has increased: ensemble forecasting, ocean wave modelling, seasonal forecasting and ozone chemistry. Although these applications put emphasis on different aspects of the model, they also provide a multi-dimensional constraint on the system and give information on model problems from a different perspective. With GEMS — see below — even more information will become available, for example on convective and turbulent transport, through modelling and verification of aerosols, trace gases and other chemical components.