SCIENCE GROUP

An Introduction to the Met Office

Peter Kaminski

Met Office

28 January 2005

Peter Kaminski has worked for 30 years in many departments of the Meteorological Office, firstly in Bracknell and then in Exeter following the Met Office relocation in 2003. He is now the manager of Observing Methods in Development. When Mr. Kaminski spoke the Met Office was preparing for a seminar on climate change to be held at Exeter 1–3 February, and to be attended by climatologists and senior officials from around the world.

Nowadays, we take the weather forecast given out regularly on television and radio for granted. Supermarkets use forecasts to predict the sales of cat litter and barbecues to ensure they keep the shelves filled with the products that will sell. The Met Office has become an integral part of our way of life.

The Met Office is considered by many around the world as the best in the area of weather forecasting. It was founded in 1854 as part of the Board of Trade, but then in 1919 it transferred to the Air Ministry, and finally in 1964 to the Ministry of Defence. In 1990 it acquired the status of Executive Agency, and then in 1996 became a Trading Fund. In 2003 it moved to Exeter, to purpose built buildings where there was better communication between the various departments, and with the latest in fast computers. The present Chief Executive, Dr David Rogers, took over in July 2004.

Out of a total of just over 2000 staff, about 1400 are scientific. There are about 1850 permanent staff, of which about 35% are in forecasting, 18% are in research, 13% development of techniques and methods, 10% sales and marketing and 7% in IT and service delivery. It has revenues of £161 million from all sources.

They are a leading member of the World Meteorological Organisation (WMO) and one their regional hubs for their communications network.

With some of the most powerful computers in Europe, they can give as accurate a forecast today for 3 days ahead, as was possible one day ahead 20 years ago. We have come a long way from the first numerical forecast made by mathematicians with slide rules that took 3 months and was wrong! The Met Office now can provide forecasting for the whole globe, and more detailed forecasts for particular clients, especially locally in the UK. The global analysis includes forecasts of hurricanes and other tropical storms.

The data for the model is provided by a network of stations around the world, supplemented by aircraft measurements for interesting situations, like large storms over the sea where there are no measurements being made either on the surface or in the air.

Their services include mobile units with the UK military, which have been active in Afghanistan and Iraq, providing up to date bulletins for the local conditions. They also provide disaster mitigation alerts, as in Mozambique in 2000, when disastrous floods hit the country. Globally, they provide reports for the aviation industry at heights over 24,000 ft, where the US service supplies the equivalent information up to that height. The US and UK services can replace the other in cases where either go offline for any reason.

After the bad publicity following the 1987 hurricane in the UK, the Met Office developed the severe weather warnings; these have now become standard features on the television forecasts when a severe weather event is likely to happen. These will give an indication of excessive snow, ice, rain, wind etc. to allow everyone to take the appropriate precautions for their interests. Services are provided to forecast storm tides (e.g. The Thames Barrier). Recently, they have added pollution and hay fever predictions.

Although they forecast only the weather, and can estimate how much rain will fall, they have not traditionally been involved with modelling the ground to determine what will happen to the water after it has fallen. The Met Office is working with the Environmental Agency to help develop suitable models so that better flood predictions can be made.

New services are also being developed for specific cases, like the Health Service. Bed usage has been shown to follow exceptional weather in a fairly predictable manner, and so if hospitals are given these warnings, they can schedule their use of their operating theatres and beds accordingly. For instance, there is evidence that incidences of heart attacks peak 2 days after a severe cold snap and strokes about 5 days after.

The models in use for forecasting are based on grid points at 50km intervals over the whole globe, and cover 35 levels in the atmosphere. Over the UK, the resolution can be finer. The models are run twice a day. At the noon run, the data is collected between approx 10:30 and 13:30, and undergoes a quality check analysis at about 13:30. Other data (e.g. satellite pictures) is used to verify how the initial state of the atmosphere compares with what the model is showing. At 13:45 the model is run and takes about one hour's computing time, with predictions being produced for each 12 hours for up to 5 days ahead. The Met Office Global model output is compared with other model output, especially across Europe. An especially valuable source of comparison is the forecast generated by the European Centre for Medium-Range Weather Forecasts (ECMWF) located in Reading. ECMWF perturbs the initial data and runs 51 separate forecasts for up to 10 days ahead; this allows the forecasters to look at the possible different forecast scenarios.

In addition to the forecast models, climate models are also being developed to try to see what will happen to the climate over long periods, for example up to 100 years. A reasonable match, using natural factors only, can be made up to about 1950, when the measured average global temperature starts to rise above the model's predictions. But when man-made factors are also introduced, the latter rise can also be accounted for. Using this model, with four different estimates of the production of carbon dioxide, shows that in the year 2100, the average temperature may be between 2º and 5º Celsius higher than in 1990. By the 2080s, we would have significantly warmer wetter winters and hotter drier summers; the 20% increase in precipitation in the winter will probably be accompanied by an increase in storminess; and the 40% decrease in summer leading to one summer in two being very dry.

The audience raised questions about the reliability of the data and models, and where they might be improved. The data stations are rather far apart, and the intermediate grid points are interpolated between them. Although more stations might improve the data, an automatic station is in fact expensive to set up and maintain, and there are no plans to significantly change the number now present. The models would benefit if the number of levels in the atmosphere could be increased, although these measurements are very hard to do. Some aircraft carry observing instruments; this data is transmitted to Exeter and input to the model.

The effect of the sea on the atmosphere is currently not represented in the model particularly well, but this is being looked at to develop the model as it is thought to be a significant factor in how the weather behaves.

Many of the computer programs to run the models are still written and developed in Fortran, a computer language that originates back to the 1950s, but is still the most suitable for this sort of scientific work.

Andy Pepperdine