Atmospheric fluctuations may mislead weather forecasts

Oxford Blueprint - the newsletter of the University of Oxford, 4(8), p 2, 18 March 2004

Scientists at Oxford have discovered that small-scale fluctuations, which are wide-spread in the atmosphere, may have a greater impact on weather systems than has previously been thought. The results, published in Nonlinear Processes in Geophysics, may have important implications for accurate weather forecasting.

The fluctuations, known as inertia-gravity waves because they are sustained by a combination of inertial and gravitational forces, are prominent in the bottom 15 km of the atmosphere. They can often be seen from the surface of the Earth as 'stripy' features in clouds. Their horizontal wavelengths can be as short as 5 km - too small to be picked up by current weather prediction models, which divide the surface of the Earth into grid-boxes measuring around 50 km by 50 km.

Meteorologists have therefore always had to assume that inertia-gravity waves do not significantly interact with weather systems, such as warm and cold fronts, but this assumption had never been rigorously tested.

Motivated by the results of laboratory experiments, which seemed to challenge the meteorologists' assumption, the Oxford scientists developed a computer model of a simple fluid system resembling the atmosphere. They represented the inertia-gravity waves as random noise in the model, since the fluctuations can be highly irregular, chaotic and transient. They found that the system could behave differently when the inertia-gravity wave representation was activated - in other words, the meteorologists' assumptions were not always justified.

In particular, the state of the fluid could undergo spontaneous transitions to quite different states, with a dramatic shift in the patterns of low and high pressure. Extrapolation of these results to the real atmosphere suggests that inertia-gravity waves could be a cause of significant errors in weather forecasts. The surprise implication of the research is that adding random noise to the forecast might actually help improve things.

Dr Paul Williams, one of the scientists involved in the study, said: 'It seems that we have observed a phenomenon which might have the potential to affect the accuracy of weather forecasts. More research is needed to find out exactly how bad the forecast error might be, but the preliminary results are very exciting.'