Long-range Forecasting

He emphasised that a vastly better understanding of how the atmosphere worked would be required before accurate long-range forecasts could be made. According to my notes he began his first lecture with the statement "a question that is sometimes raised concerns the justification for the assumption that the general circulation of the atmosphere is organised. If there is no organised system then there is no hope for long-range forecasting".

It seemed that Willet used the term 'organised' to describe an atmosphere in which a given initial state would be uniquely followed by a specific sequence over a period of months.

Many years later I read the papers of the UK's G. D. Robinson in the 1960s and those of Ed Lorenz of the US in the 1970s on predictability. These papers suggested that the limited 'memory' of the atmosphere would make it unlikely that accurate predictions of the state of the atmosphere could be made for more than a week or two ahead.

If there is randomness in the input of energy into the atmosphere it would seem impossible to make accurate long-range forecasts for more than a week to 10 days. But the demand for long-range forecasts is so great that there will always be efforts to make such predictions. Many have been made, but even with the dramatic scientific and technological advances in recent years few long-range forecasts have shown sufficient skill and precision to be practically useful.

Willet's lectures were invaluable in their breadth of vision and their emphasis on the gaps in the state of knowledge of the atmosphere at that time. I wonder whether, in these days of ready responses by scientists to the questions of media interviewers, more caution and less dogma would be desirable in scientists' replies.

Bureau of Meteorology