Climate-change scientists have called for massive investment to improve computer modelling of the effects of global warming.

There were also calls at a climate symposium at the Royal Society in London for greater co-operation between the various specialists involved, including computer modelling experts. The symposium managed to be both reassuring (for people living well inland in Britain) and terrifying.

The fear that Britain will freeze from a flipping of the Gulf Stream has receded; it is now thought that there will be a slowing of the great flow of warming water from the tropics but the loss of heat will be more than offset by the warming caused by greenhouse gases. Chart after chart at the symposium showed that climate change is both normal and scary.

Ice sheets reached down to London’s Finchley Road just 200 lifetimes ago; 100 lifetimes ago you could walk from Britain to the continent. As one speaker said: “Anything that has happened in the past can happen again.”

The question that exercised the scientists was the extent to which you can use past fluctuations to build computer models to predict future changes ­ and how you then persuade people to believe those models, especially when they are riddled with uncertainties. You can read the past to see what the world looked like under different climatic conditions.

You can test your computer models to see how well they can fit historical records. But, as several speakers pointed out, your models can only take you so far because what is happening now is unprecedented. The one certainty, for all but a small minority of scientists, is that human activity is causing the world to warm up.

What is not known for sure is how quickly this will happen, and what the effects will be. The complexities are daunting. To take two variables: global temperature and the level of carbon dioxide in the atmosphere. The relationship between these is generally depicted as a simple case of the more the CO2, the hotter the earth.

But Professor Peter Cox of the University of Exeter pointed out: “The climate is sensitive to CO2 but CO2 is more sensitive to climate.”

Higher temperatures affect the growth of CO2-absorbing plants and the absorption of CO2 in the oceans, with the result that rises in temperature historically tend to come some time after CO2 levels increase. This fact was seized upon by a recent Channel 4 documentary to dismiss global warming claims as a “swindle”. For scientists it is another complex feedback mechanism to fit into their models.

Humans are of course disturbing its damping effect by releasing CO2 trapped for millennia as oil, a natural form of carbon sequestration. Desperately in need of better modelling is the melting of ice, both at the poles and in more southerly upland glaciers that act like a reservoir for water supplies in places like northern India, the symposium was told.

The effect of ice and ocean warming on future sea levels has produced an alarming range of predictions. Professor Gerard Roe, of the University of Washington, said he had recently been to a workshop of experts, none of whom “was prepared to rule out the possibility of [a rise of] metres in a century”.

A rise of just one metre would put much of East Anglia, Holland, and the north German coast below sea level and displace millions of people in places including Bangladesh. Local impacts such as these need more study and better fo recasts.

Professor Bob Watson, chief scientific adviser to the Department of Environment, Food and Rural Affairs, called for “high-resolution, probabilistic models”. He agreed that this would require multi-petaflops of computing power that might need to be financed at a European level.

He stressed that there was no point in predicting the impact of climate change without also developing a strategy for adapting to it, and that the issue should not be divorced from others such as bio-diversity and pollution.