For the past 50 years billions have been pumped into earthquake research. Experts now believe that a lot of this has been wasted.
'Prediction is impossible,' said University of Colorado professor of geophysics Roger Bilham. 'But we can offer 30 year probabilities, though it requires knowing the history of a region.'
Dr Russ Evans of the British Geological Society added: 'It is a feature of earthquakes that they occur in time clusters. A lot of good work is being done superimposing time based clustering on geographical clustering to give statistical probabilities. There is also a lot of research being carried out on short term warning systems.'
Earthquake forecasting is based on the calculation of strain field areas in the ground using boundary element analysis. When a fault slips it produces a strain field which influences the adjacent field. This allows scientists to develop a picture of developing strain over hundreds of years.
'Every 500 years the Anatolian Fault unzips in a strong series of events,' explained Bilham. 'It has been slipping east from the Aegean and west from Iran leaving a gap at Istanbul. Now the first part of that gap has been filled. The remaining bit is just south of Istanbul.
'There has got to be at least a 50% probability of a 7-7.5 Mw earthquake near Istanbul in the next 30 years. And I think that is a conservative view.'