Predicting Predictable: Accuracy and Reliability of Earthquake Forecasts

Friday, 19 December 2014
Vladimir G Kossobokov, Institute of Earthquake Prediction Theory and Mathematical Geophysics, Russian Academy of Sciences, Moscow, Russia; Institut de Physique du Globe de Paris, Paris, France
Earthquake forecast/prediction is an uncertain profession. The famous Gutenberg-Richter relationship limits magnitude range of prediction to about one unit. Otherwise, the statistics of outcomes would be related to the smallest earthquakes and may be misleading when attributed to the largest earthquakes. Moreover, the intrinsic uncertainty of earthquake sizing allows self-deceptive picking of justification “just from below” the targeted magnitude range. This might be important encouraging evidence but, by no means, can be a “helpful” additive to statistics of a rigid testing that determines reliability and efficiency of a farecast/prediction method.

Usually, earthquake prediction is classified in respect to expectation time while overlooking term-less identification of earthquake prone areas, as well as spatial accuracy. The forecasts are often made for a “cell” or “seismic region” whose area is not linked to the size of target earthquakes. This might be another source for making a wrong choice in parameterization of an forecast/prediction method and, eventually, for unsatisfactory performance in a real-time application.

Summing up, prediction of time and location of an earthquake of a certain magnitude range can be classified into categories listed in the Table below –

Classification of earthquake prediction accuracy

Temporal, in years

Spatial, in source zone size (L)




Up to 100













Note that a wide variety of possible combinations that exist is much larger than usually considered “short-term exact” one. In principle, such an accurate statement about anticipated seismic extreme might be futile due to the complexities of the Earth’s lithosphere, its blocks-and-faults structure, and evidently nonlinear dynamics of the seismic process. The observed scaling of source size and preparation zone with earthquake magnitude implies exponential scales for territorial accuracy of predictions similar to the temporal ones. It varies from exact pinpointing the source to long-range uncertainty of about a few tens of L.

Understanding of the complexity of seismic process has led to a reliable intermediate-term middle-range predictions tested globally since 1992.