Roadmap for earthquake numerical forecasting in China-Reflection on the tenth anniversary of Wenchuan earthquake

Abstract

Failure of forecasting the 2008 Wenchuan earthquake has been a shake up for Chinese earthquake forecast research. Since 1966, Chinese have been trying to predict earthquakes by empirical method. This method obtained limited success, but has not been able to make break though after hard work of half a century. Many Chinese seismologists believe that earthquake forecast should be changed from empirical forecast mainly based on precursors to physical forecast based on understanding of earthquake mechanics. We therefore propose a frame of earthquake numerical forecast system. Earthquakes occur in the lithosphere when tectonic stress increases and exceeds the crustal strength in the dynamic earth. With growing knowledge of the geological structure, thermal state, porous fluid conditions, physical properties of the lithosphere, and in depth understanding of the earthquake mechanics in the poro-elasto-plasto-viscous rock mass cut by active faults. People now have the capacity to perform massive computation to model the evolution of stress field the occurrence of series earthquakes to produce the synthetic earthquake catalog. The synthetic sequence may help us to understand the real regional earthquakes as better data are obtained model parameters, boundary and initial conditions. It is time to start exploring earthquake numerical forecast, although there is a long way to go to be actual practiced. Weather numerical forecast took about half century from idea to reality. Earthquake numerical forecast may take even longer time. Ten years ago, China had issued the National Earthquake Science and Technology Development Program (2007-2020), which claimed to start "the earthquake numerical forecast test research". To implement such a program, a route map of numerical earthquake forecast is necessary. We suggest a frame work for such a route map. A single synthetic earthquake model can produce a deterministic series of events, but uncertainties in the model as well as in boundary and initial conditions may lead to different series of events. Ensemble forecast therefore can be made for time independent model of long term forecasts. If the sequences can be made to repeat historical earthquakes, by properly adjust the initial conditions, then, time dependent ensemble forecasts can be made for midterm forecasts. Short time numerical forecasts, however, are not practical at present time. Empirical forecasts play hard to search any possible precursors, numerical forecasts, however, try to understand the physics and obtain data to constrain the boundary and initial conditions. They have different ideas to spend financial and human resources. Numerical forecast research will be a pilot program to lead essential progresses in earthquake science.