Abstract:

China Seismological Bureau (CSB) has been〔irrying on routine annual earthquake prediction in China since 1975. In this study, we apply a scheme of R score to evaluate the disclosed annual predictions in the 90s. A completely random guess leads to an R score of 0 and a complete successful prediction has an R score of +1.The average R score of the annual prediction in China in the 90s is about 0.184, greater than 0.This score has to be compared with deferent strategies of random predictions because the background seismicity in China is different in different regions. The first strategy is to m ake chances of predictions at each location proportional to its background seismicity, which leads to an expected R score of 0.150. The CSB annual prediction is marginally higher than this value. The second strategy is predictions are made only at regions of highest background seismicity, the R seore then can reach、high as 0.270, superior than the CSB prediction, but this kind of prediction is practical meaningless. A combination of actual prediction of CSB and regions of highest background seismicity〔an raise the R score to 0.336, superior than strategy 2. This study indicates that annual earthquake prediction in China is still in a very preliminary stage. CSB annual prediction is not better than random predictions at remote areas, but behaves better than random predictions in populated and economic important regions which are the main areas direfully monitored in Chinese earthquake prediction project.

Key words: earthquake prediction, statistical test, China, evaluation