Abstract: Using the modified Cora-3 pattern recognition algorithm established by I.M.Gel-fand et al., we have studied the related date in tectonics, geomorphology and geophysics to identify destructive earthquake-prone regions in East China(30?-35癗, 116?-123癊). The results are satisfactory.Seven D (Dangerous) and eight N (Non-dangerous) "characteristic traits" are selected from sixteen original features. The geological meanings of the traits are analyzed. It shows that destructive earthquakes in East China often occur in such regions as inside subsidence areas during the neotectonic period, inside or in the margin of Cretaceous or Cainozoic structural/faulted-down basins, near the boundaries of the blocks having vertical neotectonic movement of contrasting nature, as well as where there are abundant active faults. In the above-mentioned regions some degree of geophysical anomalies generally appears at the same time.The mis-identifying rate in the course of "Voting" is low. By voting we have zoned five earthquake-prone regions in East China. The conclusions are confirmed by the control experiments.

Key words: Eastern China, Pattern recognition, Zoning of dangerous areas

摘要： 按I.M.Gelland等人建立的模式识别Cora-3修改算法研究华东地区的地质构造、地貌和地球物理特征并用于识别本地区破坏性地震。破坏性地震易发生在新构造运动的沉降区内、白垩纪或新生代构造(断陷)盆地的内部或边缘、新构造垂直差异运动的地质块体边界附近以及活动断层数较多的地方。 “投票”结果表明误识率较低(ε=0.061—0.098)。根据“投票”结果,划分出五个D节点相对密集的地震危险区。 控制试验表明识别结果是比较稳定和可靠的。

关键词: 华东地区, 模式识别法, 危险区划