基于图像信息方法的日本东北M_W9.0地震回溯性预测研究

Retrospective Forecasting Study of the Tohoku M_W9.0 Earthquake by Pattern Informatics Method

以日本局部地区(32.0°~46.0°N,136.0°~148.0°E)为研究区域,应用图像信息方法,选用8a滑动预测窗长,1°×1°网格为主要计算参数,系统计算了研究区域内2000年以来的"地震热点"(显著异常地区)并获得了区域内7级以上大震特别是2011年日本东北M_W9.0地震的热点演化图像。结果表明:①包含M_W9.0地震的预测时间窗内,其震中的邻近网格持续有热点出现,并且其中有4个连续滑动的窗口中该地震震中所在网格亦存在热点。②在回溯时间段内发生的21个M≥7.0地震中,除2004年9月5日发生于日本近畿南岸近海的M7.1地震前无地震热点外,绝大多数在震前皆有热点出现。③与其他M≥7.0地震相比,M_W9.0地震前热点图像分布范围更广,稳定性更好,持续时间更长。

In this paper, Pattern Informatics （PI） method was applied to the retrospective forecast study for M≥7.0 earthquakes, especially the Tohoku Mw9.0 earthquake in the chosen local area （32.0°-46.0°N, 136.0°-148.0°E） in Japan. With calculating parame- ters of 1°×1° grid and forecasting window of 8 years, hotspot diagrams of each forecasting window have been obtained since the year of 2000, and the relationship between the hotspots and all the M≥7.0 earthquakes during the forecasting windows were studied,es- pecially for the Mw9.0 Tohoku earthquake in 2011. The results show that： ① In all fore casting windows containing the Mw9.0 Tohoku earthquake,the hotspots were in its Moore neighborhood grids. And in four continuous sliding windows,hotspot appeared in its epi- central grid. ②Totally, twenty-one M≥7. 0 earthquakes occurred in the retrospective forecasting windows,and most of them occurred in the hotspot grid or its Moore neighbor- hood grids. Only the MT. 1 earthquake off the coast of Japan Kinki on Sep. 5th 2004 did not occurred in the hotspot grid or its Moore neighborhood grids. ③ Compared with other M ≥7.0 earthquakes,hotspots about Mw9.0 Tohoku earthquake has wider distribution zone, longer lasting time, and higher stability.