Abstract

This paper is concerned with intermediate-term prediction of large earthquakes for the Himalayan seismic belt. The intermediate-term prediction is here understood to involve times of a few months to a few years in areas of a few hundreds of square kilometers. We are going to predict earthquakes that occur once every 5--10 years on the average. The basic assumption is that the preparation of a large earthquakes affects the low magnitude seismicity in the area. To be more specific, a large earthquake is expected to be preceded by a higher rate of occurrence and nonuniformities in low magnitude seismicity. Such changes are typical of many nonlinear systems when approaching the critical state. In the case under consideration, the system in question is formed by the earthquake-generating fualts of the area. The CN algorithm, based on this hypothesis, is designed to deal with intermediate-term prediction of large earthquakes. It has been applied with some success to many seismic regions of the world. The database to use in prediction includes smaller earthquakes in the area. This work is a retrospective testing of whether the CN algorithm can be used to predict large earthquakes in the Himalaya. This area was divided into three subareas to try to predict earthquakes of magnitude 6.4 or greater. A total of six such events have occurred in the area during the period 1970--1988, five of these have been predicted retrospectively. The alarm time varies between 15 and 31 percent of total observation time. The above result inspires some hope for a tentative forward prediction of large earthquakes in the Himalaya using the CN algorithm.