| In seismic zoning,seismicity tendency forecasting is an important step.The intensity at the Songpan-Maowen region on 1990 zoning map of China is one degree lower than that on the 1977 map,which is due to the difference between the judgements on the seismicity tendency.It is taken as a case study;reliability of the forecasting and how to take the result account into zoning is deal with in deep in this dissertation.This paper proceeds as follow.(1)In those years,a working team on intensity in Southwestern China group forecasted the seismicity tendency for the time period after 1970 in the Central seismic zone of China and their result was adopted by 1977 map.T he same data and the same method for the forecasting are adopted in this study for checking.The years of 1920,1933,1955 and 1970 are taken as the starting points respectively for the seismicity tendency forecasting of the periods afterwards.Peaks in forecasting time series appeared nearly at the same time as those in 1970's.The result is consistent with that by the team in those years,as those earthquakes with magnitude no less than 7 may occur in 2007 and 2011-2013 in the Central zone,similar with the occurring times of Wenchuan earthquake and Lushan earthquake.Spatial distribution of seismicity in the zone is further simplified as latitude-time series,earthquake locations in the same periods are forecasted by the same method.The results are quite different from the forecasting by the team in those years,not near the Songpan-Maowen and the Luding area,and they are unstable too.(2)The earthquake data from 1971 to 2016 are supplied then to the series,the results by the same method are similar with the previous results both in time and space domain.From the data of 1500-2016,the time series are constructed for earthquakes with magnitudes no less than 6 and 7 respectively,and then the correlation function of them is calculated.The time series of seismicity no less than 7 in the next 100 years from 1980 and 1990 are forecasted respectively.The peaks on the series also appeared near 2008 and 2013.The results are quite stable at the two time periods.The maximum difference between the actual strong earthquake occurrence and the corresponding forecasting time is about ten years,and the more data the smaller differences.(3)Furthermore,sensitivity of the seismicity tendency forecasting on the adopted starting points of time series,threshold magnitude and the length of forecasted time,are analyzed respectively.The result shows that former two have just little impact on the forecast,while the latter should not be short for a robust forecasting,and 100 years is long enough.In conclusion,Wiener filtering is feasible to forecast seismicity tendency in the next 100 years in a zone for seismic zoning.After the earthquake data in the last 46 years supplied,forecasting on the spatial area is still unsuccessful.It is further showed that to forecast the location of earthquake roughly by Wiener filtering is still not feasible up to now.The spatial forecasting on seismicity tendency in seismic zoning is on the way to search some new approaches.(4)An Artificial Intelligence method is tried to forecast seismicity tendency in this dissertation.The earthquake data from 1500 to 2016 in the Central seismic zone are adopted.The samples for leering of Artificial Neural Network are constructed by means of acquiring characteristic values by a window moving along time.The temporal and spatial tendencies are forecasted by a 16-8-4 network and by both of a 12-8-4 and a 16-8-4 networks respectively.For checking of the forecasting,the data during the last given period are pre-kept from the network training.The results show that large earthquakes may occur at 28?N-31?N in the Central zone during 1990-2019,that temporal and spatial range may cover those of Wenchuan earthquake at 31.0?N in 2008 and Lushan earthquake at 30.3?N in 2013.The deficiency to be deal with further is that the range is large. |
PDF Full Text Request