Solar activity (SA) has been hypothesized to be a trigger of earthquakes, although it is not as intuitively associated as other potential triggers such as </span><span style="font-family:Verdana;"&g...Solar activity (SA) has been hypothesized to be a trigger of earthquakes, although it is not as intuitively associated as other potential triggers such as </span><span style="font-family:Verdana;">tidal stress, rainfall, and the building of artificial water reservoirs. Here, we in</span><span style="font-family:Verdana;">ves</span><span style="font-family:Verdana;">tigate the relation between SA and global earthquake numbers (GEN) by using</span><span style="font-family:Verdana;"> a deep learning method to test the hypothesis. We use the daily data of GEN </span><span style="font-family:Verdana;">and SA (1996/01/01</span></span><span style="font-family:Verdana;">-</span><span style="font-family:Verdana;">2019/12/31) to construct a temporal convolution netw</span><span style="font-family:""><span style="font-family:Verdana;">ork (</span><span style="font-family:Verdana;">TCN). From the computational results, we confirm that the TCN captures th</span><span style="font-family:Verdana;">e </span><span style="font-family:Verdana;">relation between SA and earthquakes with magnitudes from 4.0 to 4.9. We als</span><span style="font-family:Verdana;">o </span><span style="font-family:Verdana;">find that the TCN achieves better fitting and prediction performance compar</span><span style="font-family:Verdana;">ed with previous work</span></span><span style="font-family:Verdana;">.展开更多
文摘Solar activity (SA) has been hypothesized to be a trigger of earthquakes, although it is not as intuitively associated as other potential triggers such as </span><span style="font-family:Verdana;">tidal stress, rainfall, and the building of artificial water reservoirs. Here, we in</span><span style="font-family:Verdana;">ves</span><span style="font-family:Verdana;">tigate the relation between SA and global earthquake numbers (GEN) by using</span><span style="font-family:Verdana;"> a deep learning method to test the hypothesis. We use the daily data of GEN </span><span style="font-family:Verdana;">and SA (1996/01/01</span></span><span style="font-family:Verdana;">-</span><span style="font-family:Verdana;">2019/12/31) to construct a temporal convolution netw</span><span style="font-family:""><span style="font-family:Verdana;">ork (</span><span style="font-family:Verdana;">TCN). From the computational results, we confirm that the TCN captures th</span><span style="font-family:Verdana;">e </span><span style="font-family:Verdana;">relation between SA and earthquakes with magnitudes from 4.0 to 4.9. We als</span><span style="font-family:Verdana;">o </span><span style="font-family:Verdana;">find that the TCN achieves better fitting and prediction performance compar</span><span style="font-family:Verdana;">ed with previous work</span></span><span style="font-family:Verdana;">.
