Effect of the Gardner and Knopoff declustering method on seismicity

Effect of the Gardner and Knopoff declustering method on seismicity removal
from the earthquake catalog
K. R. Felzer
The Gardner and Knopoff (1974) routine has been traditionally used for catalog
declustering when making earthquake hazard maps. This routine consists of
empirically determined aftershock sequence time and space windows that vary as a
function of mainshock magnitude. Gardner and Knopoff (1974) specified the time
and space windows at half magnitude increments from M 2.5 to M 8.0. For
application to the full earthquake catalog I linearly interpolate between these
specified values. The Gardner and Knopoff (1974) routine also specifies that the
largest earthquake be retained from each cluster. Thus both foreshocks and
aftershocks are removed from the declustered catalog, with the traditional
mainshock remaining behind. This preferential removal of the smaller earthquakes
produces a declustered catalog that has a lower b value than the original catalog.
Implementing the routine typically results in the removal of at least 2/3 of the
earthquakes from the catalog.
Below is a chart of the fraction of earthquakes that is removed from the UCERF3
catalog as a function of magnitude by the Gardner and Knopoff (1974) method
during three independent time periods. These values give an idea of the fraction of
earthquakes that is removed and the consistency of this fraction at different times.
Magnitude
2.5≤M<3
3≤M<3.5
3.5≤M<4
4≤M<4.5
4.5≤M<5
5≤M<5.5
5.5≤M<6
6≤M<6.5
6.5≤M<7
7≤M<7.5
7.5≤M<8
1984-2011
catalog
0.78
0.74
0.72
0.66
0.67
0.54
0.42
0.55
0.14
0
N/A
1932-1984
catalog
1855-1932
catalog
0.67
0.53
0.42
0.41
0.23
0.11
N/A
0
0.37
0.12
0.12
0
0