Two techniques are applied in order to estimate the seismicity parameters a
nd their depth distribution in South America, an area of high seismic activ
ity. These techniques are: (1) Gumbel's third asymptotic (GIII) distributio
n of extremes and (2) the maximum likelihood approach. The obtained seismic
ity parameters through GIII distribution of extremes, are the omega value w
hich is considered as the upper bound magnitude and is related to the finit
e maximum stresses and strains which are currently accumulated and released
by the rocks, as earthquake, in a region, the shape parameter lambda, and
the characteristic value u, with P(u) = 1/e and P(omega) = 1. The most prob
able maximum magnitude, RM, of earthquakes for 95 years is also computed. T
he seismicity parameters obtained by the application of the maximum likelih
ood approach are the activity rate, lambda, of the seismic events, the mean
return period, RP, of earthquakes above a cutoff magnitude M greater than
or equal to m and the parameter b of the magnitude-frequency relation. The
parameter beta which is related to b with b = beta log(10)(e), is also comp
uted. The regional maximum magnitude, M-max (this is considered as the maxi
mum possible magnitude in a specific region), is also provided under some c
onstraints. Six depth-range subdivisions are defined for the examined area.
All the seismicity parameters are computed and especially for GIII distrib
ution a covariance analysis of its three parameters (omega, u and lambda) i
s performed for each individual depth range. In all cases omega values are
higher than the regional M-max estimates. The seismic activity rate decreas
es with depth, while the beta parameter, which ranges between 1.18 to 2.06,
does not show a clear regularity in its depth distribution. An important n
otification is that the parameter lambda obtained by the two methods, expre
sses totally different quantities. (C) 2000 Elsevier Science B.V. All right
s reserved.