We study source parameters of large, shallow Mexican subduction zone earthquakes (95¿W to 106¿W) which occurred between 1928 and 1986 by modeling the P waves recorded on Galitzin-Wilip seismograph in DeBilt (DBN), Holland. For post-1962 events the source parameters retrieved from DBN seismograms alone agree well with those obtained from long-period World-Wide Standardized Seismograph Network records, giving us confidence in our results for pre-1962 events. All earthquakes are shallow (H~10 to 20 km). With few exceptions the sources in Oaxaca (95¿W to 90¿W) are very simple. To the northwest of 99¿W they are simple as well as complex. The ratio of surface wave to body wave seismic moment (Mos/MoP), which is a measure of long- to-short-period radiation, is smaller in Oaxaca (~1.5¿0.5) than in the regions northwest of 99¿W (~3.1¿1.3). These results suggest a change in the plate interface characteristics near 99¿W. The shape change in the rupture mode and the intersection of the O'Gorman Fracture Zone (OFZ) with the trench occur near 99¿W. Two strike-slip events offshore, close to OFZ, suggest a segmentation of the subducting plate near 99¿W.

The age of the plate near the trench in Oaxaca is not well knwon; it is possible that it does not increase continuously from northwest to southeast in the region but jumps across 99¿W. If so, then the older age of the subducted plate southeast of 99¿W may be the cause of the distinct rupture mode of the Oaxaca earthquakes. The length of the Benioff zone, which is greatest below Oaxaca (≈400 km) and decreases toward the northwest, can be explained by the correlation between the length of the subducted slab and the product of the lithosphere age and convergence rate. The relative complexity of sources, the weaker background seismicity, and the lesser number of aftershocks northwest of Oaxaca may be explained by a stronger interface coupling resulting from subduction of younger oceanic slabs (~5 to 13 m.y. old) in this region. This, however, explains neither larger Mos/MoP values northwest of Oaxaca nor the low stress drop estimates obtained from the analysis of near-field strong-motion data for the Michoacan earthquake of 1985, both of which indicate weaker coupling of the interface. Thus the issue of whether subduction of very young plates (≤10 m.y. old) results in strong or weak coupling remains unsolved from the presently available Mexican data. ¿ American Geophysical Union 1991