Leveling observations and data modeling

     Analyzing the leveling data collected from 1995 to 1998 in the Guerrero coast, we conclude that either the geometry of the locked patch at the Guerrero gap is distinctly different from the geometry of neighboring locked zones or else, the Guerrero gap is persisting in a rather unexpected seismotectonic mode, which implies an extremely sluggish strain release. This could be due either to a slow, ongoing thrust slip event (e.g., Hirose et al., 1999) or to an abnormally long-lasting period of preseismic slip. The latter phenomenon may be related to the specific frictional properties of the seismogenic interface (e.g. rate- and state-dependent frictional laws, Kato et al., 1997) when quasi-stable sliding occurs prior to a great thrust earthquake.
     These hypotheses creates a number of essential problems which should be consequently resolved. For example,

        -  when and why has the strain accumulation been completed and the strain release commenced? Is this a slow thrust-slip event or is it precursory sliding?
         -  At what point in time will a transition from quasi-stable sliding to unstable slip take place?

We need to continue all types of observations (leveling, GPS, and high precision tilt measurements) on the Guerrero coast to resolve at least some of these questions.
     The simple model of instantaneous seismic rupture we used to fit the observed uplift in the Atoyac line leads to the conclusion that extremely slow thrust slip may be occurring. This inference appears to be self - contradictory given that such a model requires subduction to cease temporarily, while only the ocean-ward motion of the overriding continental plate persists along a patch of the plates' interface (Figure 3 d,e). This apparent contradiction is easy to resolve because the model is the linear superposition of steady-state, kinematic subduction along the entire plate interface and the additional thrust slip on the rupturing patch.
     We also considered models with a slow-growing locked zone. However, they could not fit the observed deformation in the Guerrero gap better than static locked zone models. The model with the absence of any locking at the plate interface does not match the data at all.

Fig. 1a. Modeling of the vertical uplift rate for the Barra de Potosí leveling line (Elastic dislocation model, Savage, 1983). Shaded areas below the zero fault depth line represent the configuration of the subducting Cocos plate and the interplate contact according to Kostoglodov et al., (1996). The geometric parameters of the fault segments and the slip rate are given in the tables at the top of the Figure. Locked segments are shown as elongated filled rectangles. a - Planar fault model. Different models (a, b, c) can fit the observed relative uplift rate, nevertheless, the location and width of the locked zone hardly show any variations.

Fig. 1b - Better fit to the uplift and the interplate geometry using a segmented fault model. The location of the locked zone is approximately the same as in case a. Locked segments are shown as red heavy solid lines. Other explanations are the same as for Figure 1.

Fig. 2. Comparison of the predicted vertical motion and the observed vertical uplift rate (1995-1998 epoch) for the Acapulco leveling line. Segmented fault fitting the interplate geometry. Locked segments are shown as red heavy solid lines. Other explanations are the same as for Figure 1.

Fig. 3a. Predicted vertical motion on the Atoyac line. Interseismic coupling model (a) - simple plane fault for the subsiding coastal area. The data can only be fitted if the locked segment location does not agree with the interplate contact configuration. Locked segments are shown as red heavy solid lines. Vertical bars are one sigma cumulative standard deviations of the relative uplift rate estimates. For other explanations see Figure 1.

Fig. 3b. Predicted vertical motion on the Atoyac line. Interseismic coupling model (b) - complex plane fault for the subsiding coastal area. The locked seismogenic zone is too deep and rather far from the coast. Locked segments are shown as red heavy solid lines. Vertical bars are one sigma cumulative standard deviations of the relative uplift rate estimates. For other explanations see Figure 1.

Fig. 3c. Predicted vertical motion on the Atoyac line. Interseismic coupling model (c) - complex plane fault for the uplifting coastal area. Geometry of the locked zone is reasonable. Locked segments are shown as red heavy solid lines. Vertical bars are one sigma cumulative standard deviations of the relative uplift rate estimates. For other explanations see Figure 1. For other explanations see Figure 1.

Fig. 3d. Predicted vertical motion on the Atoyac line. Coseismic slip or "Slow rupture" model (d) reveals the slipping faults, which are consistent with the interplate geometry. The model fits the data for the subsiding coast. Locked segments are shown as red heavy solid lines. Vertical bars are one sigma cumulative standard deviations of the relative uplift rate estimates. For other explanations see Figure 1.

Fig. 3e. Predicted vertical motion on the Atoyac line. Coseismic slip or "Slow rupture" model (e) reveals the slipping faults, which are consistent with the interplate geometry. The model fits well the coastal uplift. Locked segments are shown as red heavy solid lines. Vertical bars are one sigma cumulative standard deviations of the relative uplift rate estimates. For other explanations see Figure 1.

     In order to unequivocally determine the deformation mode attendant to the interseismic regime in the Guerrero gap, we will combine the differential leveling data with GPS survey data to be gathered in late 2000.