Figure Index:

Figure 1. - Measured displacements associated with the Northridge earthquake. Station locations are indicated by gray circles (and identified by a 4-character I.D. for reference to the tables). Horizontal displacement is shown by a heavy black vector and corresponding 1-sigma ellipse. Uniform-slip model vectors are shown in white; variable slip model in gray.

Figure 2. - Measured and modeled vertical displacements at the GPS stations. Measured vertical displacements are shown by the central black column at each station, and the 1-sigma error is indicated by a vertical line. Uniform-slip model columns are shown in white; variable slip model in gray.

Figure 3. - Cross-section showing the model fault planes overlain on the aftershocks relocated by Mori et al. (1995). We find that one cannot model the geodetic data as well when assuming the fault plane is coplanar with the aftershocks. The uniform slip model plane is shown as a thick gray line, and the larger co-planar fault assumed for the variable slip model is shown as a thinner black line.

Figure 4. - Total moment, subfault nrms, and data nrms (both based on reduced chi-squared values) as a function of the number of retained consecutive eigenvalues for the variable slip modeling. The total moment is the length of each subfault's slip vector, summed over all subfaults. The subfault nrms is an index of dispersion for the length of all subfaults' slip vectors. The data nrms is a similar index for the model residuals.

Figure 5. - Distribution of slip on the Northridge fault plane. Contours and shading indicate the slip amplitude (in meters), and arrows indicate the slip vector for each subfault of the model. The model has 10 subfaults along- strike and 13 down-dip, each 2 km by 2 km. Slip evidently occurred mainly updip and northwest of the hypocenter. Less than one meter of slip occurred above a depth of 5 km. This model, based solely on the GPS data, differs substantially from those based solely on seismological data (Wald and Heaton, 1994b; Dreger, in press). It differs, but less dramatically, with the "geodetic only" and "combined" models of Wald et al. (this issue).

Figure 6. - Distribution of normalized residual (NRES) values for the uniform and variable slip models. The outlier (especially in the east and vertical components) is station LOVE, whose displacement is anomalous (as discussed in the text).

Figure 7. - Residual vectors at the GPS stations. Some spatial systematics appear to remain in these unmodeled data, particularly in the vicinity of the eastern Ventura Basin, as discussed in the text. Also, residuals are significant but diversely oriented in the Santa Monica Mountains (near the down-dip edge of the rupture). Note the vector scale is several times larger than in Figure 1.

Figure A1. - GPS station history of occupations. Station names are given on the left and the total number of occupations for each station is given on the right. Each occupation is shown as a diamond symbol (except thttp://www.scecdc.scec.orgrveys, multiple sessions on the same day are represented by one symbol). Stations with a total of only 2 or 3 occupations are most susceptible to coarse errors.