Tsunami Record from the Great 1906 San Francisco Earthquake
Shortly after the Great San Francisco earthquake of April 18, 1906, a sea level disturbance (tsunami) was recorded at the Presidio tide gauge station in San Francisco (the station is now located nearby at Ft. Point). What type of mechanism (earthquake rupture, landslide, other) generated the tsunami recorded at the Presidio tide gauge station?
An Unusual Tsunami
Shortly after the Great San Francisco earthquake of April 18, 1906, a sea level disturbance (tsunami) was recorded at the Presidio tide gauge station in San Francisco (the station is now located nearby at Ft. Point). This disturbance puzzled Lawson (1908), author of the comprehensive report of the earthquake, and Henry Reid, proponent of the elastic rebound theory of earthquakes, primarily because it appeared as a small and solitary negative amplitude wave. The graph shows what the record looked like with the tidal component removed (the red arrow indicates the approximate time of the earthquake).
This tsunami record is unusual because the initial lowering of sea level is not followed by a subsequent positive amplitude wave. The lowering of sea level over about 16 minutes was followed by a series of oscillations with an approximate period of 40-45 minutes. Lawson (1908) originally ascribed these later oscillations to reverberation of the tsunami within San Francisco Bay. As shown in the animations below, however, it appears that these oscillations are caused by complex wave effects outside the Golden Gate and not within the bay. The central question remains—what type of mechanism (earthquake rupture, landslide, other) generated the tsunami recorded at the Presidio tide gauge station.
New Fault Interpretation
High-resolution aeromagnetic data was collected to better illuminate the geometry of the San Andreas fault offshore of the Golden Gate. Interpretation of the data by Jachens and Zoback (1999) and Zoback et al. (1999) indicated that the San Andreas fault makes a 3 km right step north of Lake Merced and a smaller 1 km left step south of Bolinas Lagoon, as shown by the red line on the map.
We can further test the hypotheses that rupture of the San Andreas fault during the 1906 earthquake occurred (1) on the newly defined discontinuous fault segments or (2) on a continuous fault trend defined by the older interpretation (orange line in the figure above). At the same time, we must consider other non-seismogenic sources for the tsunami, including massive cliff failures caused by the ground shaking as reported by Lawson (1908).
The Tsunami Model: A Convergence of Disciplines
Geologic investigations in the San Francisco Bay region are, of course, not limited to earthquake studies. The USGS has a vigorous research program related to environmental aspects of San Francisco Bay. As part of the broader effort, researchers from the Water Resources Division of the USGS developed an estuarine circulation model to study sediment and pollutant transport in the bay. The hydrodynamic model used for environmental studies is modified to model tsunami propagation, using initial sea level conditions specified by the parameters of earthquake rupture. Slip during the 1906 earthquake is resolved from geodetic measurements before and after the earthquake (Thatcher et al., 1997), as shown in the figure.
Aside from the amount of slip during the earthquake, we need to specify parameters that describe the geometry of faulting. Two of the cases tested include a tsunami generated from continuous rupture of the fault represented by the green line in Figure 2 (orange line in Figure 1) and discontinuous rupture indicated by the new interpretation. The calculated subsidence of the earth's surface for the two cases is shown in the two map plots.
Below each figure we show a section of the tide gauge record (solid line) in comparison with a synthetic record using the assumed source geometry. The tsunami record is consistent with rupture of discontinuous segments of the San Andreas fault along the Golden Gate platform. Using this hydrodynamic model, we test other possibilities such as compound rupture involving nearby faults in addition to the San Andreas and tsunamis generated by cliff failures. Of all the possibilities, discontinuous rupture of the San Andreas (above) seems to best explain the Presidio tide gauge record.
Animation of the 1906 Tsunami
The results above indicate the tsunami from the 1906 San Francisco earthquake was caused primarily by downdropping of the sea floor north of Lake Merced, between overlapping segments of the San Andreas fault. Three observations are apparent from hydrodynamic modeling of this tsunami:
- the tsunami propagated from the source region to the Golden Gate as a trapped wave (i.e., a particular class of waves that propagate parallel to the shoreline);
- trapped waves generated by this earthquake were reflected and scattered, resulting in the 40-45 min. period oscillations apparent on the tide gauge record; and
- relatively little wave energy is transmitted through the Golden Gate to San Francisco Bay.
The snapshots of the tsunami derived from the discontinuous fault model described above (black lines are current vectors located at every model grid point spaced 250 m apart) are shown here.
The animation is available at two horizontal scales: small scale animation (centered near the Golden Gate) and large scale animation (covering the San Francisco Bay region).
Concluding Remarks
In hindsight, it is remarkably fortuitous that a tsunami was recorded from the 1906 San Francisco earthquake. First, the most likely epicenter for the earthquake was located within 15 km of the only tide gauge station operating in northern California at the time. Second, if the San Andreas fault was continuous offshore, a tsunami probably would not have been recorded. The fact that the San Andreas fault makes a right step in the offshore region means that during earthquake rupture the sea floor is downdropped in the stepover region, resulting in the generation of a tsunami. It is evident from this study that tide gauge records can provide additional and corroborating information on the rupture process of historic earthquakes.
Even though the magnitude of the 1906 earthquake was large (M 7.8), it generated a tsunami wave only approximately 10 cm in height. In contrast, a tsunami from a similar magnitude subduction zone earthquake in other regions bordering the Pacific basin would have (on average) generated a much larger tsunami. The primary tsunami threat along the central California coast is from distant tsunamis generated by earthquakes along subduction zones, such as the 1964 Great Alaska earthquake. For more information on tsunami inundation maps, see the NOAA Center for Tsunami Research.
Tsunami and Earthquake Research
Preliminary Simulations of Recent Tsunamis
Shortly after the Great San Francisco earthquake of April 18, 1906, a sea level disturbance (tsunami) was recorded at the Presidio tide gauge station in San Francisco (the station is now located nearby at Ft. Point). What type of mechanism (earthquake rupture, landslide, other) generated the tsunami recorded at the Presidio tide gauge station?
An Unusual Tsunami
Shortly after the Great San Francisco earthquake of April 18, 1906, a sea level disturbance (tsunami) was recorded at the Presidio tide gauge station in San Francisco (the station is now located nearby at Ft. Point). This disturbance puzzled Lawson (1908), author of the comprehensive report of the earthquake, and Henry Reid, proponent of the elastic rebound theory of earthquakes, primarily because it appeared as a small and solitary negative amplitude wave. The graph shows what the record looked like with the tidal component removed (the red arrow indicates the approximate time of the earthquake).
This tsunami record is unusual because the initial lowering of sea level is not followed by a subsequent positive amplitude wave. The lowering of sea level over about 16 minutes was followed by a series of oscillations with an approximate period of 40-45 minutes. Lawson (1908) originally ascribed these later oscillations to reverberation of the tsunami within San Francisco Bay. As shown in the animations below, however, it appears that these oscillations are caused by complex wave effects outside the Golden Gate and not within the bay. The central question remains—what type of mechanism (earthquake rupture, landslide, other) generated the tsunami recorded at the Presidio tide gauge station.
New Fault Interpretation
High-resolution aeromagnetic data was collected to better illuminate the geometry of the San Andreas fault offshore of the Golden Gate. Interpretation of the data by Jachens and Zoback (1999) and Zoback et al. (1999) indicated that the San Andreas fault makes a 3 km right step north of Lake Merced and a smaller 1 km left step south of Bolinas Lagoon, as shown by the red line on the map.
We can further test the hypotheses that rupture of the San Andreas fault during the 1906 earthquake occurred (1) on the newly defined discontinuous fault segments or (2) on a continuous fault trend defined by the older interpretation (orange line in the figure above). At the same time, we must consider other non-seismogenic sources for the tsunami, including massive cliff failures caused by the ground shaking as reported by Lawson (1908).
The Tsunami Model: A Convergence of Disciplines
Geologic investigations in the San Francisco Bay region are, of course, not limited to earthquake studies. The USGS has a vigorous research program related to environmental aspects of San Francisco Bay. As part of the broader effort, researchers from the Water Resources Division of the USGS developed an estuarine circulation model to study sediment and pollutant transport in the bay. The hydrodynamic model used for environmental studies is modified to model tsunami propagation, using initial sea level conditions specified by the parameters of earthquake rupture. Slip during the 1906 earthquake is resolved from geodetic measurements before and after the earthquake (Thatcher et al., 1997), as shown in the figure.
Aside from the amount of slip during the earthquake, we need to specify parameters that describe the geometry of faulting. Two of the cases tested include a tsunami generated from continuous rupture of the fault represented by the green line in Figure 2 (orange line in Figure 1) and discontinuous rupture indicated by the new interpretation. The calculated subsidence of the earth's surface for the two cases is shown in the two map plots.
Below each figure we show a section of the tide gauge record (solid line) in comparison with a synthetic record using the assumed source geometry. The tsunami record is consistent with rupture of discontinuous segments of the San Andreas fault along the Golden Gate platform. Using this hydrodynamic model, we test other possibilities such as compound rupture involving nearby faults in addition to the San Andreas and tsunamis generated by cliff failures. Of all the possibilities, discontinuous rupture of the San Andreas (above) seems to best explain the Presidio tide gauge record.
Animation of the 1906 Tsunami
The results above indicate the tsunami from the 1906 San Francisco earthquake was caused primarily by downdropping of the sea floor north of Lake Merced, between overlapping segments of the San Andreas fault. Three observations are apparent from hydrodynamic modeling of this tsunami:
- the tsunami propagated from the source region to the Golden Gate as a trapped wave (i.e., a particular class of waves that propagate parallel to the shoreline);
- trapped waves generated by this earthquake were reflected and scattered, resulting in the 40-45 min. period oscillations apparent on the tide gauge record; and
- relatively little wave energy is transmitted through the Golden Gate to San Francisco Bay.
The snapshots of the tsunami derived from the discontinuous fault model described above (black lines are current vectors located at every model grid point spaced 250 m apart) are shown here.
The animation is available at two horizontal scales: small scale animation (centered near the Golden Gate) and large scale animation (covering the San Francisco Bay region).
Concluding Remarks
In hindsight, it is remarkably fortuitous that a tsunami was recorded from the 1906 San Francisco earthquake. First, the most likely epicenter for the earthquake was located within 15 km of the only tide gauge station operating in northern California at the time. Second, if the San Andreas fault was continuous offshore, a tsunami probably would not have been recorded. The fact that the San Andreas fault makes a right step in the offshore region means that during earthquake rupture the sea floor is downdropped in the stepover region, resulting in the generation of a tsunami. It is evident from this study that tide gauge records can provide additional and corroborating information on the rupture process of historic earthquakes.
Even though the magnitude of the 1906 earthquake was large (M 7.8), it generated a tsunami wave only approximately 10 cm in height. In contrast, a tsunami from a similar magnitude subduction zone earthquake in other regions bordering the Pacific basin would have (on average) generated a much larger tsunami. The primary tsunami threat along the central California coast is from distant tsunamis generated by earthquakes along subduction zones, such as the 1964 Great Alaska earthquake. For more information on tsunami inundation maps, see the NOAA Center for Tsunami Research.