Preliminary Modelling of the March 11, 2011 Tohoku-Chiho Taiheiyo-Oki Earthquake Sequence

Point of contact: Eric Kiser

Japanese page

The results showing the sequence of plate interface rupture are generated with single a priori information, i.e., hypocentral location. No other constraints, such as inhibition of rupture on the location that has already experienced rupture, has been imposed. This can be clearly seen by aftershocks that occur within the mainshock rupture area.

Continuous back-projection of data for the first 1500 seconds after the mainshock shows that a number of large aftershocks occurred in this 25-minute window. We can identify them independently.

Earthquakes that break parts of the region that has not previous slipped typically start at the edge of a previous event as expected for triggered events. Note that there are two events that rupture almost simultaneously at about 550 seconds after the mainshock as well as two events (at about 1050 seconds and 1250 seconds) that are nearly identical. Note that the colour saturation is time dependent, i.e., the colours should not be used to identify regions of high or low energy release.

Based upon the back-projection results, we can obtain an independent estimates of the earthquake occurrence time, duration, and size. The table below compares the back-projection catalogue with that from Japan Meteorological Agency (JMA) and USGS National Earthquake Information Centre (NEIC). The time given in the table is with respect to the hypocentral time of the Mw 9.0 event, i.e., March 11, 2001 at 05:46:18 UTC as determined by the JMA. The back-projection results use hypocentral time and location as input parameters.

All the earthquakes with magnitude estimates above 6.0 in JMA and NEIC catalogues are obtained by the back-projection analysis. Furthermore, additional events are identified. In fact, there are many more events we can recognize if the amplitude level is lowered, but for this preliminary report, we are focusing on the cleanest of the events. This result suggests that the back-projection technique can be an effective way to identify large earthquakes occurring very close in time to another large event.

Based upon the rupture area overlap with the mainshock, event M, events A, C, D, E, G, and I are aftershocks of the mainshock. Many of them occur very close in time to the mainshock time, and highest concentration of large aftershocks occur in regions where energy release during the mainshock was relatively low. Events F, H, and J, on the other hand, rupture areas that do not seem to have been involved in the mainshock M, and are likely to be triggered by the mainshock activity. It is interesting to note that they occur later in this time series compared to real "aftershocks". The event B is probably a crustal earthquake.

If we sum moment of all the earthquakes that happen in the first 25 minutes, the resulting moment magnitude is 8.8, 0.4 units higher than the initial event. This would have been the size of the earthquake if all the segments failed in a single event. Because there are uncertainties in the rupture area to moment magnitude conversion, we could take the first event which, based upon back-projection analysis, is Mw 8.4, and scale it up to the size reported by JMA and NEIC, i.e., Mw 9.0. This implies that the potential size of the giant earthquake involving all segments would be Mw 9.4.

The back-projection can identify small events, but for this region using the TA array data, the magnitude for small events (< 6.3) are likely to be overestimated to 6.3 or slightly above. We are working to improve the magnitude estimates.

Back to main page for the Japanese earthquakes