- © 2005 by the Seismological Society of America
Since the beginning of June 2004, near-real-time ShakeMaps have been produced in Ontario for earthquakes of M > 2.8 and are posted at http://www.shakemap.carleton.ca/ within minutes of occurrence. ShakeMaps, originally conceived by Wald et al. (1999), provide rapid online assessment of locations, shaking intensities, and expected levels of damage to specific areas, The Nuttli magnitude (MN) 5.4 earthquake which occurred 17 km southwest of Rivière-du-Loup, Québec on 6 March 2005 was the first moderate, well recorded event since the implementation of the Ontario ShakeMap project. It provided a good opportunity to evaluate the performance of ShakeMap in eastern Canada.
The ShakeMap location and moment magnitude, based on the ground-motion centroid, are very close to traditional estimates of these parameters. ShakeMap intensities agree with the preliminary observed intensity results collected based on felt reports submitted online. Recorded ground-motion parameters from this earthquake agree very well with the predictions of empirical ground-motion relations developed for ShakeMap applications by Kaka and Atkinson (2005a), as well as with relations developed by Atkinson and Boore (1995).
A moderate earthquake of MN 5.4 occurred in the Charlevoix seismic zone on 6 March 2005 (01:17:49 EST) at 47.75°N 69.73°W (see earthquake location determinec by the Geological Survey of Canada at http://www.seismo.nrcan.gc.ca/nedb/eq_db_e.php). The earthquake was felt strongly in Rivière-du-Loup, Saint-Pascal, and La Malbaie and Saint-Pacome regions. It was also felt widely throughout Québec, eastern Ontario, and the border regions of the northeastern United States. No significant structural damage has been reported.
The Riviére-du-Loup event was the first moderate well recorded event in eastern Canada since the implementation of the Ontario ShakeMap project. ShakeMaps, originally conceived by Wald et al. (1999), provide rapid online assessment of the locations, shaking intensities, and expected levels of damage to specific areas, q-he Rivière-du-Loup event provided a good test of the performance of ShakeMap and an opportunity to examine the validity of different relations that are implemented in the Ontario ShakeMap program.
ShakeMap sent e-mail notifications within 3 minutes of the event, and maps were automatically produced and posted within 7 minutes. Initial maps were not accurate because real-time ShakeMaps are currently based only on POLARIS stations (within Ontario, far from the earthquake). The event was postprocessed to include all of the available CNSN (Canadian National Seismograph Network) stations, to give an indication of the performance of the system that can be obtained if all eastern Canadian stations were used. As of the end of 2005, all CNSN stations will be available in real-time for ShakeMap applications, in addition to the POLARIS stations. Thus the use of all stations is indicative of the ShakeMap performance that can be readily obtained with the current eastern Canadian station distribution.
PERFORMANCE OF SHAKEMAP
Location and Magnitude
Ontario ShakeMap uses the ground-motion centroid concept, originally developed by Kanamori (1993), to estimate the earthquake magnitude and location (see Kaka and Atkinson, 2005b). In Figure 1, the ShakeMap location is compared to that of the epicentral location determined by the Geological Survey of Canada; the centroid is located approximately 17 km from the epicenter. The ShakeMap estimate of moment magnitude (M) is 5.0, whereas the GSC value of MN 5.4 provides an estimated value of M 4.9, using empirical relations (Atkinson, 1993a) to convert MN to M. Preliminary moment-tensor evaluations of the event provide a value of M 4.7 to 4.8 (D. Boore, personal communication, 2005, based on estimates made by W. Y. Kim and R. Herrmann, respectively). Thus the magnitude determined by the ground-motion centroid approach appears to be accurate, while there is an offset in the location of the centroid from the epicenter, attributable to the effects of ground-motion amplitude variability on the centroid for sparse station distribution.
Recorded ground motions include broadband three-component and short-period data from the Canadian National Seismographic Network (CNSN) and Portable Observatories for Lithospheric Analysis and Research Investigating Seismicity (POLARIS) network (see Figure 1). Instrument response is removed as described by Atkinson (2004). Ground-motion parameters, listed in Table 1, include peak ground velocity (PGV), peak ground acceleration (PGA), and 5% damped pseudo-acceleration (PSA) at various frequencies.
An important element of ShakeMap is the combination of recorded ground motions with predicted values based on location and magnitude. The use of the predicted values is particularly important in eastern Canada to “fill out” the ground-motion pattern, due to the sparse distribution of seismograph stations. In Ontario ShakeMap, ground-motion values between stations are predicted using the empirical ground-motion equations given by Kaka and Atkinson (2005a), using the magnitude and location of the centroid. Figure 2 shows that the ground-motion amplitudes predicted by the Kaka and Atkinson relations agree well with the recorded amplitudes; the predictions of Atkinson and Boore (1995) are also in close agreement with observations. Only vertical-component data are shown on Figure 2, as the Kaka and Atkinson relations are specified for vertical-component amplitudes. The Atkinson and Boore (1995) horizontal-component predictions were converted to the equivalent vertical component using a relation for hard-rock sites given by Atkinson (1993b). The vertical-component amplitude is chosen because the site amplification of the vertical component is relatively small, making it a more useful basis for the estimation of location and magnitude. The fact that the vertical-component soil amplitudes agree relatively well with the rock amplitudes, as seen in Figure 2, supports this approach. Note that when predicting intensities from the ground motions, the vertical-component amplitudes are converted to equivalent horizontal values based on H/V ratios that depend on site soil conditions, using site-specific H/V ratios for the recording stations.
ShakeMap was designed as a rapid-response tool to portray the extent and variation of ground shaking and expected damage throughout an affected region immediately following a significant earthquake (Wald et al. 1999). The felt-intensity level is more useful to those who are not familiar with ground-motion parameters such as PGV, PGA and PSA (f). A key element of ShakeMap is the requirement for a relationship to relate ground-motion amplitude to felt intensity that is applicable to eastern North American earthquakes. In our ShakeMap implementation, felt intensity is derived from an empirical relation between PGV and Modified Mercalli intensity (MMI) as given by Kaka and Atkinson (2004): (1) where PGV is the horizontal component PGV in mm/s and R is hypocentral distance in km. It is assumed when predicting intensity that all sites with unknown soil condition are NEHRP C (firm ground) and are accordingly assigned H/V = 2.38 (Adams and Halchuk, 2003). An exception is made for the POLARIS stations, which are classified by their site-specific H/V ratios (Kaka and Atkinson, 2005b). The CNSN stations are hard-rock sites and are thus assigned H/V= 1.21 (Atkinson, 1993b); this ratio is also assumed for in rock POLARIS stations sited with unknown H/V.
Comparing observed and predicted intensities is difficult for this event because a Canadian internet intensity reporting system is not yet in place comparable to the U.S. Geological Survey “Did You Feel It?” (DYFI) Web site. Canadians can report their experiences in descriptive format at the GSC Web site (http://www.seismo.nrcan.gc.ca/), but these responses need to be processed manually to provide a reliable intensity. This processing is in progress but will take time, as there were about 3,000 responses (mostly in French). Americans in the northeast who felt the earthquake reported their observations on the USGS DYFI site (http://pasadena.wr.usgs.gov/shake/), while some Canadians reported to the USGS international DYFI site (also available at http://pasadena.wr.usgs.gov/shake/). Overall the reported intensities of IV to VI near the epicenter, with intensities of II to IV in northern Maine, New Hampshire, and Vermont, agree well with ShakeMap intensities, which can be viewed at http://www.shakemap.carleton.ca/. In Figure 3, the predicted ShakeMap intensities are compared to MMI based on felt reports submitted online to the automated USGS sites. It is concluded that the agreement between observed and predicted intensity appears to be satisfactory. It is noted that observed MMI data are preliminary in nature; efforts are underway to evaluate intensity in more detail. An interesting observation is that there may be a bias in the Internet intensities to high values at large distances, due to the greater tendency of those who felt the event to report their observations online, as compared to those who did not feel the event.
Given the vast area covered and the relatively small number of seismographic recording stations, ShakeMap offers a reasonable picture of ground shaking across the region. The performance of ShakeMap for this event is very encouraging and can be viewed as a significant step forward in the development of reliable near-real-time seismic information relevant to postearthquake rapid warning systems in eastern Canada. Improvements are needed to reflect specific local variation in ground shaking due to soil conditions.
Ontario ShakeMap development is supported by Ontario Power Generation and the Ontario Research and Development Challenge Fund. The assistance of the Geological Survey of Canada and the U.S. Geological Survey in ShakeMap development work is gratefully acknowledged.