S24B-01
Determination of earthquake early warning parameters, τc and Pd
We explore a practical approach to earthquake early warning in southern California, Taiwan, and Japan by determining a ground-motion period parameter τc and a high-pass filtered displacement amplitude parameter Pd from the initial 3 sec of the P waveforms. At a given site, we estimate the magnitude of an event from τc and the peak ground-motion velocity (PGV) from Pd. The incoming 3-component signals are recursively converted to ground acceleration, velocity and displacement. The displacements are recursively filtered with a one-way Butterworth high-pass filter with a cutoff frequency of 0.075 Hz, and a P-wave trigger is constantly monitored. When a trigger occurs, τc and Pd are computed. We found the relationships between τc and magnitude (M) for southern California, Taiwan, and Japan, and between Pd and PGV for both southern California and Taiwan. Those relationships can be used to detect the occurrence of a major earthquake and provide onsite warning in the area around the station where onset of strong ground motion is expected within seconds after the arrival of the P-wave. When the station density is high, the methods can be applied to multi-station data to increase the robustness of onsite early warning and to add the regional warning approach. In an ideal situation, such warnings would be available within 10 sec of the origin time of a large earthquake whose subsequent ground motion may last for tens of seconds.
S24B-02
A Bayesian Method for the Real Time Estimation of Magnitude From Early P- and S-wave Displacement Peaks
It has been shown that the initial portion of P- and S-wave signals can provide information about the final earthquake magnitude in a wide magnitude range. This observation opens the perspective for the real-time determination of earthquake source parameters. In this paper we describe a probabilistic, evolutionary approach for the real time magnitude estimation which can have a potential use in earthquake early warning. The technique is based on empirical prediction laws correlating the low-frequency peak ground displacement measured in a few seconds after the P and/or S phase arrival (PD) and the final event magnitude. The evidence for such a correlation has been found through the analyisis of 256 shallow crustal events in the magnitude range MJMA 4-7.1 located over the entire Japanese archipelago. The PD measured in a 2-sec window from the first P-phase arrival correlates with magnitude in the range M 4- 6.5. While a possible saturation effect above M 6.5 is observed, it is no more evident in an enlarged window of 4- sec. The scaling of S peaks with magnitude is instead observed also at smaller time lapses (i.e., 1-sec) after the first S-arrival. We developed a technique to estimate the probability density function (PDF) of magnitude, at each time step after the event origin. The predicted magnitude value correspond to the maximum of PDF, while its uncertainty is given by the 95% confidence bound. The method has been applied to the 2007 (MJMA 6.9) Noto-Hanto and 1995 (MJMA 7.3) Kobe earthquakes. Results of this study can be summarized as follow: - The probabilistic algorithm based on predictive model based on PD vs final magnitude is able to provide fast and robust estimation of the final magnitude. - The information available after few seconds from the first detection of P phase at the network is useful to predict the peak ground motion at a given regional target with uncertainties which are comparable to those derived from the attenuation law. - The near-source, S-phase data can be used jointly with P data for regional early warning purposes, thus increasing the magnitude estimation accuracy and reliability. http://www.saferproject.net/
S24B-03
A Kinematic and Dynamic Investigation of Early Magnitude Estimation
Recent seismological analysis has shown a general scaling between final magnitude and the averaged first few seconds of ground motion recorded by seismic arrays. One possible explanation for this scaling is that an earthquake's final size is somehow related to it's initiation. Another possible hypothesis is that the initial recorded ground motion relates to a substantial section of the overall fault size. By using simple dynamic and kinematic models we investigate both theories. Our dynamic models show that scaling is valid only over a narrow magnitude range for a particular class of rupture. Simple kinematic models on the other hand provide a more fruitful result whereby scaling between magnitude and a simplified peak ground displacement is similar to that observed in real data.
S24B-04
Does The Slip In The Early Steps Of The Rupture Scale With The Final Magnitude Of The Event ?
Earthquake early warning systems are real-time monitoring infrastructures designed to provide a rapid notification of the potential effects of an impending earthquake at critical distant targets. The ability in predicting if an earthquake is going to be large from the first few seconds on the accelerograms is crucial in resolving the trade-off between the lead-time of the system and the accuracy in the evaluation of the missed/false alarm probabilities. Here we investigate the scaling of the magnitude with the ground velocity integral computed from the first 4s of P wave and 2s of S wave. We come up with such a new parameter because it is expected to be more stable than peaks and it is directly linked to the energy radiated by the propagating rupture. Therefore it can give insights into the physics of the early stages of the rupture. We analyze more than 2000 high-resolution strong motion records from the Japanese databases K-net and Kik-net with earthquake magnitude ranging between 4 and 7. After normalization by the hypocentral distance, assuming constant stress drop and computing effective rupture areas from the Brune model, we draw the average slip corresponding to the early stages of the rupture against the final magnitude. Despite of the dispersion of data, we found that standard scaling relations still hold in the whole investigated magnitude range. Results of our analysis suggest that the initial slip is an indicator of the rupture size and the total moment in the explored magnitude range implying that the rise time has to be negligible at this time scale.
S24B-05
Estimating the Fault Rupture Geometry in Real Time for the 2007 Noto Hanto and 2007 Niigata-ken Chuetsu-oki Earthquakes
In October 2007, the Japan Meteorological Agency starts providing earthquake early warning publicly. They broadcast the early warning message on TV and radio in real time, so that public people can take damage- mitigating actions before strong shaking arrives. The research topic of earthquake early warning attracts public people as well as seismologists and engineers. Japan had two major earthquakes in 2007 which caused significant damage on buildings and infrastructures. The 2007 Noto Hanto earthquake (Mw6.7) occurred on March 25 at off Noto peninsula near west coast of this peninsula. After 4 months, July 16, the 2007 Niigata-ken Chuetsu-oki earthquake (Mw6.7) hit near the west coast of Honshu Island, which killed 11 people. This research applies earthquake early warning algorithms to the dataset of these earthquakes. We applied two different approaches to determine the ongoing fault rupture geometry from accelerograms. Based on these methods, the current direction of the fault trace the current fault rupture length, and the current fault extent can be estimated. The near-source/far-source discriminant function can tell the probability that the station is located in the near- source region based on the amplitude of vertical acceleration and horizontal velocity. The estimation of near- source station agreed with fault model very well. Stations on the fault model have high probability that the station is near-source. The real-time rupture geometry estimation agrees with the actual earthquake rupture geometry quite well. For the Noto Hanto earthquake dataset, the rupture direction can be estimated 12 seconds after the event onset and the final solution is achieved after 15 seconds. The rupture estimate for the Niigata-ken Chuetsu-oki earthquake dataset is more difficult to converge since the station distribution is limited in one side of the fault. These methodologies to characterize rupture geometry in real time were originally designed for larger earthquakes with larger rupture dimensions, but it is shown that they also work well for the size of these earthquakes (Mw 6.7).
S24B-06
Loss Estimates in Scenario Mode may Help to Harden Those in Real-Time: Repeat of the 1356 Basel Earthquake as an Example
Estimating losses within minutes after earthquakes worldwide can be difficult because of error sources, unexpected issues and the pressure of time. Therefore, we propose to compile a catalog of scenario loss estimates for locations where future earthquakes may be expected. These scenarios could then be consulted in real-time to assist with estimating the order of magnitude of the losses and identifying local problems that may exist. As an example, we present scenario loss estimates for a repeat of the 1356 M6.9 earthquake near Basel, Switzerland, which was the largest and most devastating historic earthquake north of the Alps. The losses we estimated are defined as average damage to buildings for all settlements affected (intensity equal to and larger than V on the modified Mercalli scale), number of injured and number of fatalities. The results in which we have most confidence are the ratio of human losses within the city of Basel to that outside of it, because any errors in the absolute values of the loss estimates tend to cancel. For a repeat of the 1356 earthquake with assumed magnitude of 6.9 and three possible epicenters (distances 6, 10, and 15 km from Basel), we calculated that the countryside would sustain 2 to 4 times the number of human losses than the city itself. In cases of smaller earthquakes (M6.5 and 6.0), at the same distances from Basel, the countryside will sustain 2 to 16 times the losses of the city. With the current population, the number of fatalities is expected to lie in the range of 6,000 to 22,000 for M6.9, 1,700 to 8,400 for an M6.5, and 160 to 1,400 for an M6.0. However, these values should be taken as preliminary, pending recalculation with recent information on building stock properties. Our preliminary estimates suggest that the number of persons requiring hospitalization may range from 6,000 to 8,000 in case of an M6.9, from 2,300 to 4,500 in case of an M6.5, and from 350 to 1,000, in case of an M6.0 earthquake. The portion of losses sustained by the neighboring countries, France and Germany are estimated to range from 0% to 5% of the total losses, strongly depending on the selected epicenter and magnitude. http://www.wapmerr.org