JP2005114380A - Real-time earthquake risk prediction method - Google Patents

Abstract

PROBLEM TO BE SOLVED: In a conventional method for determining an epicenter using a small number of observation data, (1) accuracy is low, (2) it takes time until a solution is obtained, and (3) earthquakes occurring at different points simultaneously. There were problems such as being considered the same earthquake at the time.
When a seismic wave arrives at one seismic wave observation point 21 at time t ₁ ⁽⁰⁾ and the coordinates of the observation point are (x1, y1, z1), the arrival data at one point and another point arrive. The location of the epicenter and the time (x, y, z, t0) are determined by the least square method using the difference between the measured value t′n + 1 and the theoretical value tn + 1. The solution is determined by simultaneous equations, and the weight W _i is selected empirically, but the distance from the measurement point is a function.
[Selection] Figure 1

Description

  In an environment where the seismic network is at a certain level and data is collected immediately (real-time) or near real-time, the information is communicated to the user after the earthquake occurs and before the seismic wave arrives. The present invention relates to a method of real-time earthquake risk prediction that performs warning or automatic disaster prevention measures.

  The real-time seismic element determination method based on the data from the national seismic observation network is typically the Uredas method (Railway Research Institute report Vol. 19, No. 3, pp. 1-12) and the non-arrival method (Japanese Patent Application No. 2001-308218). (See the official gazette) has been proposed, and verification tests are being conducted. The former feature is that the information on the epicenter is obtained from one observation point, and the latter is that the epicenter parameter is obtained with considerably high accuracy from the data of two or more observation points.

However, in such a real-time earthquake risk prediction method, the conventional method is not sufficient for a specific user to estimate the earthquake risk sufficiently quickly and accurately and to provide maintenance of the target object. There are aspects that cannot be said, and there is a need for an improvement method.
As earthquakes occur, real-time earthquake information begins to be transmitted from seismic observation networks with a certain density. In order to use such information mainly for emergency automatic control that requires high reliability, a system in which a specific user estimates the earthquake risk with sufficient accuracy is required.

The feature of the Yuredas method is that the information on the epicenter is obtained from the seismic waveform at one inspection point. This method has a problem that the accuracy is slightly low and the depth of the epicenter cannot be obtained. In the non-arrival method, the epicenter parameter is obtained with considerably high accuracy using data of two or more observation points. By the way, when determining the epicenter using a plurality of observation data that is not so large at the beginning by the non-arrival method, if there are earthquakes that occurred at different points at the same time, they are regarded as the same earthquake, and there is a large error in the magnitude of the earthquake. There was a possibility of entering.
In the initial arrival and arrival method, after the seismic wave arrived at the first observation point, it was necessary to wait until the seismic wave arrived at the two observation points to determine the epicenter.
The present invention presents a method for solving these misidentification problems. And in order to solve the problem of the method of real-time earthquake risk prediction, the following two terms are provided.

1) Method of high-precision / high-speed determination of hypocenter information 2) Method of real-time high-accuracy earthquake risk estimation on the user side (hereinafter referred to as “one-point-not-arrival method”)

The present invention meets the above-described problems by presenting a new method.
In order to achieve this object, the present invention is configured as described in the claims. That is, the present invention as described in claim 1,
A method of determining the time and position of a hypocenter by seismic observation, comprising a method of real-time seismic risk prediction, characterized by using one observation data and determining a seismic source element by a one-point non-arrival method.
Further, the present invention provides the following, as described in claim 2.
A method for determining the time and position of a hypocenter by seismic observation, wherein the one observation data is used to determine a hypocenter by fusing the one-point arrival / no-arrival method and the Uredas method. Configure the method.
Further, the present invention provides a method as claimed in claim 3.
A method for determining the time and position of the epicenter by seismic observation, which constitutes a method for real-time earthquake risk prediction, characterized by identifying earthquakes that may have occurred at two locations almost simultaneously by vector differences .
Further, the present invention provides the following, as described in claim 4.
Using the method of determining the time and position of the epicenter by seismic observation, the real-time seismic risk prediction method is constructed by processing the input seismic source information in real time and estimating the seismic risk of a specific site.

  The effect obtained by the method of real-time earthquake risk prediction according to the present invention will be described for each claim.

First, in the first aspect of the invention, by solving the simultaneous equations (9a, 9b) by the least square method, the source time and position can be obtained with high accuracy using one observation point data by the one-point arrival / no-arrival method.
In the invention of claim 2, the location of the epicenter can be determined with high accuracy by using one observation data by the fusion method based on the Uredas method and the one-point non-arrival method.
Further, in the invention of claim 3, it is possible to determine whether or not the seismic waves observed at two or more locations almost simultaneously are due to the same earthquake based on the allowable value | Δr |. .
Further, in the invention of claim 4, it is possible to estimate the earthquake risk of a specific site by processing the input epicenter information in real time.

  Examples are shown below.

Method for Determining the Epicenter by One-Point Arrival / De-arrival Method Using One Observation Data FIG. 1 is a diagram showing an earthquake wave arrival model according to the present invention. As shown in FIG. 1, it is assumed that an earthquake wave arrives at _one seismic wave observation point 21 at time t ₁ ⁽⁰⁾ . Let the coordinates of the observation point be (x1, y1, z1). Assume that the time width of this data transmission packet is Δt, and the time t1 is between nΔt and (n + 1) Δt (FIG. 2). The epicenter time and t0, the unknown hypocenter 22 shown in FIG. 1 r (x, y, z ) and the spherical surface of radius Rt1 = (t1-t0) × V p and Ct1 (Figure 3). Here, V _p is the velocity of the P wave. Assuming isotropic properties, the observed seismic wave arrival time data t′n + 1 up to time (n + 1) Δt is

t′n + 1 = (t′1, t′2, t′3...) (1)

It is. t′1 = t ⁽⁰⁾ 1 is a measured value, but t2 ′, t3 ′,... are undecided until this time,

t2 ′> (n + 1) Δt (2)
t3 ′> (n + 1) Δt
・ ・ ・ ・ ・ ・
ti ′> (n + 1) Δt
・ ・ ・ ・ ・ ・
tN ′> (n + 1) Δt

It is only known to satisfy the inequality. Here, N is the total number of observation points. In the example of FIG. 1, FIG. 2 and FIG. 3, it is assumed that the seismic wave has reached by time (n + 1) Δt only at station 21, but when it reaches other stations (j), In Equation (1), t′j = t _j ^(0), and in Equation (2), the index j can be easily extended by removing it from the conditional expression. The same applies when there are three or more observation points.
And to obtain the epicenter parameter, we do as follows. That is, (x1, y1, z1, t ₁ ⁽⁰⁾ ) is

Satisfy the condition of
Also, the theoretical time tj to reach point 2... N is

Meet. The measured value t'n + 1 (formula 1) and the theoretical value tn + 1 shown in the following formula

t′n + 1 = (t1, t2, t3...) (5)

The seismic center position and time (x, y, z, t0) are obtained so that the “difference” between the two points is minimized in the sense of least squares.
The evaluation function ε in this case is

And Here, H (t) is a step function, W1,... Wn are weights and are arbitrary variables.
For simple handling, it is assumed that t ₁ ⁽⁰⁾ is correct.
At this time, the evaluation function ε is

Thus, a solution that minimizes ε is obtained. The solution is as follows.

This equation (8) becomes the following simultaneous equations.

Here, δ _ji is a Kronecker symbol (δ ₁₁ = 0, δ _ji = 0,...).
Then, the choice of the weight W _i is, but those should be determined empirically,
The distance from the measurement point is a function.
In particular, it is appropriate to increase the weights for the eight points around the inspection point and to make the weights considerably smaller (1/3 or 1/4) than these weights.
The magnitude is determined using a distance attenuation formula that is often used.
As described above, by solving the simultaneous equations (9a, 9b) by the method of least squares, the time and position of the epicenter from one observation point data can be obtained (one-point arrival method). As a result, real-time earthquake risk can be predicted at high speed and high accuracy.

(2) Seismic source determination method using one observation data by hybrid method of uredas method and one-point arrival method The uredas method obtains epicenter arrival direction θ and epicenter distance Δ. These errors are defined as dθ and dΔ. The location of the hypocenter calculated from now is ru, and the probability distribution is Prob (ru). On the other hand, the location of the hypocenter by the single point arrival method is rA, and the probability distribution is Prob (ra). This distribution is determined by the S / N (signal-to-noise) ratio of the data in the measurement in the Uredas method, and is mainly a function of the number of observation points and the location in the single arrival method.
In the hybrid method, the results of the Yuredas method and the one-point arrival / absence method are averaged with weights. That is,

Here, the positive number λ ₁ is a weight for both methods and is a value obtained from the probability distribution.
If the variances of the probability distributions of the Yuredas method and the one-point non-arrival method are δ _{u and} δ _A , respectively, the weight λ is determined as follows, for example.

λ = δ ′ _u ⁻¹ / (δ ^′ _u ⁻¹ + δ ^′ _A ⁻¹ ) (11)

The hypocenter time is estimated in the same way.
As described above, the location of the epicenter can be determined with high accuracy using a single observation data by the hybrid method based on the Yuredas method and the single point non-arrival method.

(3) Judgment method of whether or not they are caused by the same earthquake In the single point arrival and arrival method, two or more observation data, or in ordinary earthquake observations, there are three or more earthquake observations. It is necessary to check whether the data corresponds to the same earthquake. Here, the method is presented.
The determination as to whether or not they are caused by the same earthquake is performed each time the seismic wave is detected at a new observation point by the method of the first embodiment or the second embodiment. That is, the location and time of the epicenter by the first station data are (r1, t01), and the location and time of the seismic center by the second station are (r2, t02). The difference between the two vectors

| Δr | = | r1−r2 | + Vp | t01−t02 | (12)

Compared with the allowable value | Δr | _S ,

| Δr | ≦ | Δr | _S ; Same earthquake (13a)

| Δr | >> | Δr | _S ; different earthquakes (13b)

Is determined. | Δr | _S is determined according to the equation (12) with reference to an error by each method. The determination may be made by performing a statistical test (null hypothesis).
If the score is three or more, it is determined whether or not the group occupies the majority method.
As described above, it is possible to determine whether or not they are caused by the same earthquake, which was measured at two locations almost simultaneously with the allowable value | Δr | as a reference.

(4) Method for Estimating Seismic Risk at a Specific Site in Real Time 4.1 Risk Estimation Method A detailed description of the present invention will be given using FIG. In this example, the case where the epicenter S (symbol 1) of the earthquake and the user U (symbol 6 (U)) are separated from each other to some extent (assuming that the average interval between observation networks is L and 3L to 4L) will be described.
Let Ta (allowed time) be the minimum time required to perform a certain emergency disaster response, and Tp (preparatory time) be the time required for preparation. It can be considered that Ta is approximately 1 to several seconds and Tp is approximately 10 seconds. Rather, it seems that devices with such characteristics are suitable for using real-time earthquake information. The user performs emergency control at Ta before the arrival of the main movement, and starts a preliminary movement at Tp. However, it should be noted that the pre-arrival information may be smaller than the minimum time Ta when an earthquake occurs most recently.
In such a case, emergency measures based on seismometer data on the user side (specific site) will be taken.
Hereinafter, each observation data processing method will be described.
(A) When all observation data can be handled in combination FIG. 5 is a diagram showing a calculation algorithm for risk (seismic value, time). As shown in FIG. 5, it is assumed that observation data and epicenter information are delivered sequentially. Data in the user zone (attention zone P _Z , operation zone R _Z ) (which may be network or user side, but merged). As a user area data D (u), the calculation algorithm for risk (seismic value, time) in FIG. 5 will be described below.
Assume that the i-th epicenter information Pi is issued from the center. The accuracy is Prob (Pi). The risk level (seismic value, time) at that time is calculated. Let Ri be the probability distribution and Prob (Ri).
P-wave until the scheduled time to arrive at P _Z: checking the information transmitted on the transmission side of the algorithm independent calculation method. That is, wide area data D (r)
D (u) including is used to select the weight W for obtaining the residual by slightly increasing the weight on the user side data.
When the P-wave has entered the P _Z:
Check and correct the parameters (eg, the inventor (Fujinawa etc.) (see Japanese Patent Application No. 2001-257765))
A method for checking this parameter will be described below.
In Japanese Patent Application No. 2001-257765, correction is performed using data on the user side, but in the present invention, the user side and the wide-area observation data are handled in a unified manner. For this reason, considering the accuracy of each measurement data,
(A) Whether the observation is performed correctly.
(B) It is always determined whether or not the signal is lower than the background noise level. When the user-side data and the observation network data include waveform data, the original determination using the waveform is performed.
(C) An example of accuracy ranking is shown below.
Rank A: 99% -100%
Rank B: 95% -99%
Rank C: 90% -95%
Rank D: 90% or less.

        This rank table is an example and has a property determined by the risk manager in consideration of the characteristics of the target system.

It is determined that the event is determined to be an earthquake at rank α, and the parameter Ω _j enters (ω _j + iΔω _j , ω _j + (i + 1) Δω _j ) at rank α _ji .

(D) Event determination method
(D-1) When there is one station:
* Depends on noise identification method.

・ Distinction by amplitude
・ Waveform discrimination (P, S wave discrimination, utilization of 3 component waveforms)
(D-2) When measuring point 2 or more:
* Depends on noise identification method.
-Discrimination by amplitude-Waveform discrimination (P / S wave discrimination, utilization of 3 component waveform relationship)
* Consistency check method
The following determination is performed individually or in combination, and is performed by statistical processing.

            -Time axis: For the same event that cannot be determined as an earthquake, seismic measurement is performed based on the travel time curve and distance attenuation formula each time new data is acquired.

・ Position judgment
・ Waveform ・ When the degree of risk Ri is above a certain threshold, a preparation action signal is issued.
When the P wave enters the danger zone RZ (7 in FIG. 4):
・ Detected at two stations in the RZ and outputs a control start signal when the seismic risk is above a certain level.
The above is repeated until the S wave arrives.

In addition, it is clear that an analogy can be made when the S wave is within the attention region O _Z (8 in FIG. 4).
(B) When user data is handled independently At a certain time t _i , it is assumed that the event has a risk rank β _j and its accuracy is δ _ij . At this time, the estimated value of the parameter Ω _k at the user's site is ω _k ⁽ⁱ⁾ and the accuracy is δ _k ⁽ⁱ⁾ .
On the other hand, the measured values based on the user data are (ω _k ^{* (i)} , δ _k ^{* (i)} ), respectively.
Statistically test whether the hypothesis of ω _k ⁽ⁱ⁾ = ω _k ^{* (i)} holds.
If established:
Let the accuracy be Δ _k ⁽ⁱ⁾ . When the accuracy limit is Δ _C , when Δ _k ⁽ⁱ⁾ is larger than the accuracy limit, the correction risk rank is defined as β _j , and the accuracy is defined as δ ′ _ij = δ _ij × Δ _k ⁽ⁱ⁾ .

If not established:
For example, it is determined as follows according to the characteristics of system management.

(2.1) Selection on the safety side:
(Ω _k ^{* (i)} , ω _k ⁽ⁱ⁾ )) is assumed as a control estimated value.

(2.2) Intermediate selection:
Let mean (ω _k ^{* (i)} , ω _k ⁽ⁱ⁾ ) be the estimated value for control.

(2.3) Maximum side selection:
Let max (ω _k ⁽ⁱ⁾ , ω _k ^{* (i)} ) be an estimated value for control.

(2.4) General expressions:
ε × ω _k ⁽ⁱ⁾ + (1−ε) × ω _k ^{* (i)} ) is set as an estimated value for control. ε is a positive number between 0 and 1, and is determined according to the purpose.

(2.5) More general expressions:
f = f (ω _k ⁽ⁱ⁾ , ω _k ^{* (i)} , ε)
ε = 0: f = ω _k ⁽ⁱ⁾
ε = 1: f = ω _k ^{* (i)}

4.2 User-side seismic network
Using the above results, the facility is located at the user point U, and an earthquake observation network is designed to protect this facility from earthquakes. Let ρ _{r be} the distance from the user point to R _Z , and ρ _{r ′ be} the distance to P _Z. If the warning time is Tr and the preparation time is Tp,

ρ _r = Vs × Tr (14)
It is determined.
The approximate values are 3.1 km with Vs = 3.1 km / s and Tr = 1s.
By placing a certain seismograph to P _Z, and to perform the verification of the official announcement data. It is assumed that the seismometers are arranged in a circle centered on U and evenly at an azimuth angle θ. If it is the nth square,
ρ _r / ρ _{r ′} = cos θ (15)
θ = 2π / n
It becomes.
For example, when θ = 60 °, n = 6, ρ _r / ρ _{r ′} = 1/2, and ρ _{r ′} ≈6.2 km, so that it can be installed at a substantially reasonable distance. Incidentally, when n = 4, ρ _{r ′} = ∞, which is an almost impractical distance.

In this way, at least two points can be prepared for an earthquake arrival in any direction of arrival. In this case, one preparation time Tp ⁽¹⁾ is
√3 / 2sec ≦ Tp ⁽¹⁾ ≦ 1sec
Also, the two preparation times Tp ⁽²⁾ are
0 ≦ Tp ⁽²⁾ ≦ √3 / 2sec
It becomes the range.
Furthermore, when it is desired to set Tp ⁽²⁾ to Tp ′,
Δρ _r = Tp ′ × V _S
Just spread. Accordingly, ρ _{r ′} is increased by Δ2ρ _r .

4.3 Evaluation of estimated earthquake risk level It is often necessary to determine whether the estimated value is valid. Explain the basic idea. FIG. 6 shows an evaluation graph for estimating the intensity of an earthquake (usually measured seismic intensity is used, but parameters are selected depending on the object). As shown in FIG. 6, the wide funnel-shaped area is the acceptable range. The minimum seismic intensity and maximum non-damaged seismic intensity vary depending on the target.

On the other hand, FIG. 7 is a diagram showing the arrival time which is another index of the earthquake risk R in the evaluation method of the predicted value of the earthquake risk. In FIG. 7, the minimum time required to perform a certain disaster prevention measure is T _r (minimum time to take an emergency procedure). In addition, the time required for taking the preparatory action is defined as T _p (minimum time for preparation action).
Assuming that the correct arrival time is T _t , first, unless T _t ≧ T _{tr, it will} not be in time.
In other words, the limited measure using this information cannot be performed.
If the estimated value is Te and the error is ΔT,

T _e = T _t + ΔT (16)
And therefore

T _e −ΔT ≧ T _r , T _e ≧ T _r + ΔT (17)
Is required. In the case of preparatory actions,

T _e ≧ T _p + ΔT (18)
Is a necessary condition.
In addition, the estimated value and the true value must match within a certain range.
That is,

T _t −ΔT ₁ ≦ T _e ≦ T _t + ΔT ₂ (19)
Here, ΔT ₁ and ΔT ₂ are maximum errors with respect to the lower limit and the upper limit.
As described above, it is possible to estimate the earthquake risk of a specific site by processing the input source information in real time.

It is a figure which shows the seismic wave arrival model which concerns on this invention. It is explanatory drawing of the seismic wave shown in FIG. It is explanatory drawing of the seismic risk estimation method by the seismic wave shown in FIG. It is a detailed explanatory view about P wave arrival by the seismic wave shown in FIG. It is explanatory drawing of the risk (seismic source value, time) calculation algorithm shown in FIG. It is detailed explanatory drawing about the evaluation graph of the estimation of the intensity | strength of an earthquake. It is explanatory drawing which shows the evaluation method of the predicted value of the earthquake risk shown in FIG.

Explanation of symbols

1 Epicenter 2-5 Seismic observation point 6 near the epicenter 6 (u) User's location 7 Envelope (Rz) of wave front of main motion at the time of issuing emergency command
8 Waveform envelope (Oz) of the main motion at the time of issuing the preparation command
10-15 Seismic observation point near the user 21 Seismic observation point (first seismic wave arrival point)
22 Position of wave front when seismic wave for a certain time reaches first observation point 21 to 32 Observation point where seismic wave has not yet reached

Claims (4)

A method for determining the time and position of a hypocenter by seismic observation, wherein a single seismic data is used to determine a source element by a single point arrival / departure method. A method for determining the time and position of a hypocenter by seismic observation, wherein the one observation data is used to determine a hypocenter by fusing the one-point arrival / no-arrival method and the Uredas method. the method of. A method for determining the time and location of a hypocenter by seismic observation, wherein the earthquakes that may have occurred at two locations at the same time are identified by vector differences, and a method for predicting real-time seismic risk. A method of real-time earthquake risk prediction, which uses the method of determining the time and position of the epicenter by seismic observation and estimates the earthquake risk of a specific site by processing input source information in real time.

Images