Fisika Bumi Bab 1

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 9

Initial reference models in local earthquake tomography

Abstract. The inverse problem of three-dimensional (3-D) local earthquake tomography is


formulated as a linear approximation to a nonlinear function. Thus the solutions obtained and the
reliability estimates depend on the initial reference model. Inappropriate models may result in
artifacts of significant amplitude. Here, we advocate the application of the same inversion
formalism to determine hypocenters and one-dimensional (l-D) velocity model parameters,
including station corrections, as the first step in the 3-D modeling process. We call the resulting
velocity model the minimum 1-D model. For test purposes, a synthetic data set based on the
velocity structure of the San Andreas fault zone in central California was constructed. Two sets of
3-D tomographic P velocity results were calculated with identical travel time data and identical
inversion parameters. One used an initial 1-D model selected from a priori knowledge of average
crustal velocities, and the other used the minimum 1-D model. Where the data well resolve the
structure, the 3-D image obtained with the minimum 1-D model is much closer to the true model
than the one obtained with the a priori reference model. In zones of poor resolution, there are fewer
artifacts in the 3-D image based on the minimum 1-D model. Although major characteristics of
the 3-D velocity structure are present in both images, proper interpretation of the results obtained
with the a priori 1-D model is seriously compromised by artifacts that distort the image and that
go undetected by either resolution or covariance diagnostics.

Introduction

Seismic tomography applied to the solid Earth is a non-linear process [Pavlis and Booker, 1983].
In general, solutions are obtained by linearization with respect to a reference Earth model [e.g.,
Aki and Lee, 1976; Nolet, 1978]. The tomographic images resulting from such linearized inversion
are dependent on the initial reference models and hypocentral locations [Michael, 1988; Vander
Hilst and Spakman, 1989; VanderHilst et al., 1991]. This dependence, in conjunction with
ambiguities intrinsic to the inversion of seismic data, is an issue that has to be addressed in any
application of seismic tomography [Lees and Shalev, 1992]. Most velocity models of the Earth's
interior are derived by first establishing a simple smooth model that explains some weighted
average of the observations. Subsequently, this initial model is modified until a sufficient degree
of coincidence between the bulk of the observations and predicted values is obtained. This
procedure is followed in seismic tomographic studies, with a one-dimensional (l-D) model usually
serving as the initial reference model for the three-dimensional (3-D) inversion. In many previous
studies the influence of the initial reference 1-D model on the results of the tomographic inversion
has not been fully appreciated. In the following we show that an inappropriate initial reference
model may not only affect the quality of the 3-D image by introducing artifacts, but it may also
influence the confidence calculations by underestimating the uncertainties of the results. To
overcome these problems, Kissling et al. [1984] and Kissling [1988] proposed a two-step
procedure to obtain 3-D tomographic results with minimal dependence on the reference model:
First, the travel time data are jointly inverted to obtain a 1-D tomographic solution, together with
revised hypocenter coordinates and station corrections. We call this new model the "minimum 1-
D model" [Kissling, 1988]. Second, the 3-D tomographic inversion is determined using the
minimum 1-D model as the starting model. In this study we present theoretical arguments for such
an approach and demonstrate its importance for 3-D tomographic results for a simple synthetic
test case mimicking San Andreas Fault structure in central California.
Coupled Hypocenter Velocity Model Problem

The arrival time of a seismic wave generated by an earthquake is a nonlinear function of the station
coordinates (s), the hypocentral parameters (h, including origin time and geographic coordinates),
and the velocity field (m). tob s = f( s,h,m ). (1) In general, neither the true hypocentral parameters
nor the velocity field are known. With arrival times and station 19,635 location being the only
measurable quantities, we cannot solve (1) directly. To proceed, we have to make an educated
guess of the unknown parameters. Using an a priori velocity model, we trace rays from a trial
source location to thereceivers and calculate theoretical arrival times (tcalc). The differences
between the observed and the calculated arrival time, the residual travel time (tres), can be
expanded as functions of the differences {A) between the estimated and the true hypocentral and
velocity parameters. To calculate suitable adjustments (corrections) to the hypocentral and model
parameters, we need to know the dependence of the observed travel times on all parameters. For
hypocenter parameters except the origin time this dependence is strongly nonlinear [e.g., Thurber,
1985], and for velocity parameters it is moderately nonlinear, even in a 1-D model [Pavlis and
Booker, 1983]. Applying a first-order Taylor series expansion to (1), we obtain a linear relationship
between the travel time residual and adjustments to the hypocentral (Ahk) and velocity (Ami)
parameters:

In matrix notation, the coupled hypocenter velocity model parameter relation can be written as

t vector of travel time residuals;


H matrix of partial derivatives of travel time with respect to
hypocentral parameters;
h vector of hypocentral parameter adjustments;
M matrix of partial derivatives of travel time with respect to model parameters;
m vector of model parameter adjustments;
e vector of travel time errors, including contributions from errors in measuring the observed travel
times, errors in t calc due to errors in station coordinates, use of the wrong velocity model and
hypocentral coordinates, and errors caused by the linear approximation;
A matrix of all partial derivatives;
d vector of hypocentral and model parameter adjustments.

Neglecting the effect of (Mm) in equation (3) while locating the earthquakes, for example, has the
potential to introduce systematic errors into the estimated hypocenter locations [Thurber, 1992;
Eberhart-Phillips and Michael, 1993]. Similarly, neglecting of (Hh) in equation (3) may result in
biased velocity parameters [Michael, 1988; VanderHilst and pakman, 1989]. As we demonstrate
below, inclusion of both hypocenter and model parameters in the estimation procedure does not
necessarily guarantee successful recovery of either true distribution. Unless we have "guessed" the
correct hypocentral coordinates, tomographic imaging with local earthquake data demands the
updating of both hypocenter and velocity parameters. We concur with Thurber [1992] that this is
most reliably achieved by solving the coupled hypocenter-velocity model problem, rather than
alternating independent hypocenter and velocity adjustment steps. To reduce the computational
burden of solving the very large system of equations (3), Pavlis and Booker [1980] and Spencer
and Gubbins [1980] independently introduced an algorithm, to separate A into the two smaller
matrices, one containing the hypocenter location information, and one containing the model
parameter information. The reduced form of A pertaining to the velocity model may then be solved
separately to obtain the same solution as if the entire matrix were being inverted. Kissling [1988]
demonstrated the effectiveness of the parameter separation procedure for local earthquake
tomography in a test with data from Long Valley, California. (3) Parameter separation does not,
however, reduce the dependency of the solutions on the reference velocity model or the reference
hypocenters. Generally, the solution to (3) is taken as the least squares solution that minimizes a
weighted combination of the squared SAF Figure 1. Schematic velocity models of the San Andreas
Fault (SAF) in central California. a: Synthetic velocity model of SAF used to calculate synthetic,
yet realistic data sets for the Loma Prieta earthquake series (see text). Different shading denotes
different velocities (values are in kilometers per seconds). Note the lack of vertical velocity
gradients and the strong lateral change in velocity. b: Schematic 1-D model used to approximate
the unknown velocity structure for earthquake location and used as the reference model for 3-D
tomographic inversions. The two reference 1-D models (i.e., the refraction- based and the
minimum 1-D model) have identical layer thicknesses but different layer velocities. Different
shading denotes different velocities. For exact velocity values, see Figure 3.
A
B
error (eYe) and a measure of the velocity model change, typically its Euclidean norm
(mTm)[Spakmart and Nolet, 1988]. Under the assumption of normally distributed errors and
model perturbations, the least squares formulation results in the most likely solution that belongs
to the same family of solutions as the initial reference-model. By seeking a solution in the
neighbourhood of the initial reference model, defects in the reference model may lead to artifacts
in the 3-D tomographic images, particularly when the solution is not refined through model
updating and iteration. Because the parameter space commonly contains several thousand
unknowns, simple iteration schemes may be incapable of avoiding local minima and thus may not
converge to the global minimum. Furthermore, the size of A and/or the application of parameter
separation may make it difficult, if not impossible, to compute the normal diagnostics of model
performance, such as the model and data resolution matrices and model covariance matrix. Thus
potential trade-off between hypocentral parameters and velocities are difficult to identify.

Concept of the Minimum 1-D Model


The chances for successful estimation of the true model using (3) can obviously be improved by
selecting a starting model in the neighborhood of the true model, one for which the linearization
assumption holds. The problem is how to find such a model. We suggest that the natural starting
point is the 1-D velocity model that itself represents the least squares solution to (3). Not only are
the algorithms for developing 1- D models well studied [Crosson, 1976; Ellsworth, 1977; Roecker,
1981; Pavlis and Booker, 1983; Kissling, 1988], but also the smaller dimension of the model space
makes it practical to calculate the full suite of linear diagnostics. We refer to this optimal 1-D
model as the minimum 1-D model. Use of the minimum 1-D model also permits us to test for the
significance of the 3-D model, since the ratio of the variance reduction for the 3-D model to the
residual variance of the 3-D model, properly adjusted for degrees of freedom will follow an F
distribution. This test only makes sense, however, when the reference model is itself the least
squares solution for a model with fewer degrees of freedom, which will not be the case for an
arbitrarily selected starting point. In the minimum 1-D model, the layer velocities will
approximately equal the average velocity in the 3-D tomographic solution within the same depth
range. Note that it is not the spatial average of the model. Rather, the velocities of the 1-D model
approach the average of the 3-D model elements in each layer weighted by the total ray length in
each element. The initial reference model should thus be constructed using similar distribution of
sources and receivers as the data set being analyzed for 3-D structure. The determination of the
minimum 1-D model is a trial and error process that ideally starts with the collection and selection
of a priori information about the subsurface structure. Since this process can lead to ambiguous
results, particularly when more than one a priori 1-D models have been established, several
parameters that control the inversion need to be varied and the corresponding results need to be
evaluated. In the appendix, guidelines for the calculation of minimum 1-D models are provided.
Note that such procedures do not guarantee convergence to a best fit solution. Rather, specific
characteristics of the data set and of the velocity structure need to be implemented in the calculation
process. The results also depend on the effectiveness of the data selection process [Kissling, 1988].
Performance of a Minimum 1-D Model as Initial Reference Model for 3-D Local Earthquake
Tomography To test the minimum 1-D model approach, we have designed a difficult target, based
upon the structure of the central creeping segment of the San Andreas fault in California. In this
region, the fault juxtaposes two distinct crustal velocity structures [Walter and Mooney, 1982] and
contains a tabular zone of extremely low velocity extending to seismogenic depths [Healy and
Peake, 1975; Feng and McEvilly, 1983]. To amplify the effects of the horizontal gradient across
the fault, our synthetic velocity model has no vertical gradient and medium-to-strong horizontal
gradients (Figure l a). Note that our minimum 1-D model with horizontal layering (Figure lb) has
an orthogonal basis to the true model. The suite of stations and earthquakes for the test (Figure 2)
corresponds to the data of Eberhart-Phillips et al. [1990], which are essentially the same as those
analyzed by Lees [1990] and by Foxall et al. [1993]. The reader should bear in mind, however,
that the synthetic structure bears no resemblance to the structure in the Loma Prieta region, aside
from the average crustal velocity outside of the anomalous zone in the synthetic model. Using the
synthetic velocity model of Figure l a for each of the 12,000 source-receiver pairs in the Eberhart-
Phillips et al.[1990] data set (199 earthquakes), we calculated travel times by solving the eikonal
equations with a finite difference algorithm [Vidale, 1990]. Computational rounding errors in the
travel time calculation, typically +0.01 s, provide the only noise in the travel time data. This noise
level corresponds to the standard error of "0 weight" readings for the actual U.S. Geological Survey
(USGS) data [Allen, 1982]. In the data creation phase of the experiment, W. L. Elsworth
established the synthetic model, calculated the true travel times, and computed new hypocenters
using a simplified 1-D model. These trial hypocenters have mislocation errors of several hundred
meters to a few km. The modelling was done by E. Kissling and D. Eberhart-Phillips without prior
knowledge of any characteristics of the synthetic velocity structure. Thus we have attempted to
recreate the actual situation in which the modeler knows only the travel times and station
coordinates. With this synthetic data set, routine tomographic procedures for the inversion of local
earthquake data [Kissling, 1988; Eberhart-Phillips, 1990] were performed using two similar
algorithms based on work by Ellsworth [1977] and Thurber [1981], respectively. Both
tomographic procedures achieve nonlinear inversion by iterating over linear inversion steps with
updating of hypocenters, velocities, and ray paths. The major differences between the two methods
are the approaches used for forward ray tracing (an approximate local 2D ray tracer [Thurber and
Ellsworth, 1980; Kissling, 1988] versus an approximate 3-D ray tracer [Urn and Thurber, 1987]),
the size of the inverse problem that may be handled, and the amount of controlling and testing of
the data [Kissling, 1988; Eberhart-Phillips, 1990]. While the algorithm of Kissling
[1988] employs a velocity model with constant layer velocities, the algorithm of Eberhart-Phillips
[1990] accounts for velocity gradients. Fortunately, the results obtained by the two inversion
routines for the same parameters and assumptions were almost identical. Consequently, in the
following we only show results from the inversion routine of Eberhart-Phillips [1990]. Since this
method includes a 3-D search algorithm for the ray path, large systematic errors in the recognition
of seismic phases due to 3-D structural effects may be excluded. On the basis of refraction seismic
studies [Mooney and Colburn, 1985] and earthquake travel time data Dietz and Ellsworth [1990]
obtained a 1-D velocity model (a priori 1-D model) for the Loma Prieta area. Following the
procedure outlined in the appendix, we calculated a minimum 1-D model for synthetic data with a
layering identical with the a priori 1- D model (Figure 3). We also constructed a second 1-D model
that adopted the a priori velocity values but included optimal station corrections and hypocenters,
just as was done with the minimum 1-D model. This initial part of the inversion process thus
produced two parallel estimates, one for each 1-D model (a priori 1-D model and minimum 1-D
model). The performance of these two 1-D models is very similar, achieving rms misfits of 0.08
and 0.07 s, respectively. The epicenters of each model are systematically repelled from the tabular
low velocity zone of the true model, as would be expected, by an average of 2 km. Focal depths
are biased toward too shallow depth by 2 km in the a priori 1-D model but are unbiased, on average
in the minimum 1-D model. This result is not unexpected, since the a priori model projects velocity
model errors into depth and origin time, on average. Origin times are too early by a median value
of 0.2 s in both models, however. Subsequently, we performed two complete 3-D inversions of the
synthetic travel time data set, one with the a priori 1-D model and one with the minimum 1-D
model. Except for the 1-D model aspect, the control parameters for the two 3-D inversions were
identical. Synthetic quarry blast from 16 surface sources with known coordinates, but unknown
origin times, were also included in the 3-D data set. The two 3-D tomographic images were
calculated by inverting the travel times in an iterative process (six iterations) without using station
corrections. Convergence to a stable minimum was achieved in both cases, with iteration
termination controlled by an F test. The results of these parallel inversions (Figures 4 and 5) are
tomographic images of the same synthetic model (Figure la) for the same original local earthquake
data set. The differences in these 3-D results, therefore, may be fully attributed to the use of
different starting layer velocities and trial hypocenters. The resulting velocity models (Figures 4
and 5) reproducemany of the long- wavelength features of the true model, particularly for the
model nodes at depths of 3 and 7 km. Each model also significantly reduces the hypocentral errors.
Median epicentral errors are 0.6 and 1.0km for the minimum 1-D and a priori 1-D solutions,
respectively, while median depth errors are 0.2 and -0.6 km, and median origin time errors are -
0.01 and -0.08 s, respectively. The resolution matrix for the two cases are virtually identical,
suggesting that there are no important differences in sampling between the two ray sets. The
standard error estimates for the velocity parameters average about 0.08 km/s for both test cases in
areas of good resolution. While thestandard error estimates are equivalent for either initial
reference model, the standard deviation of the actual velocity errors is 0.67 km/s for the a priori 1-
D model and 0.26 km/s for the minimum 1-D model. Overall, the solution obtained with the
minimum 1-D model outperforms the a priori 1-D model results. A comparison of the Figures 4
and 5 reveals a strong dependence of the tomographic results on the initial reference model and
reveals the potential of the minimum 1-D model to reduce artifacts in the solution that distort the
3-D velocity field in these layers for the a priori 1-D model. A quantitative assessment may be
made by comparing the model and true velocity fields in regions with acceptable model resolution.
Using a diagonal resolution value of 0.25 as the cut-off, the correlation coefficient between the
minimum 1-D solution and true model is 0.88, while for the a priori model it is only 0.54. The
correlation is highly significant for both models (p>0.001). The performance of the a priori model
for estimating the relative horizontal velocity pattern in each layer is somewhat more satisfactory
(correlation coefficient is 0.76), although the minimum 1-D model is clearly superior (0.92).
Discussion and Conclusions Solutions to the local earthquake tomography are based on iterative
refinement of a linearized approximation to a nonlinear function. Reliability measures for the least
squares solution are also dependent on the validity of the linearization. We have shown here a clear
dependence of the solution and its diagnostic measures on the initial reference model. Systematic
error in the starting model will not only influence the 3-D results but will also distort the resulting
error estimates. The particular method we have used [Thurber, 1981; Eberhart-Phillips, 1990] to
solve the inverse problem simultaneously updates the hypocenters, ray paths, and travel times at
each iteration step and terminates the iteration only after perturbations became statistically
insignificant. Perhaps another inversion strategy would succeed in finding a better minimum when
the a priori 1-D model is the starting point, but this must be demonstrated. Our results thus illustrate
the potential ambiguity in this specific inverse problem With real data, however, we are forced to
rely upon other tests, including sensitivity tests Lees and Shalev [1992], "checkerboard" tests
Spakman [1991], error and resolution estimates Koch [1993], and smoothing test Sambridge
[1990] to select the most probable results. Cross validation is also highly recommended,
particularly as a means to select the proper damping to regularize the solution [Segall and Du,
1993]. One worrisome problem with both our derived models is the appearance of long-
wavelength artifacts in the lower half of each model that are just of the type that might be mistaken
for structure (Figures 4 and 5). Each model contains a high-velocity body immediately to the
Southwest of the fault zone (below the zone in Figure 4 and to its left in Figure 5) that
approximately corresponds to the distribution of the deepest hypocenters (those that correspond to
the main dipping aftershock zone of the Loma Prieta earthquake [Dietz and Ellsworth, 1990]).
Note that the hypocenters (dots in Figure 5) are slightly out of place due to the velocity pull-up
effect of the high velocity artifact. Because this feature correlates well between our two estimates,
there would be a strong temptation to interpret it as a true part of the 3-D velocity field. It is also
somewhat troubling that a similar high velocity body appears in the tomographic models of the
Loma Prieta region [Eberhart-Phillips et al., 1990; Lees, 1990]. One common link between these
studies is their use of the routinely determined U SGS hypocenters as the initial locations. These
hypocenters were computed using the minimum 1-D model of Dietz and Ellsworth [1990], which
make them our ideal recommendation for the initial reference model. We cannot say, without
further investigation, if the appearance of a similar high-velocity body in our synthetic example
casts doubt on the reality of this body, but caution is clearly suggested. Using a similar data set
from Loma Prieta region, Lees and Shalev [1992] investigated the stability of 3-D tomographic
results for the actual Loma Prieta data set. They found that various initial velocity models, different
parameterization, and either linear or nonlinear inversion procedures all produced similar 3-D
velocity fields. In particular, Lees and Shalev [1992, p. 1838] conclude, that" at least for this
particular data set, the initial starting models used routinely for earthquake location [i.e., the Dietz-
Ellsworth minimum 1-D model] are close enough to the final three-dimensional non linear models
such that first-order perturbation theory is sufficient for deriving the primary signal in the data and
determining the geometry of three-dimensional slowness field". In their study, Lees and Shalev
[1992] employed the stepwise inversion strategy in which adjustments to the velocity model are
determined without perturbing the hypocenters. The hypocenters may be updated in the revised
model when the procedure is iterated. As they always linearize around the same hypocenter
locations and station corrections from the Dietz-Ellsworth minimum 1-D model in the initial
velocity step and regularize the solution with horizontal smoothing, it is not surprising that their
first velocity step moves them toward a common model, regardless of the other assumptions being
tested. Correlation between models derived with different starting models may not, however, be a
good measure of reliability, as artifacts in our synthetic model results (Figures 4 and 5) correlate
well between models. We also note an apparent difference between our result of a strong
dependence on the initial model and those of Pavlis and Booker [1983], who found only a moderate
influence of the starting model on recovery of 1-D velocity models. Their results agree well with
our experience in determining minimum 1-D models in many areas around the world [e.g., Kissling
et al., 1984; Kissling, 1988; Kradolfer, 1989; Hauksson and Jones, 1989; Dietz and Ellsworth,
1990; Eberhart-Phillips, 1990; Scott, 1992; Castillo and Ellsworth, 1993]. The stronger
dependence of the tomographic images presented here on the initial 1-D model suggests that the
local earthquake tomography problem in three dimensions is less well behaved than its 1-D
counterpart. This should not be surprising, as in going from one to three dimensions, we replace
an infinitely long horizontal averaging kernel with one of finite length. Use of the least squares
solution to the 1-D tomography problem with earthquake hypocenters stabilised in that model as
the point of departure for 3-D modeling can therefore do nothing but improve the chances for
successful 3-D modeling.

Step 1. Establishing the a Priori 1-D Model(s)


Obtain all available a priori (prior to the one or three-dimensional inversion) information regarding
the stratification of the area under study (velocities, layer thicknesses, etc.). In general, use
refraction seismic models, simplified where necessary to constant velocity layers. If no controlled-
source seismology models are available, use phase correlations and cross over distances [e.g.,
Deichmann, 1987] from well-recorded earthquakes and/or infer the layered structure from
geologic information. Define the media by several layers of increasing velocity with depth.
Thicknesses of the layers in the upper crust should be about 2 km and in the lower crust about 4 to
5 km. Estimate layer velocities according to a priori information or a general crustal model. In case
of incomplete or inconsistent information or, if the area under consideration confines two or more
distinctly different tectonic provinces, establish several 1-D models. Choose a reference station
with a continuous or nearly continuous record of events. It must be a reliable station, preferably
located toward the center of the network and should not show extreme site effects. The model(s)
and the reference station are called the a priori 1-D model(s). If several significantly different a
priori 1-D models are established the following steps 2 through 5 are repeated for each 1-D model
seperately.

Step 2. Establishing the Geometry and the Velocity Intervals of Potential 1-D Model(s)
Select about 500 of the best events in the data (i.e., those with the most high-quality P arrivals)
that cover the entire area under consideration. Relocate them with routine VELEST using a
damping coefficient of 0.01 for the hypocentral parameters and the station delays and 0.1 for the
velocity parameters. Invert for hypocenters every iteration and for station delays and velocity
parameters every second iteration. Repeat this procedure several times with new (updated)
velocities in the reference 1-D model, with perhaps the new station delays, and with new
hypocenter locations. Repeat the procedure also for reduced number of layers where possible by
combining adjacent layers with similar velocities. Unless clearly indicated by the data, in most
cases it is preferable to avoid low velocity layers, as they normally introduce instabilities. Our
experience suggests that shot or blast data should not be included in the 1-D model inversion.
Rather, such data should be used to set the near-surface velocities, and to test the performance of
the resulting minimum 1-D model when used for locating hypocenters. This countrintuitive
suggestion may be understood by considering that ray paths with both endpoints near the surface
sample, on average, a much more heterogeneous part of the Earth than do ray paths from events in
the seismogenic crust. The goal of this trial and error approach is to establish reasonable geometry
of the crustal model and corresponding intervals for the velocity parameters and station delays. In
addition, this approach provides valuable knowledge about the quality of the data. Procede to the
next step when (1 the earthquake locations, station delays, and velocity values do not vary
significantly in subsequent runs; (2 the total RMS value of all events shows a significant reduction
with respect to the first routine earthquake locations; and (3 the calculated 1-D velocity model and
the set of station corrections make some geological sense (e.g., stations with negative travel time
residuals should lie in local high-velocity areas with respect to the reference station, etc.) and do
not violate a priori information. If all these requirements are satisfied, the result may be called the
"updated a priori 1-D model with corresponding station residuals".

Step 3. Relocation and Final Selection of Events


Relocate all events using the updated a priori 1-D model with station residuals with a routine
location procedure (HYPO71 [Lee and Lahr, 1975], HYPOINVERSE [Klein, 1978],
HYPOELLIPSE [Lahr, 1980]) or with VELEST in the single- event mode (fixing the station and
velocity parameters). Reselect the best (consider gap, number of observations, distance to next
station) 500 or so events that should be well distributed over the volume under investigation. If
more than one such subset of about 500 events can be extracted, proceed for each subset separately
with step 4 but try to obtain similar results.

Step 4. Calculation of Minimum 1-D Model for One Subset


In general terms, repeat step 2 with the updated a priori 1-D model and station residuals and with
a damping of 0.01 for the hypocentral, 0.1 for the station, and 1.0 for the velocity parameters. The
goal of this step is to calculate the 1-D model (velocity parameters and station residuals) that
minimizes the total estimated location errors for a fixed geometry. Test the stability of the result
by systematically and randomly shifting hypocenters and by underdamping the velocity
parameters. If you are pleased with the performance of the solution, fix the updated velocity
parameters by overdamping and calculate thestation residuals. The resulting velocity model with
corresponding station residuals is called "minimum 1-D model".

Step 5. Calculation of Minimum 1-D Model for Several Subsets

If several subsets of 500 events were extracted, test the dependence of your minimum 1-D model
on specific data. Find the 1-D model and station residuals that will best fit the results from all
subsets, mix data from different subsets, and repeat step 4. If the results are unsatisfactory, evaluate
the best 1-D model by the procedure described in step 6.

Step 6. Evaluation of Different Minimum 1-D for Same Area


If several significantly different a priori 1-D models were established and steps 2 through 5 were
successfully completed for each of them, you may base your choice of one minimum 1- D model
on the result of the following performance test: Select all travel time data from quarry blasts or
shots (i.e., from sources of known location) and relocate these events for the different minimum
1-D models without fixing the depth during the location process. If the near-surface velocities for
several station locations are known, compare the station residuals with the differences between the
average layer velocity and the local velocities. Finally, select the minimum 1-D model that best
resembles the a priori information.

You might also like