Estimation of shaking over the regional extent for an earthquake in California is obtained by the spatial interpolation of the measured ground motions with geologically based frequency and amplitude-dependent site corrections. We use the California Site Condition Map (California Geological Survey, CGS) maps of National Earthquake Hazard Reduction Program (NEHRP) classification site conditions as the basis for our site corrections. These site condition maps have coverage throughout the state at 1:250,000 scale (Wills et al., 2000). We use the amplification factors of Borcherdt et al. (1994). In addition, ShakeMap ground motions in regions of sparse station spacing are estimated using ground motion regression, initially from a point location at the epicenter. Later, as information about fault dimensions became available (in the form of aftershocks, source rupture models, and observed surface slip), the fault location and rupture dimensions are used as the basis for ground motion estimation.
First, peak ground motion parameters are recovered for each station and associated with a particular earthquake origin time and epicenter. We then create a coarse, uniformly spaced grid of 30-km spaced "phantom" stations. Peak ground motions and spectral acceleration values are assigned to each coarse grid point using a ground motion attenuation relationship for rock sites given the magnitude of the earthquake and distance to each grid point. In practice, we apply a static correction to the amplitudes of the regression by using the network-determined magnitude, predicting the observed amplitudes, and correcting for an amplitude bias term between the predictions and the data.
ShakeMap uses estimated ground motions in areas with sparse seismic station coverage. For earthquakes of magnitude 5.3 and greater, the southern California ShakeMap system uses the Boore-Joyner-Fumal (SRL, V68, #1, 1997) attenuation relation for peak ground acceleration (PGA) and pseudo-spectral acceleration (PSA), and PGV is computed from 1Hz PSA using the Newnark-Hall (1982) equation. For smaller earthquakes, we use a refit (by Vince Quitoriano) of the BJF relationship to a southern California-specific database of small earthquakes.
Site corrections are then used to interpolate from ground motions recorded on a fairly sparse, non-uniformly spaced network of stations to maps showing spatially continuous functions (i.e., contours). Prior to interpolation, we reduce the ground motion amplitudes to a common reference, in this case bedrock motions. Peak ground motion amplitudes from the seismic stations are corrected to rock site conditions; and the observations (corrected to rock) and the coarse phantom stations (computed for rock) are then interpolated to a fine rock site grid (roughly 1.6-km spacing). We scale the peak acceleration (PGA) amplitude with the Borcherdt et al. (1994) short-period amplification factors while the peak ground velocity (PGV) values are corrected with the mid-period factors. Response spectral values are scaled by the short-period factors at 0.3 sec, and by the mid-period response at 1.0 and 3.0 seconds. The site correction procedure is applied so that the original data values are returned at each station; hence, the actual recorded motions are preserved in the process and the final contours reflect the observations wherever they exist.
Next, the interpolated rock grid is corrected at each point for local site amplification and instrumental intensity map is generated by relating the peak ground acceleration or velocity at each grid point to intensity as described by Wald et al. (1999). This fine grid is saved and exported to the file "grid.xyz". A continuous surface is also fit to the fine grid to produce the contour maps and GIS formatted maps.
Values of the peak amplitudes at the ShakeMap map grid nodes. File is ASCII text in the following format:
First line is a header, space-delimited, with all of the following:
<name/CUSPID of event> <mag> <epicentral lat> <epicentral lon> <MMM DD YYYY> <HH:MM:SS timezone> <W bound> <S bound> <E bound> <N bound> (Process time: <time>) <Location String>The first 'time' field is the time of the event. 'Process time' is the time this file was last updated.
For large or historic earthquakes the "Location String" will usually be the name of the earthquake, otherwise it will be something of the form "12.1 mi. SSW of Carpinteria, CA"
The remaining lines are of the form:
<lon> <lat> <pga> <pgv> <mmi> <psa03> <psa10> <psa30> pga = peak ground motion (acceleration, as percent of g) pgv = peak ground motion (velocity in cm/s) mmi = estimated instrumental intensity psa03 = spectral acceleration at 0.3 s period, 5% damping (percent of g) psa10 = spectral acceleration at 1.0 s period, 5% damping (percent of g) psa30 = spectral acceleration at 3.0 s period, 5% damping (percent of g)
Specifies the earthquake and ShakeMap parameters, and values of the peak amplitudes at the ShakeMap map grid nodes. File is XML text.
These are automatic computer generated maps and have not necessarily been checked by human oversight, so they may contain errors. Further, the input data is raw and unchecked, and may contain errors.
Contours can be misleading since data gaps may exist. Caution should be used in deciding which features in the contour patterns are required by the data. Ground motions and intensities can vary greatly over small distances, so these maps are only approximate; when maps are enlarged beyond the limits of the original data in an effort to show small areas, the maps are unreliable.
These maps are preliminary in nature and will be updated as data arrives from distributed sources.
The estimated intensity map is derived from ground motions recorded by seismographs and represents Modified Mercalli Intensities (MMI's) that are likely to have been associated with the ground motions. Unlike conventional Modified Mercalli Intensities, the estimated intensities are not based on observations of the earthquake effects on people or structures.
Locations within the same intensity area will not necessarily experience the same level of damage since damage depends heavily on the type of structure, the nature of the construction, and the details of the ground motion at that site. For this reason more or less damage than described in the MMI scale may occur.
Large earthquakes can generate very long-period ground motions that can cause damage at great distances from the epicenter; although the intensity estimated from the ground motions may be small, significant effects to large structures (bridges, tall buildings, storage tanks) may be notable.
The ground motion levels and descriptions associated with each intensity value are based on recent damaging earthquakes. There may be revisions in these parameters as more data become available or from further improvements in methodology.
First line is a header containing <name/CUSPID of event> <mag> <epicentral lat> <epicentral lon> <MMM DD YYYY> <HH:MM:SS timezone> <W bound> <S bound> <E bound> <N bound> (Process time: <time>) <Location String>
The remaining lines contain <lon> <lat> <pga> <pgv> <mmi> <psa03> <psa10> <psa30>