The Role of Fluids in Promoting Seismic Activity in Active Spreading Centers of the Salton Trough, California, USA

We interpret seismic activity in the active spreading centers of the Salton Trough to indicate 1) a magmatic intrusion in the lower crust beneath the active Brawly, Cerro Prieto, Imperial, Elsinore, and San Jacinto fault systems; and 2) fluids in the upper crust that have been released from that magmatic body. The absence of a magmatic body and fluids at the location of fossil spreading centers along the Sand Hill and Algodones faults ndicated by little or no seismic activity in those areas. We show several lines of evidence to point out that both melt and fluids related to the seismic activity. In particular, receiver function analysis, Vp/Vs ratios, and tomographic data reveal low velocity zones coincide with the location of the active spreading centers. High Vp/Vs ratios and low velocity zones in the lower crust and upper mantle attributed to melt inclusion, while low Vp/Vs ratios in the upper crust are attributed water inclusions. Frequency-magnitude distributions characterized by high b-values in southern California; high b-values have also been associated with crustal fluids. A crustal scale model developed from the receiver functions, gravity, and magnetic data supports the existence of a magmatic intrusion within about 20 km of the surface southwest of the Salton Sea, that intrusion extends for 70 km in a SW-NE direction.


Introduction
Mantle fluids have been interpreted to be associated with the San Andreas fault system in central and southern California [1].The most promising source of mantle derived fluids comes from mantle wedge material, which was potentially hydrated and serpentinized during subduction prior to the formation of the San Andreas fault and which would subsequently dehydrate after the plate boundary along California transitioned from subduction to strike slip tectonics [2].Fluids appear to enter the brittle fault zone at or near lithostatic pressure and thus may contribute to fault weakening because of high fluid pressure at seismogenic depths [3,4].Seismicity and tomography suggest that for seismic events to occur in the lower crust, fluids as well as a rheologically strong lower crustal layer must be present in order to restrict the rise of fluids.There has been little consideration of whether ponded magma and exsolved fluids at the base of the crust could similarly trigger lower crustal events [5].One of the few studies to point out that high pore pressure would be necessary to trigger such lower crustal events has been that of [6], who studied lower crustal earthquakes in the East African Rift and found no difference in rupture characteristics between upper and lower crustal events (based on m b -M w relation).This suggests that similar fluid triggering processes may occur throughout the crust as proposed by [5].
In this study, we examine the relationship between fluids and seismicity in active spreading centers (Brawly, Cerro Prieto, Imperial, Elsinore, and San Jacinto) of the Salton Trough and compare those areas with areas of little or no earthquake activity in nearby fossil spreading centers (Sand Hill and Algodones).We used data from the southern California broadband seismic network and data collected by the EarthScope US Array seismic deployment for our analysis.We also obtained data from existing gravity databases at the University of Texas, El Paso (UTEP) (http://www.research.utep.edu/paces)-PanAmerican Center of Earth and Environmental Studies-(PACES) that is currently hosted at the CYBER-ShARE Center of Excellence at UTEP.Aeromagnetic data were obtained from the US Geological Survey.The results of this study indicate a magmatic intrusion at a depth of about 20 km to the southwest of Salton Sea.That intrusion extends for 70 km in SW-NE direction.

Tectonic Setting
The Salton Trough (Figures 1 and 2) is a highly extended terrane at the northern extent of sea-floor spread-  ing in the Gulf of California [7].The area has high heat flow and inferred to be in the early stages of transitioning from continental crust to oceanic crust [8].
The Salton Trough results from northward progression of California [8][9][10].The initial opening of the Gulf of California occurred about 12 -10 Ma, shortly after subduction ceased along the continental margin of Mexico [9,11].The Salton Trough apparently represents a present day onshore analog to theses Gulf of California evolving spreading centers [10].
The Salton Trough region is seismically active [12] with most of the seismicity in the region occurring along the Imperial and Brawley fault zones (Figure 2, [8]).The Cerro Prieto, Imperial, Brawley, and San Andreas faults form an echelon series of east-stepping, right lateral faults linked by pull apart basins referred to as the Cerro Prieto, Brawley, and Salton Buttes (Figure 2) spreading centers by [10].These spreading centers have probably changed position more than once.Thus, for example, the parallel Sand Hills and Algodones faults inferred to represent fossil transforms or spreading centers parallel to, but approximately 45 km east of the Imperial fault zone [9,13].
The Imperial Valley-Salton Trough is highly active seismically [8].Large earthquakes have occurred on Imperial fault in 1940 (Ms 7.1) and in 1979 (Ms 6.9).Reference [8] suggested based on geodetic, seismic, and heat flow data that the imperial and Brawley fault systems are young (3000 to 100,000 years old) and have migrated northwest into Imperial Valley.The southern most part of the active San Andreas fault extends northwest from the southeast corner of the Saton Trough [12].Further northeast, smaller Late Cenozoic strike slip faults also known from surface exposure and these likely represent parts of the evolving transform system [7].We suppose it is possible to classify The Cerro Prieto, Imperial, Brawley, and San Andreas as seisomogenic fault based on the seismic activity along these faults.

Receiver Functions
A receiver function is the seismic response of the earth beneath a seismic station to an incoming P-wave.In particular, a receiver function maps P-to-S converted energy due to impedance contrasts (i.e., changes in velocity and density) in the earth.First-order information about the crustal structure can be derived from the radial receiver function, which is dominated by P-to-S converted energy from velocity discontinuities in the crust and upper mantle [14].Receiver functions can be used to determine crustal thickness, Vp/Vs ratios, and the lateral variation of the crustal thickness [15].In receiver function estimation, the foundation of the iterative deconvolution approach is least squares minimization of the difference between the observed horizontal component seismogram and predicted signal generated by convolution of an iteratively updated spike train with the vertical component seismogram [16].The iterative time-domain approach has several advantages, such as the ability to estimate the percent fit and the long period stability by a priori constructing the deconvolution as a sum of Gaussian pulses [16].We compute receiver functions using the iterative time deconvolution with Gaussian width (Ga) factors of 2.5,

 
1.75, and 1 which is equivalent to applying low pass filters with cutoff frequencies of 1.2, 0.9, and 0.5 Hz, respectively.
components to radial and transverse directions.From the waveform data, we computed the radial and transverse receiver functions using iterative deconvolution, keeping data with an 80% or greater fit.We also manually inspected each radial receiver function to ensure quality.We then stacked the radial receiver functions using the approach of [15].For stacking we ran the H-K stacking codes, we set the weight to be (0.3/0.5/0.2) (0.4/0.3/0.3) and (0.5/0.4/0.1).
We collected data from 27 broadband seismograph stations (listed in Table 1 and shown in Figures 3 and 4) in the vicinity of the Salton Trough, which recorded data from 2000 to 2009.
Specifically, we collected broadband seismic waveform data for teleseismic earthquakes with M > 5.5.Our approach of analysis consist of several steps, the first step is copying receiver function for stations into different directories, stepping through each year and 100 day Julian day increment.Loop over Julian days 000, 100, 200, 300 for each K (0, 1, 2, 3).
The time separation between Ps and P can be used to estimate crustal thickness H, given the average crustal velocities of Vs and Vp: Data available directly from the Incorporated Research Institutes for Seismology (IRIS), Data Management Center (DMC), using the Standing Order of Data (SOD), which allowed for automated rotation of the horizontal where p is the ray parameter of the incident wave [15].Of concern is the trade-off between the crustal thickness and crustal velocities, since t Ps represents the differential   travel time of S with respect to P in the crust.
In our analysis, we ran script to calculate the ray parameters for all events and stations in the study area.
The dependence of H on Vp is not as strong as the dependence on Vs or more precisely on the Vp/Vs ratio K.For example using a Vp of 6.
We contour the crustal thickness and Vp/Vs ratios (Figures 3 and 4 respectively) using a minimum-curvature algorithm to interpolate values to a rectangular grid.Moho dome shape beneath the area of active upper crust extension suggested being primarily the product of magmatic activity in the lower crust and upper mantle following the interpretation of [20] for a similar feature in Death Valley, California.
Reference [20] suggests that the doming of the lower crust of Death Valley accommodated by the addition of magma to the lower crust.
The Moho is about 20 km deep southwest of the Salton Sea and increases to 32 km east of the Salton Sea.The average Vp/Vs ratio is 1.80 in the study area in general; this ratio increases to the south and southeast of the Salton Sea becoming 2.1 and 2.3, and decreases to the east to 1.6.Reference [21] concluded that high concentration of high Vp/Vs ratio might be caused by the plagioclase rich mafic composition of the lower crust; because high crustal S-wave and Rayleigh wave phase velocity in the same area exclude the possibility of crustal fluids or partial melting as a possible reason for high Vp/Vs.Reference [22] concluded that the low velocity with low Vp/Vs zones in the upper crust are caused by the inclusion of H 2 O and that low velocity with high Vp/Vs zones in the lower crust and uppermost mantle are caused by melt inclusions.Magnetotelluric survey in similar areas revealed conductive zones located at a depth of below 15 -20 km; this is consistent with the low velocity with high Vp/Vs zone [23].Reference [23] inferred that the low velocity anomaly could be explained by the partially molten rock.

Frequency-Magnitude Distribution (b-Values)
The b-value is a critical parameter for seismic hazard assessment and an indicator of the presence of fluids and/ or partial melt.The earthquake distributions (number of events vs magnitude) typically follow a power law; and the negative slope of that plot is the b-value [24].Normal Copyright © 2012 SciRes.IJG fault events have the highest (b) values (b ≈ 1.1), thrust events the lowest (b ≈ 0.7) and strike slip events intermediate values (b ≈ 0.9) [24].Southern California Earthquake Data Center (http://www.data.scec.org/index.html)seismic activity record shows numerous seismic events for September, 2009, that are focused along the San Jacinto, Elsinore, Cerro Prieto and Imperial faults, while seismic activity is weak or absent along the fossil spreading centers of the Sand Hill and Algodones faults (Figure 5).Reference [25] analyzed earthquakes from the local catalog of events in the Southern California region from 07/01/1944 to 03/01/1990 using magnitudes between M = 4.1 and M = 7.7.The cumulative and the binned density are shown in Figure 6.Slopes in Figure 6 are negative.This observation may imply the number of normal fault events falls off more rapidly with increasing magnitude than is the case for strike-slip or thrust events.b-values for the complete catalog are shown in Table 2.The cumulative distribution seems to have the property of self-similarity across the entire range of magnitude to about M = 6.6 (Figure 6; from [25]).Reference [26] shows the relation between the depth and the b-values in southern California (Table 3).
They separate the data into five depth ranges from 0 -3, 3 -6, 6 -9, 9 -12, and 12 -15 km for calculating b-values, they restricted the magnitude to larger than M = 2.0 for detection and completeness and smaller than M = 5.5.Table 3 shows a decrease in b-values with increasing earthquake depth, with exception for the intervals 9 -12 km and 12 -15 where the b-value is increasing by a very small portion.
Reference [27] showed that areas of high b-values from 1.3 ± 0.05 up to 1.5 ± 0.05 could be attributed to the    presence of high thermal gradients due to the emplacement of magmatic fluids.As magmatic fluids injected into the system and temperatures expected to rise around the intrusion and the crust would weaken due to its inability to accumulate high amounts of stress.The high temperature, weaken crust and expanding sill could cause the formation of numerous small fractures [27].
A relatively large number of smaller earthquakes is expected to accompany the formation of small fractures and would alter the frequency-magnitude distribution of earthquake toward higher b-values because of developing deviatoric tension [27].
Reference [28] plotted b-values in Koyna-Warna region, India, and concluded that a systematic increase in b-values from 8 km depth to 12 km for Koyna earth-quakes indicates the presence of pore fluids at a depth greater than 8 km, and stated that low b-value areas at 5 km depth show localized high stresses that are favorable for future rupture.magnetic anomaly map (Figure 8).We used Montage Oasis soft ware for contouring and Adobe illustrate for final production.Both maps were projected using UTM projection.

Model Development
We obtained gravity data from UTEP-PACES We created a crustal scale model (Figure 9) to explore the deep structures of the active and fossil spreading centers using the unfiltered Bouguer gravity and magnetic anomaly maps with the receiver function analysis.We divided our targeted features into "active" and "fossil" spreading centers based on the amount (magnitude and frequency) of seismic activity corresponding with known faults and the style of the potential field anomaly.
That is, we expect high amplitude Bouguer gravity anomalies to correspond to the rifts because of the thin crust and shallow Moho which is typical for rift settings.
(http://www.research.utep.edu/paces)website that is currently hosted at the Cyber-ShARE Center of Excellence at UTEP.We used 40,784 Bouguer gravity points to create the Bouguer gravity anomaly map (Figure 7).The gravity data were merged from a variety of surveys and cover the US and the border region.Average error for this data set ranges from 0.05 to 2 mGal (Al-Douri, personal communication, 2009).Reference [29] of the US Geological Survey using a digital elevation model calculated terrain corrections and a technique based on the approach of [30].A Bouguer gravity correction was made using 2670 kg/m 3 as the reduction density.Aeromagnetic data were obtained from the US Geological Survey with a grid spacing of 1 km [31].We used 112,436 aeromagnetic measurement points to create the

Crustal Model
The starting point for our model was to incorporate the depth to the Moho that was determined from the receiver  function analysis and the densities for the upper, middle, lower crust, and upper mantle from previous studies [7,12,32].Magnetic susceptibilities estimated from [33].
The density (D) for the upper crust (sediments and metasedimenatry composition) ranges from 2500 to 2600 kg/m 3 , while magnetic susceptibility (S) ranges from 0.002 to 0.007, density of the middle crust is 2750 kg/m 3 , and (S) is 0.065 and mostly quartzofeldspathic composition, density of the lower crust is 2950 kg/m 3 (gabbroic composition), and (S) is 0.004.Density of the magmatic body is 3100 kg/m 3 , its magnetic susceptibility is 0.0, and the mantle density is 3300 kg/m 3 with a magnetic susceptibility of 0.0.
The profile (A-A') is about 450 km long, and crosses the central part of the study area (see Figure 7 for profile location).The depth to the Moho, determined from the receiver function, varies from 38 km at the starting point (A), decreases to 26 km southwest of the Salton Sea, and increases to 32 km at the end point (A') of the model.Magmatic intrusion is modeled at a depth of about 20 km to the southwest of Salton Sea as shown by the crustal model (Figure 9).That intrusion extends for 70 km in SW-NE direction as interpreted from the model.Magmatic intrusion released fluids that promote seismic activity in the active spreading centers.

Discussion
We used several data sets to interpret seismic activity in the active spreading centers of Salton Trough.Table 4 summarizes the differences in observations and models between the active and fossil (inactive) spreading centers.We incorporate receiver function analysis with gravity and magnetic data to create a subsurface crustal scale model that reveals magmatic intrusion underlies the active spreading centers.Vp/Vs ratios are high in the area where magmatic intrusion presents and indicates partial melt in the lower crust; b-values are high in these same areas and indicate fluids in the upper crust.
We interpret upper crustal seismicity within the active spreading centers to promote by fluids ex-solved from  the magmatic intrusion that underlies the spreading centers at a depth of about 20 km because cyclic fluid pore pressure increase weakens faults [34].Earthquake activity in the lower crust has been identified at several rifts (e.g.Taupo Volcanic Zone, New Zealand, [5]) suggesting the possibility of fluid involvement.Both seismicity and tomography suggest that in order for these lower crustal events to occur, fluids needed, as well as a rheologically strong lower crust that restricts these fluids to its base.The strength of the crust is governed by strain rate, temperature and rock composition as well as pore fluid pressure [5].
The locations of the active spreading centers are characterized by high amplitude gravity anomalies because the crust is thin and the Moho is shallow beneath active spreading centers according to our receiver function analysis (Figures 3 and 4).Low gravity anomaly observed to the west and south of the Salton Sea, we relate this low anomaly to the existence of fluids or partial melt.We assume the crust is gabbroic in composition, consistent with our model density of 2950 kg/m 3 .This composition could provide the strong lower crust that restricts fluids to its base.The active spreading centers are characterized by a low magnetic anomaly inferred to be the response from molten material and/or fluids.In contrast, high amplitude magnetic anomalies prevail in the fossil spreading centers due to the magnetization of cooled gabbro and less fluid in the crust.
Several lines of evidence indicate the role of fluid involvement in seismic activity within the active spreading centers of the Salton Trough.High Vp/Vs to the south and southwest of the Salton Sea provide evidence for such activity (Figure 4).
The crustal scale model suggests a partially molten, mafic intrusion underlies the location of the active spreading centers (Figure 9).We interpret the fluids to be derived from that magma intrusion.Our model also shows a doming of the Moho beneath the spreading centers (between San Jacinto and San Andreas faults, Figure 9) which is suggested to results from magmatic activity and intrusion in the lower crust and upper mantle.Reference [35] concluded that the regional upper mantle is very heterogeneous with the greatest amplitude of velocity and Vp/Vs perturbations in the upper 200 km and mantle structure that correlate spatially with major physiographic provinces of southern California.
The largest low velocity, high Vp/Vs anomaly is found beneath the Salton Trough and is attributed to relatively high partial melt in the asthenosphere resulting in lithospheric thinning.Reference [4] concluded from the magnitudes of Vp, Vs, and Vp/Vs perturbations, and knowledge of regional 1-D velocity and attenuation, that the asthenosphere contains up to 1% partial melt extending to depths of 150 -200 km.
b-values from frequency-magnitude plots for southwest California are high.According to [25,26], areas of high b-values are attributed to the presence of a high thermal gradient due to the emplacement of magmatic fluids [27,28].Earthquakes occur on strike slip and/or normal faults in these areas.The lack of coincidence between the epicenteral area of seismic activity and the surface outlet of sub-crustal fluids suggest that the escaping fluids do not follow vertical pathway on moving from their deep source to the surface [36].Pressurized fluids can escape from their source along high permeability pathways in the upper crust [37].

Conclusion
Seismic activity is restricted to the active spreading centers of the Brawly, Cerro Prieto, Imperial, Elsinore, and San Jacinto fault zones, these are strike-slip faults that extend from 15 km (Brawley fault) to 210 km (San Jacinto fault zone) in the active spreading centers of the Salton Trough rift (Figure 2).Spreading is limited or absent in the fossil spreading centers at the Sand Hill and Algodones faults (Figure 2).We associate seismic activity with the presence of a magmatic intrusion beneath the active spreading centers and with fluids derived from this intrusion.Several lines of evidence support the existence of magma intrusion and fluids: 1) low velocity and high Vp/Vs ratios in the lower crust as shown by tomography and receiver function analysis; 2) low amplitude magnetic anomaly in the location of active spreading centers attributed to the fluids and/or melts; and 3) a high amplitude gravity anomaly that underlies the active spreading characteristics of the inferred magma intrusion can be recognized over a distance of 70 km in a NE-SW direction along the fault zones of the active spreading centers (Figure 9).Frequency-magnitude distribution b-values are high in southwest California including the study area; high b-values are attributed to high thermal gradient due to the emplacement of magmatic fluids.

Figure 1 .
Figure 1.Location map showing the study area and adjacent areas.Dashed line A-A' showing the location of our model.Square box showing the area of Figure 2.

Figure 2 .
Figure 2. Topographic and structural map of the Salton Trough showing the active spreading centers (gray polygon), and fossil spreading centers (After Lonsdale, 1989).

Figure 3 .
Figure 3. Contour map of the Moho depth (crustal thickness) in km based on receiver function data.Data were collected from 27 stations (shown as red triangles).Stations codes are shown inside small squares beside each station.

Figure 4 .
Figure 4. Vp/Vs contour map, high Vp/Vs located to the south and southwest of the Salton Sea.Gray circle shows an approximate location of the magmatic intrusion.
3 km/s and Vp/Vs ratio of 1.732 for 30 km thick crust, one gets: ∆H = (H/Vp) ∆Vp = 4.3 ∆Vp (km) which means that the uncertainty of H is <0.5 km for a 0.1 km/s in Vp.However, the thickness is highly dependent on the Vp/Vs as shown by: ∆H = (H/Vp) ∆k = −40.2∆k (km) i.e., a 0.1 change in k can lead to about 4 km change in the crustal thickness.This ambiguity reduced by using later phase, which provide additional constraints:

Figure 5 .
Figure 5.The seismic map of September 2009.Seismic activity focused along the active spreading centers and absent along the fossil spreading centers (Map from Southern California Earthquake Data Center).

Figure 7 .
Figure 7. Bouguer gravity anomaly map of the Salton Trough area.Fossil spreading centers and active spreading centers are gravity high related to basaltic volcanism of Salton Buttes, and due to the presence of shallow Moho and magmatic intrusion.Dashed line A-A' shows the location of the crustal profile.Circle labeled as 1 shows the approximate location of the active spreading centers; circle 2 shows the approximate location of the fossil spreading centers.SD: San Diego; T: Tijuana; SF: San Felipe.

Figure 8 .
Figure 8. Magnetic anomaly map of the Salton Trough.Fossil spreading centers are magnetic medium to high related to basaltic rocks; active spreading centers are magnetic low associated with the fluids in the upper crust.Dashed line A-A' shows the location of the crustal profile.Circle labeled as 1 shows the approximate location of the active spreading centers; circle 2 shows the approximate location of the fossil spreading centers.SD: San Diego; T: Tijuana; SF: San Felipe.

Figure 9 .
Figure 9. Interpretative model for gravity and aeromagnetic data profile A-A' (Central region of Salton Trough).This model is about ~450 km long, covers the central part of the study area.The depth to the Moho, according to the receiver functions, varies from 25 km at the central point of the model south of Salton Sea [D = Density (kg/m 3 ), S = Susceptibility (dimensionless), M = Magnetization (A/m), MI = Magnetic inclination (degree), MD = Magnetic declination (degree)].