P Wave Reflection and Refraction and SH Wave Refraction Data Processing in the Mooring, TN Area Abstract: Author: Duayne Rieger Home Institution: Slippery Rock University of Pennsylvania REU Institution: The University of Memphis REU Advisor: Dr. Charles A. Langston In the Mooring area, in NW Tennessee (part of the New Madrid seismic zone), the subsurface consists of unconsolidated sediments down to the Paleozoic limestone basement. On October 28, 2002, as part of The Embayment Seismic Excitation Experiment, a 500lb explosion was detonated in the Mooring, TN area to generate surface waves in effort to investigate sediment anelasticity. P wave reflection and refraction data along with s wave refraction data was collected in between the explosion site and the strong motion array of the 2002 Embayment Seismic Excitation Experiment. Through the processing of this data, using the ProMAX package in the LandMark system, the velocity structure of this area can be further studied. Introduction to the Study: Earthquakes are very devastating to society and can cause the loss of life along with massive amounts of structural damage. At the present time there exists no method of accurately predicting when and where an earthquake will occur. With this lack of warning, preventative actions are the best options for earthquake mitigation. Earthquake mitigation relies heavily on understanding how the energy released by earthquakes responds to different types of Earth materials. The New Madrid Seismic Zone is an active seismic zone in the Mississippi embayment, which is located in the central part of the United States. The Mississippi embayment straddles the Mississippi river from Southern Kentucky and Missouri down into the Gulf of Mexico. The embayment contains upper Cretaceous to the present unconsolidated sediments that range up to a kilometer in depth, with the depth generally decreasing with distance away from the river. A number of studies have presented evidence that when seismic waves propagate up through these unconsolidated sediments they tend to experience amplification (Langston, 2005). In light of this, understanding how seismic waves propagate through these unconsolidated sediments is of great importance. On October 28, 2002, as part of The Embayment Seismic Excitation Experiment, a 500lb explosion was detonated in the Mooring, TN area to generate surface waves in effort to investigate sediment anelasticity in the embayment. To aid in this study, P wave reflection and refraction data along with SH wave refraction data was collected in between the explosion site and the strong motion array of the 2002 Embayment Seismic Excitation Experiment. Through paper 03 1 Seattle, Washington August 8-12, 2007
the processing of this data, using the ProMAX package in the LandMark system, the velocity structure of this area can be further studied. Data Collection: The seismic data was collected by the United States Geological Survey on November 16 th, 2006. The P wave data was collected using a vibroseis 12s, 15-120Hz, linear sweep as a seismic source (shot). The seismic array used was of straight line geometry and consisted of one 8Hz vertical geophone located at each of 144 recording stations (located 5m apart), ranging from station 101-244 (station 101 being the most Eastern station), which recorded, using a Geometrics 24bit Geode Seismograph, during every shot for 14s. The Easting, Northing, and elevation of every fifth station were recorded using GPS. The locations of the stations in between every fifth station were generated using a linear interpolate in between each GPS reading. Two initial shots were located at stations 101 and 103 respectively, after station 103 two shots were located at every other station for the remaining length of the array. Outside the seismic array, eight shots were performed 550m west of station 144, four shots were performed 0.4 miles East of station 101, and eight shots were performed 1.4 miles east of station 101. Figure 1 Seismic array for p-wave data collection The SH wave data was collected using only a slightly different seismic array than the P wave data collection. The array used the same recording stations as the P wave seismic array, but only recorded at stations 101-172. The seismic source for the SH wave data was created by driving the vibroseis truck up onto a wooden beam and striking the beam end on with a hammer. The seismic sources were only located at stations 101, between stations 136and 137, and station 172. paper 03 2 Seattle, Washington August 8-12, 2007
Figure 2 Seismic array for s-wave data collection P Wave Reflection Data Processing: Each trace of the data contains signal and noise. The main objective of reflection data processing is to increase the signal to noise ration of the data. Fig. 3 is raw data of a shot gather that was recorded when the shot was located at station 173. The signal to noise ratio is very low and requires processing. There is not set method for increasing the signal to noise ration, but there are many useful tools and techniques to be used when necessary to achieve a desirable signal to noise ratio. Figure 3 Raw p-wave data with vibroseis located at station 173 paper 03 3 Seattle, Washington August 8-12, 2007
Reflections are often small in amplitude due to geometric spreading and/or anelasticity of the medium. In order to increase the reflection amplitudes, automatic gain control (AGC) is applied. The sweep trace from each shot is correlated with the recorded data from that shot. Also, traces that contain a large amount of noise are killed from each shot gather. In signal processing Fourier analysis is used, where a non-monochromatic signal is decomposed into a sum of monochromatic signals. Using Fourier analysis makes it possible to omit unwanted frequencies of the signal, which is commonly known as filtering. A band pass filter was applied to the data in attempt to retain frequencies that make up the signal and omit frequencies that do not contribute to the signal. Refraction Acoustic Wave Ground Roll Reflection Figure 4 P wave data with AGC, sweep correlation, and a (20-30-90-120)Hz band pass filter applied The data in Fig. 4 has a higher signal to noise ratio than the data in Fig. 3. Some reflections are now identifiable as the waves that have hyperbolic moveout. Other identifiable features of the data are the refractions and acoustic waves, which each have linear moveout, and the ground roll. These are all considered noise and are muted out, leaving only the reflections, which, in doing so, will also increase the signal to noise ratio. paper 03 4 Seattle, Washington August 8-12, 2007
Next a Common Depth Point (CDP) sort is done, where the data is sorted in a way that traces that record the Earth response from the same depth are grouped together (Fig. 5). The remaining processing is done to these CDP sorts. The Earth response is best seen from CDPs from locations in the array where the fold is the largest, where with linear geometry, this is around the middle of the array. The hyperbolic reflections are more easily seen in the CDP, the red hyperbola on Fig. 5 traces out one of them. Figure 5 CDP sort located at station 178 After the CDP sorts have been created, velocity analysis is performed in order to determine the stacking velocity that generates the most semblance at different times. In detail, velocity analysis applies Normal Moveout Correction (NMO) at each time using a whole range of velocities and determines the semblance of each possible stacking velocity. Semblance is the amount of constructive interference present when the traces are stacked. The stacking velocities that create the most semblance at each time are the stacking velocities that will be used with the NMO on the CDP sorts. Fig. 6 show the velocity analysis of a particular CDP sort, high semblance is indicated by red. paper 03 5 Seattle, Washington August 8-12, 2007
Figure 6 Velocity analysis of a CDP sort Normal Moveout Correction is now applied to the CDP sorts, using the stacking velocities determined by velocity analysis. NMO (Eq. 1) is a mathematical correction to account for the hyperbolic moveout with time of reflections. NMO with the correct stacking velocity v ) for a particular reflector will provide a first arrival time correction (t) that will pull the ( stack first arrival times of the reflector at all offsets (x) to the two-way travel time ( t 0 ). The two-way travel time is the time it takes for a P wave to arrive back to a receiver when it has an angle of incidence of ninety degrees with the reflector, or the arrival time at the geophone with zero offset from the source. 2 2 t = t0 + x 2 2 v stack (Eq. 1) Once NMO has been applied to each reflector, the CDPs are then stacked together to form a CDP stack. This CDP stack can be looked at as a cross-sectional view of the Earth where reflecting layers at various times are visible. In Fig. 7, which is the CDP stack, there are many visible reflectors. Many reflectors are visible from 200-450ms which are thought to be reflectors in the upper Cretaceous to present unconsolidated sediments, along with a noticeable reflector around 650ms, which is thought to be the Paleozoic basement. This processing gives an estimation of the interval velocities of the subsurface from the stacking velocities used to paper 03 6 Seattle, Washington August 8-12, 2007
perform the NMO. Much further analysis can be done to infer more information from this processed data about the velocity structure of the shallow subsurface in the Mooring, TN area. Refraction Data Processing: Figure 7 CDP stack with NMO applied Some basic processing needed to be done to the P wave and SH wave data so the first arrivals of the refractors are more easily detected. The P wave data was correlated with the sweep trace and had the same band pass filter applied as in the reflection processing. The SH wave data only had the same band pass filter applied to it as the P wave data did. In looking at long offset refractions in the P wave data refraction and refractions in the SH wave data, apparent velocity can be determined. The apparent velocity is the velocity of a wave along the surface of the Earth. This is found by determining the time that first motion occurred at each geophone, which has a particular offset from the source, as illustrated in Fig. 8. With the offset vs. first arrival time data (x vs. t), a seismic inversion technique call the Weichert-Herglotz method can be used to generate a continuous depth vs. wave velocity model. paper 03 7 Seattle, Washington August 8-12, 2007
Figure 8 First motion time picks for SH wave data Conclusion: Through the processing of the P wave reflection and refraction data, and the SH wave refraction data, a detailed velocity model of the Mooring, TN area subsurface can be constructed. Using the velocity models from this processed data with help interpret the results of the Embayment Seismic Excitation Experiment and will aid in the investigation of sediment anelasticity. With a further understanding of sediment anelasticity in the embayment, more informed measures can be taken to increase earthquake mitigation in the embayment. Acknowledgements: Professor Charles A. Langston The National Science Foundation The Mid America Earthquake Center United States Geological Survey CERI at The University of Memphis paper 03 8 Seattle, Washington August 8-12, 2007
References Langston, C. A. (2006). Bulk sediment Q p and Q s in the Mississippi embayment, central United States, Bull. Seism. Soc. Am. 95, 2162-2179. paper 03 9 Seattle, Washington August 8-12, 2007