Sismik Yorumlama Applying the Seismic Refraction Techinique to Exploration for Transportatiom Facilities APPLYING THE SEISMIC REFRACTION TECHNIQUE TO EXPLORATION FOR TRANSPORTATION FACILITIES Michael L. Rucker AMEC Earth & Environmental, Inc. Phoenix, Arizona 85009; ABSTRACT The seismic refraction technique provides a simplified characterization of relatively large volumes of the subsurface in 2-dimensional (distance and depth) profiles. Compression wave (p-wave) velocities, the typical measured geologic material parameter, are a function of the moduli of the various unsaturated material masses in the subsurface profile. In saturated media, p-wave velocities are increased relative to equivalent unsaturated media. Within constraints of basic laws of physics, seismic refraction profiles are a representation of subsurface profiles as long as the p-wave velocities (material strengths) increase with depth. Soil/rock contacts or contrasts between weaker to stronger geologic material horizons can be interpreted from seismic refraction data. Preliminary subsurface profiles can be developed from this information, and characterization of subsurface profiles between geotechnical borings or test pits can be accomplished. Correlations between p-wave velocities and prediction of rock mass rippability or excavatability have been published, and correlations with other geotechnical parameters for local conditions can be compiled. ASTM D5777 provides suggested practices for seismic refraction investigations. INTRODUCTION Seismic refraction methods provide an effective and efficient means to obtain general information about large volumes of the subsurface in the two dimensions of depth and horizontal (or slope) distance. Information provided by seismic refraction includes compression wave (p-wave) velocities within the investigated subsurface profile. Traditionally, these velocities are interpreted to be present within layers or horizons whose depths are also interpreted. Newer interpretation methods are making it possible to interpret velocity changes as gradients as well as discrete layers. Limitations due to subsurface geometries such as thin layers and lower velocity horizons underlying higher velocity horizons, must be understood and, if necessary, accounted for. Cross- correlation with other exploration methods such as drilling, test pits and geologic mapping, can greatly increase the value of refraction seismic data. In return, refraction seismic data can significantly enhance the value of other exploration data. Both basic field operations and basic interpretations of the resulting data can be performed by properly trained and experienced geotechnical or geological engineering personnel as well as by geophysical specialists. This paper is intended to review seismic refraction practice for geotechnical engineering work as related to transportation facilities such as highways. Basic equipment and methods typically used by geotechnical engineers, and interpretation by simple, classic methods and an example interpretation by automatic optimization software will be reviewed. Equipment and methods deployed by geophysical specialty groups may be considerably more complex than those described in this paper. For more general application of seismic refraction methods, Redpath (1973) and Mooney (1984) provide classic introductions to the seismic refraction technique. The U.S. Army Corps of Engineers Engineering Manual EM 1110-1-1802 adapted by ASCE (1998) provides a more current review of the method. ASTM D5777-95, "Standard Guide for Using the Seismic Refraction Method for Subsurface Investigation" (ASTM, 2000) outlines procedures and quality control aspects of performance and analysis. Typical Applications Seismic refraction work for transportation facilities is typically performed to support geotechnical site characterization. ASTM D6429-99, "Standard Guide for Selecting Surface Geophysical Methods" (ASTM, 2000) lists seismic refraction as a primary method of choice for investigating soil/unconsolidated layers, depth to bedrock or water table, and soil and rock properties. ASTM D6429-99 lists the method as a secondary choiceor alternate method for rock layers, fractures and fault zones, voids and sinkholes, landfill and trench boundaries and archeological features. Assessment of rippability/excavatability (Caterpillar; 1984, 1993) is an important application in highway engineering where large rock cut excavations are a significant construction cost. Other applications include landslide characterization (TRB, 1996), assisting in earthwork factor determination (CALTRANS, 1978; Rucker, 2000) and earth fissure locating (Rucker and Keaton, 1998). Equipment portability can be a profound advantage of the seismic refraction method for obtaining subsurface information. Systems capable of investigating to depths of up to about 75 to 100 feet are typically small enough to be carried by a two-person crew to specific locations on a project. In the firm where the author practices, seismic refraction is considered by other geotechnical engineers and geologists to be among the 'light cavalry' of subsurface exploration methods. Proposed rock cuts through mountainous terrain, potential bridge pier locations in canyons, and sensitive areas such as archeological sites (Figure 1) or biological habitat closed to vehicular-based exploration methods are routinely investigated using seismic refraction. In addition, both packhorse and helicopter mobilization have been used in remote route studies for highways and pipelines. Figure 1. Subsurface profile at unmitigated archeological site based on seismic interpretation from four 120- foot seismic setups (lines) and surficial geologic mapping as part of the 30% design submittal for rural highway section. The area will be investigated by drilling after archeological clearance and mitigation is completed before final design. As can be seen from Figure 1, large areas of subsurface material can be characterized as a simple set of layers or horizons with different compression wave (p-wave) velocities (units of feet per second or meters per second) that increase with depth. These parameters can effectively complement other subsurface investigation methods such as exploratory drilling and test pits that provide precise subsurface information at a single location. A seismic refraction line encompassing that location can provide simplified two-dimensional trend data to extrapolate knowledge laterally from the precise subsurface exploration information. Correlations with Other Geotechnical Parameters P-wave velocities are an important parameter to assist in site characterization, especially when constrained by other geologic information. P-wave velocities can range from a few hundred feet per second in very loose soil deposits, to about 1,500 to 2,500 f/s in engineered fills, to greater than 15,000 f/s in intact, competent rock. For engineering rock characterization purposes, p-wave velocities indicate overlying soils and fracturing and 5670 5680 5690 5700 5710 5720 5730 5740 5750 707 709 711 713 715 717 Project Stationing (English) Elevation, feet p-wave velocities are in ft/sec seismic lines are 120 feet long with about a 30-ft depth of interpretation 1400 3000 1600 3400 1100 5000 ~3200 5600 ~4000 1200 highly weathered/ decomposed granite ground surface residual soil older alluviumweathering in a rock mass rather than the rock type. A classic use of p-wave velocity is for rippability studies. Figure 2 summarizes anticipated rippable, marginally rippable and non-rippable conditions as a function of p- wave velocities in granites and conglomerates. Caterpillar (1984, 1993) presents these parameters for a range of geologic materials; limitations of their use are also discussed. Correlations between p-wave velocities and other common geotechnical parameters indicate the wide range of materials that can be characterized using seismic refraction methods. Figure 3 presents correlations of p-wave velocity and average of standard penetration test (SPT) blow counts within a subsurface horizon at over 20 projects in the southwest US. Figure 4 presents correlations of p-wave velocity and average of rock quality designation (RQD) within a subsurface horizon in granitic materials for three rural highway projects and one hotel complex in Arizona. Data scatter in both Figures 3 and 4 are indicative of the nature of SPT and RQD measurements as well as limitations of the geophysical measurements and analyses. Unsaturated, cohesionless, well graded sand, gravel and cobble streambed deposits in the Southwest US (Rucker, 1996) show a correlation between mean particle size D 50 and p-wave velocity as shown in Figure 5. The effect of saturation, which increases p-wave velocity, is also shown in Figure 5. It should be noted that the speed of sound in air is about 1,150 f/s, and the speed of sound in water is about 5,000 f/s. Figure 2. Rippability as a function of p-wave velocity for various sizes of bulldozers (D7-D11) and trackhoes (235-245) in granites and conglomerates (Caterpillar, 1984, 1993). Figure 3. Correlation between p-wave velocity and average of standard penetration test (SPT) blow counts for about 18 sites in southwest US. Refusal blow counts as recorded on boring logs, greater than 100 per foot, are extrapolated to blow counts per foot for comparison purposes only. Above the water table, p-wave velocity is a measure of rock or soil mass low-strain or dynamic Young's modulus. Young's and shear modulus may be estimated by assuming a Poisson's ratio and measuring or estimating the soil or rock mass unit weight. Formulas for this calculation are presented in ASCE (1998) and ASTM D2845-95 (ASTM, 2000). At Poisson's ratio of 0.22, the shear wave velocity is 60 percent of the p-wave velocity, and at Poisson's ratio of 0.33, the shear wave velocity is 50 percent of the p-wave velocity. Figure 6 presents correlations of calculated low-strain modulus values for sites investigated by the author where both p- wave and s-wave velocities were measured. These relationships are not valid below the water table, where saturation profoundly effects p-wave velocity. Relationships between modulus as estimated by refraction seismic p-wave velocity and density have been further investigated and developed into relationships to estimate earthwork factors for large cuts in weathered rock and associated embankments for highway design (Rucker, 2000). This work is an extension of empirical earthwork relationships using p-wave velocity developed by Caltrans (1978). Figure 7 presents p-wave velocity and unit weight (density) relationships for weathered granite rock cuts and embankments on two highway projects in Arizona. The relationship is calibrated using estimated low strain modulus values based on 0 2000 4000 6000 8000 10000 12000 14000 0 200 400 600 800 1000 Equipment horsepower P-wave velocity, ft/sec granites conglomerates small backhoe D7G 235 D8L, D9N 245 D9L, D10N D10, D11N RIPPABLE UNRIPPABLE Marginal Rippability: 0 1000 2000 3000 4000 5000 6000 7000 1 10 100 1000 SPT blows per foot P-Wave Velocity, ft/sec SPT Refusalunconfined compressive strengths (Bieniawski, 1989; van Heerden, 1987) of tested core samples for these projects. Figure 4. Correlation between p-wave velocity and rock quality designation (RQD) values for three highway projects and one resort project in granites in Arizona. Average values and ranges are shown. Figure 5. Correlation between p-wave velocity at depth of about 10 feet and mean particle size D50 for well graded cohesionless streambed deposits at 20 sites in southwest US. Figure 6. Correlation between p-wave velocity and low-strain Young?s and shear moduli based on measured p-wave and shear wave velocities at 25 sites in southwest US. Figure 7. Estimation of earthwork factors based on weathered granite rock mass and existing embankment unit weights from refraction p-wave measurements (Rucker, 2000). Low-strain modulus values estimated from unconfined compressive strengths of tested core samples are used to verify unit weight-modulus trend. BACKGROUND THEORY & CONCEPTS The theory for refraction seismic methods is outlined or described in many references (ASTM, 2000; ASCE, 1998; Richart and others, 1970; Mooney, 1984; Redpath, 1973). A theoretical description is most commonly treated as an optical refraction problem using Snell?s Law to quantify wave propagation geometries. P-wave velocity is a function of material modulus, as presented in numerous references, including ASTM 2845, as shown in Figure 6. Figure 8 presents a schematic of the refraction seismic method and equipment over a dipping geologic interface. The true velocity of the lower layer is calculated as the harmonic mean of the forward 3000 5000 7000 9000 11000 13000 15000 0 20 40 60 80 100 RQD P-wave Velocity, f/s 1000 10000 0.1 1 10 100 Mean Particle Size D50, mm P-wave Velocity, ft/sec saturated unsaturated 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 1 10 100 1000 Modulus E & G, ksi P-wave Velocity, f/s Young's modulus E Shear modulus G 0.1 1 10 100 110 120 130 140 150 160 170 180 190 Unit Weight, pcf Dynamic Modulus E, GPa 1 seismic velocity, f/s 15,000 10,000 7000 5000 3000 2000 1500 embankment p-wave vel rock mass p-wave vel UCS-modulus dataand reverse profiles. Proper determination of p-wave velocities requires measurements of arrival times in forward and reverse directions to account for dipping subsurface interfaces. Figure 8. Conceptual sketch of refraction seismic method over a dipping interface. Seismograph system for engineering application is shown. Depth interpretations occur at changes in the time-distance trend slopes. P-wave ray directions are shown for forward and reverse profiles. EQUIPMENT Equipment to perform a seismic refraction survey includes a seismic energy source with an electrical timing signal, geophone(s) to detect signals from the energy source, a seismograph to measure, process and record energy source timing signals and received geophone signals, and cabling to bring source and receiver signals to the seismograph. These essential equipment elements are reviewed below and detailed in ASTM D5777. Energy Sources For geotechnical work in the American southwest, the author has found that a 10- to 12-pound sledgehammer hitting a roughly equivalent-mass metal plate is an appropriate energy source for most applications using a 120-foot long geophone array. It is also acceptable for many 300-foot long geophone arrays where ambient noise is relatively low or signal attenuation by the subsurface media is not excessive. With a sufficiently capable hammer-person, a 20-pound hammer can provide a larger signal for marginal conditions. An electrical timing device is taped to the hammer handle immediately below the hammer head to generate the initial time pulse. A cable connects the timing device to the seismograph. Seismographs with signal enhancement capability, where signals from several hammer strikes are stacked (added together) permit the effective use of these relatively low-energy sources. A sledge hammer energy source is fully portable for difficult or remote access work. It is usable in rural and urban settings where the seismic signal can be effectively determined over ambient background noises such as traffic or aircraft noise. Hammer energy sources have been used for seismic refraction surveys for urban freeway design in suburban settings using alleys as the -30 -25 -20 -15 -10 -5 0 5 10 -20 0 20 40 60 80 100 120 140 seismograph hammer & cable geophones & cable shotpoints soil; velocity 1 rock; velocity 2 velocity 1 velocity 1 velocity V2A velocity V2B Time-Distance Plot velocity 2 = 2xV2AxV2B/(V2A+V2B) Depth Time Distance velocity 1 & 2 interfaceinvestigation access. Timing cables are subject to being hit and damaged on occasion by the sledgehammer; with simple tools and electrical tape, field repairs can be made. Explosive energy sources may be appropriate where deeper investigations are performed or greater energy is needed and permitted. Use of explosives in urban settings or near residences or businesses is typically impractical or impossible due to severe regulatory and liability constraints. Cartridge-type energy source guns are available to provide a flexible and safe type of small explosive source. Explosive charge energy sources require appropriate personnel and safety procedures. Timing devices and special blasting initiators are available to simultaneously trigger an explosive source and the seismograph. When electric-detonating blasting capabilities are unavailable, some seismographs can be triggered by simply opening the timing circuit. Triggering is accomplished by wrapping wire completing the timing circuit around the explosive charge, which can then be initiated by a fuse or other non-electric detonator. Other energy sources including heavy drop weights, pneumatic systems, etc. are available. Typically, such systems are vehicle or trailer-mounted for portability, and are thus limited to use in situations where energy source locations are accessible by vehicle. Geophones and Cabling Deploying cables and geophones is a significant portion of the time and labor required to complete a seismic refraction survey. A typical geophone cable has takeouts (geophone connection points) at intervals along the cable and connectors for the seismograph at each end. Geophones normally include a spike that is shoved into the ground for effective coupling to detect the seismic signal, and a pair of electrical clips for electrical connection to the geophone cable. Appropriate geophone spacings are determined by the objectives of the survey. Onsite, a measuring tape may be laid out along the survey axis and geophones are deployed at the desired spacing. The geophone cable is then deployed with a takeout placed at each geophone location. The geophones are clipped to the cable prior to data acquisition. A properly sized geophone cable with takeouts at the geophone spacing intervals and marks on the cable for energy source locations can eliminate the need for tape layout and reduce array deployment time. Cables are relatively delicate and need to be handled with care for maximum useful life. Seismographs Early signal enhancement seismographs (circa 1970's) intended for civil engineering use were frequently single channel units with only one geophone. A survey was conducted by moving the hammer shotpoint to successive positions along the array and repeatedly hammering. The signal was displayed on a CRT screen, a knob-controlled on-screen pointer was used to identify the first arrival, and the travel time was displayed on the screen. Travel time was written down manually and the signal trace then erased for each shot point. No hard copy traces were generated. The 12-channel seismograph is an effective general-purpose tool for geotechnical investigations. The geophone array can be deployed in a short time and the resulting data can effectively resolve up to 3 subsurface horizons. Signals for all 12 channels are displayed on CRT (older) or LCD (newer) displays. Signal acquisition gains and display amplitudes are varied as needed to assist in determination of first arrivals. Filters to reduce noise are typically available, but, to minimize effects on seismic signals, should be used only when needed. Paper copies of traces are printed out in the field for permanent records. Recent instruments operating under microcomputer control can also record data onto disk. Vehicle or gel-type batteries are typically used to provide power; one or two 17-amp hour batteries can typically power the instrument for a day's work in remote areas. Seismographs with considerably more channels and capabilities are commercially available. However, the additional equipment and labor costs, including additional cables and geophones, and specialist skills needed for larger and deeper investigations, make such systems more useful tools for geophysical specialists rather than geotechnical/geological engineers and geologists.SURVEY DESIGN & DATA ACQUISITION Design of a seismic refraction survey for a project will be influenced by many factors, including the project requirements or information needs, personnel and equipment resources available, and budget constraints. This discussion is relevant for straightforward applications of the method by experienced personnel in a geotechnical group or consulting organization. Equipment can be either rented or owned by such groups. Safety Safety is the most important issue concerning survey design. Data cannot be collected where crews cannot safely work. Safety must be addressed in the office phase of survey design, and safety must be central to field modifications of the survey design. Site access is frequently a critical safety issue, especially in highway investigations. Access to highway project work areas may involve a field vehicle turning off of and onto roadways carrying traffic at highway speeds. Weather related safety issues include heat exhaustion or heatstroke as well as lightning storms. Considerable lifting and carrying, in addition to perhaps 15 to 30 or more strokes per line if a sledgehammer is used, may be required to complete work in some areas. Survey planning and design must consider availability of field crews physically as well as technically capable to safely complete the work. Many personnel cannot be expected to perform intensive, physically hard field operations, where the field crew is most or all of the days actively setting up and completing lines, or considerable backpacking of equipment, for more than two or three days without one to several days rest. Survey Design Within the constraints of safe operations, the number of lines (array setups) needed to complete a survey becomes a central survey design issue. Adequate coverage will depend upon the geotechnical issues being addressed. Fewer lines may be needed to determine appropriate excavation methodologies than would be needed to determine volumes of each type of excavation. Whenever practical, it may be effective to pair at least some seismic lines with geotechnical borings or test pits. The boring or test pit with sampling provides identification and vertical characterization of the subsurface horizons at a point, while the seismic line provides a second dimension (horizontal) to the subsurface horizon interfaces while providing further information on excavatability. Such ground truthing of the seismic data also improves interpretation of seismic lines performed in adjacent areas which are inaccessible or cost prohibitive to deploy drill rigs or backhoes. Seismic lines completed along a profile with occasional borings serves to fill in the geologic or geotechnical profiles between the borings. Cost and available resources are ultimate, realistic constraints on survey size. Details in a survey design which might be addressed in the office or field can simplify the interpretation and analysis process following completion of field work. Orient the seismic line geophone array on a uniform grade, horizontal or slope, whenever practical. This simplifies or eliminates the need for terrain corrections or variations in topography. If a grade break or change in slope cannot be avoided, set the midshot at the slope change so that the line can be interpreted as two half lines without terrain corrections. Plan instrument locations (foreshots) to be consistent to minimize the need to 'flip' or reverse the data order to maintain orientation during interpretation, analysis and presentation. Anticipated Survey Resolution & Depths of Investigation Up to three or four layers or horizons can normally be resolved using the seismic refraction method in geotechnical applications. Interpretations based on a typical 12-geophone array may have three horizons as an effective maximum. In mixed soil and rock geologic settings, these horizons may be soil, highly weathered and fractured rock, and less weathered, less fractured rock. An acknowledged general achievable accuracy for velocity depth interpretations is about 10 percent as reported in ASTM D5777. Hidden layers or velocity reversals in the subsurface profile can significantly reduce depth interpretation accuracies. Subsurface boring data may identify the presence of subsurface profile conditions that can degrade interpretation accuracy.Due to the physics of refraction propagation, the anticipated maximum depth of investigation is about one quarter to perhaps one third of the farthest energy source to geophone spacing. For 120-foot and 300-foot geophone arrays, this depth of investigation is about 30 to 40 feet and 75 to 100 feet, respectively. However, the actual depth of investigation depends upon the subsurface profile, and absence or presence of velocity reversals. Furthermore. the maximum depth is attained only around the center portion of the array; the actual zone of investigation may be roughly half ellipse shaped from the farthest shotpoints. Conservatively, the maximum depth of investigation for a particular seismic line can be assumed to be the depth of the deepest interpreted horizon. Figure 9. Investigation for a 4-foot deep waterline trench at a roadway using a 12-channel seismograph, sledgehammer energy source and simple 2-layer interpretation. Note lateral variations in interpreted p-wave velocities in a fractured rock horizon using simple intercept time method interpretation. The rock underlies an approximately 2-foot thick surficial soil horizon. The project only needed characterization to a depth of 4 feet; interpretation for deeper information was not needed. A Vermeer T-655 rock trencher was able to complete the excavation. Distances between the energy source shotpoint and near geophones determines a minimum depth of investigation. This minimum may be estimated as about one quarter to one third of the shotpoint to near geophone spacing. Long geophone arrays with large spacings between geophones and shotpoints will not resolve relatively thin, low velocity surficial soils horizons. A 5-foot and 10-foot shotpoint to geophone spacing might be able to resolve a 2-foot and 4-foot thick or less surface layer, respectively. Even with relatively close geophone spacings, such thin surficial horizons may only be interpretable as a trend between the shotpoint and shot data -5 0 5 10 15 20 25 0 20 40 60 80 100 120 Distance, ft Time, millisec. Depth, ft ground 1000 1300 1300 1 1100 1300 p-wave velocities are in ft/sec ~4500 ~7900 ~8700 ~9100 geophone locations source locationsa single nearest geophone; a trend through the source and the two nearest geophones would be needed to assure that an interpretation is not subject to a 'hidden layer' scenario. Minor errors in the timing trigger signal could also influence interpreted velocities. A sledge hammer trigger device must be attached to the handle very close to the hammer head to minimize this error. Horizontal variations in p-wave velocity within the two shallowest layers can be resolved within a 12- geophone array if several shotpoints are used. Interpretation can be performed for a two-layer case using the interval time method for data between adjacent shotpoints. For a seismic line with five shotpoints, up to four separate velocities for the intermediate horizon may be practical. Such interpretations may be of especial use in shallow rippability or excavatability investigations (Figure 9). An interpretation which presents an average velocity for a layer may under-report excavation difficulties in part of the layer. Some interpretation methods or analyses may inadvertently imply a greater resolution or accuracy than is warranted. For example, a method may determine a single representative velocity for each layer or horizon. Interface depths are calculated based on that velocity, and then iteratively readjusted to minimize errors with the observed data. The resulting interface depth plots might present geometric detail which is ultimately based on the model assumption of uniform velocity in each layer. The actual subsurface geologic profile could include zones where velocities are uniform, where velocities vary due to the condition of the subsurface materials, and where velocities vary with the overburden or confining stresses. Calculations of interpretation accuracy quantify the data goodness of fit in the context of the modeling assumptions rather than on actual subsurface conditions. Field Execution - Example Procedure for a Simple Setup with 2-Person Crew Successful field execution of seismic refraction work begins with location control appropriate for the project. Once location control has been established, execution of a typical 12-geophone array refraction line involves layout of the geophone array, preparation of the seismograph for data acquisition, deployment and application of the energy source at several locations, verification and recording of the seismic data, and picking up the equipment for the next line. For geotechnical work with a two-person crew using a sledge hammer energy source, the specific steps might be as follows. The equipment is mobilized to the foreshot position. Assuming the geophone cable is correctly sized for the line (no measuring tape is for geophone positions), person one holds the end at the foreshot position while person two walks the array length and lays out the geophone cable. Person one picks up the geophones and walks the length of the array, dropping a geophone at each cable takeout point. Person two walks back halfway and then places geophones 6 to 1 in the ground and clips them to the cable. Person one drops the farthest geophone (geophone 12) and then places geophones 12 to 7 in the ground, clips them to the cable, and returns to the seismograph. Person one opens the seismograph, attaches the geophone cable and battery cable. He ties the trigger cable to the seismograph handle, leaving slack between the connector and the handle. If the trigger cable is pulled, the force will be absorbed at the handle rather than at the connector. Person two picks up the hammer, the target plate and the trigger cable, and walks back to the far end of the array. Person one hooks up the battery, turns on the seismograph and prepares for data acquisition. Person two places the target plate at the backshot position and connects the hammer trigger to the trigger cable. Person one sets the seismograph to acquire data. Person two completes 3 to 6 hammer blows, and moves to the next source position. Person one directs person two when to hammer to acquire data when ambient noise is at a minimum. Person one checks the data on the screen, sets the gains, prints and saves the data, marks the first arrivals on the paper hard copy, clears the instrument and sets the instrument to acquire data. Person two performs hammer blows and source setup along the line at several intervals, perhaps the ends, middle and quarter points, while person one performs instrument operation and data verification and quality control. Once data acquisition is complete, the equipment is disconnected and picked up and cables are reeled up in approximately reverse order to deployment. The geophone cable is always the last item picked up; other items will not be far from the cable. This assures that no equipment becomes lost somewhere along the array. Other procedures will be appropriate with different equipment, energy sources, field conditions or project requirements. Possible InterferencesNoise is the most common source of interference when conducting a seismic refraction survey. Obtaining a clear first arrival requires that the seismic first arrival energy be considerably greater than ambient noise or ground vibrations (good signal to noise ratio). Signal enhancement, where signals from several hammer hits are summed, improves the signal to noise ratio. The sum of the desired seismic signals increases more rapidly than the sum of the (hopefully) random noise. Vehicles traveling nearby, aircraft, intermittent stationary equipment, pedestrians in the immediate vicinity, etc. are sources of interference which can be avoided or minimized by waiting until they have passed or shut down. If practical, use of shorter arrays reduces the source-geophone distance and thus signal attenuation. Very low, very high or very specific frequency noise might be reduced using filters built into the seismograph. A 60-Hz notch filter may be very useful when doing refraction work in the vicinity of high voltage power lines or large electrical installations. However, such filters might also effect the desired first arrival signal. Weather can cause interference in any of several ways. Wind can generate noise at the geophones, especially if there is considerable vegetation near ground level. Shallow burial of the geophones may reduce wind noise, and clearing vegetation from around the geophones may reduce noise by preventing stalks, stems or leaves from tapping or scraping on the geophones or adjacent cable. Rain, of course, generates profound noise with raindrop impact on the geophones and ground. In addition, geophone and trigger cables should not be left deployed or attached to the instrument or near personnel during or with imminent lightning storms. Direct, intense sunlight (typical in the desert southwest) can cause an instrument LCD display to become dark and unreadable; displays should be kept out of the sun or covered; a clipboard works fine. Finally, frozen ground can act as a high velocity layer at or just below the ground surface which blinds the refraction method with a velocity reversal. Wait until spring. High voltage power lines can cause interference with seismograph timing as well as with the geophone signals. When raised before striking, the sledgehammer and trigger switch can act as an antenna and the instrument can falsely trigger in response to induced electrical currents. The author has found that wrapping one end of a coil of bare wire around the hammer head, placing the other end of the bare wire coil on the ground, and then standing on the coil, can sufficiently ground the hammer to eliminate false triggering. On occasion in areas of dry ground, pouring water on the coil to improve the grounding may also be necessary. Recording & Verification of Field Data Whether the geophone signals are recorded on paper, on disk, or manually written down, it is imperative that the first arrivals be clearly identified (or positively identifiable) before the opportunity to repeat the data collection is past. The instrument operator (or a competent member of the crew) must review the traces and make that decision. That person must also answer the question "Is this data interpretable, and does it make sense?" At this point, poor or marginal quality data can be erased, and data acquisition repeated to try to improve data quality. Field personnel experienced in interpretation, and who participate in interpretation, is thus an important part of field data quality control. Advice to the neophyte from the 'old geophysicist' was "run the gains wide open, get the first arrival; nothing else counts." Only about 7 percent of the energy imparted at the seismic signal source propagates as body (p-) waves (Richart et al., 1970); thus only a small fraction of the seismic signal detected by the geophones is the desired first arrival. None of the following seismic signal is used in standard p-wave refraction work. Part of the instrument operator's craft is to set sufficient gain on the geophone channels to clearly distinguish the first arrival, yet not overwhelm the following portion of the data recording with later arriving seismic signal. Signal amplifier gains on seismographs from the 1980's typically were manually preset before data acquisition. Knowledge learned by experience of the anticipated signal amplitudes was part of successful data acquisition. More modern instruments have the option of dispensing with preset gains. However, autogain options might not provide accurate presentations of the first arrivals without further adjustments. Quality Control of Field DataFurther quality control and occasional rough preliminary depth interpretations can frequently be made by visual inspection of the trace data in the field. ASTM D5777 describes three quality control tests which can be applied to data in the field. The irregularity test checks for consistent first arrival times. If arrival time differences, such as deviation from straight line slopes, are excessive, then first arrival picks may be in error, or the data plotting or entry may be wrong. Also, the geological conditions may be complex or considerable noise may be present in the data. The reciprocal time test is used to verify a consistent travel time between forward and reverse profile data. Excessive differences between the profiles may indicate errors in first arrival picks or data plotting or entry. Parallelism checks for similarities between time-distance relations for various shotpoints or lines on the same geologic profile. Excessive differences may indicate an error in first arrival picks or data entry or plotting. The author considers good quality control practice to include physical marking of the initial choice of first arrivals on a paper printout of the geophone data by the field instrument operator. Data trends can be observed and visual quality control checks, including irregularity, reciprocal time and parallelism, performed when making these markings. Locations of velocity changes (change in the trend line of the first arrivals) can be used to estimate depths of velocity changes. The depth of the velocity change is estimated to be roughly about one- fourth to one-third the distance between the shotpoint and the change in the trend line. P-wave velocities can also be estimated by observing the time needed for the first arrivals to reach different geophones. The velocity is distance divided by time. Markings on trace printouts should not obscure the trace data, especially if data is not stored digitally. DATA PROCESSING Data processing for the seismic refraction method consists primarily of accounting for energy source and geophone locations, making adjustments or topographic changes along the geophone array profiles, and determining the first arrival times at the geophones. This step can be relatively simple, and has been effectively performed manually on analog trace records for decades. Where uniform grades are present or flexibility in array orientation is practical, topographic considerations can be minimized. Processing begins in the field as the data is inspected and verified during acquisition and topographic conditions along the arrays are documented. For single channel seismographs with no permanent record, the first arrival field pick is the only data available for analysis. For larger seismographs with paper trace record or digital storage, post-field data processing begins with the final determination of (picking) first arrivals. Manual picking first arrivals off of analog paper traces consists of verifying, and if appropriate, modifying the field picks. Digitally stored field traces can be further processed or manipulated in the office. Gains may be modified and digital filtering might be applied. It should be noted that filtering has the potential to modify the desired signal as well as undesired signals. Automatic and/or computer assisted first pick routines within interpretation software packages might be used. Source and geophone array geometry and first arrival time data can then be put into an appropriate format for interpretation by whatever methods. The obvious choice for first arrival picks at geophones close to energy sources in some geologies may not always be correct. Near surface, low modulus materials exist which have p-wave velocities slower than the speed of sound in air (about 350 m/s or 1,150 f/s). Assuming such a surficial horizon is present and the geophone spacing is sufficiently close to detect that horizon, the geophone response to the air wave from the energy source impact noise will occur before the desired p-wave response. Figure 10 presents geophone trace data with a readily identifiable air wave arriving before the first ground-borne p-wave. Apparent p-wave velocities close to 350 m/s (1,150 f/s) must be treated with suspicion, and used in interpretation with caution. In cases with relatively thin low velocity surficial horizons with actual p-wave velocities close to the speed of sound, interpretation errors may be small. However, relatively thick, very low velocity horizons could represent important or crucial geologic conditions such as very low density soils or landslide masses.Figure 10. Energy source air wave (impact noise from 5 sledge hammer impacts) arrival occurring before the first ground-borne p-wave arrival. Note field markings of first arrival picks. The surficial horizon was interpreted to be a low density, potentially collapsible soil with a thickness of about 6 feet. Seismic line was performed at a proposed highway embankment site; permits for vehicular access for subsurface exploration had been delayed pending cultural resource considerations. INTERPRETATION Interpretation of the first arrival data into a profile of subsurface velocities (also known as inversion) can be accomplished in several ways. Methods involving fairly laborious calculations or dependence on graphical chart solutions have been in use for decades. Hand calculators with transcendental functions (in the 1970's), followed by personal computers (by the 1980's), eliminated much drudgery and facilitated effective interpretations of refraction seismic data for geotechnical engineering use. Commercial software packages are now available for a wide range of interpretation methods and concepts. In the manual EM 1110-1-1802, the U.S. Army Corps of Engineers has recognized three main groups of interpretation methods. These are intercept-time methods (ITM), reciprocal or delay-time methods, and ray- tracing (ASCE, 1998). These methods, and their applicability to different situations, are also reviewed in ASTM D5777. Recently developed, perhaps extreme versions of ray-trace interpretation include optimization routines coupled with finite-difference techniques that converge on solutions after tens of thousands of iterations. For purposes of this paper, two very different interpretation concepts will be discussed. The intercept-time method (ITM) represents relatively simple interpretation methods with calculations that can be done with a calculator or implemented on a computer spreadsheet. Also operating in a PC environment, automatic optimization provides a very different type of interpretation which effectively compliments more traditional interpretation results. Time-Distance Plots Time-distance plots are a basic format for presentation of first arrival data needed for refraction interpretation. Distance which encompasses the geophone array and shotpoints is plotted along the x-axis. Recorded time for the energy to travel between shotpoints and individual geophones are plotted along the y-axis at each geophone distance. Figure 11 presents an example set of time-distance plots for a twelve geophone array seismic line with five shotpoints. This line was completed in an unmitigated cultural resource site on a highwayalignment east of Payson, Arizona. Until mitigation and clearance are complete, borings and test pits cannot be completed at that site. The site was nearly level, so that topographic considerations were not needed for interpretation. Mooney (1984) provides an extensive set of time-distance plot examples with possible interpretations for a variety of subsurface geometries. Manual EM 1110-1-1802 (ASCE, 1998) also provides several examples of complex subsurface geologic conditions with corresponding time-distance plots. Such catalogs are a valuable tool for the novice interpreter. However, it must be understood that interpretations are non-unique. Site characterization or design needs may require further verification and exploration to meet project objectives. Figure 11. Example time-distance plot for 12-geophone array with 5 shotpoints at the ends, center and quarter points along the array. Note complex arrival time data from the middle shotpoint, with a distinct apparent velocity reduction between geophones 4-5 and 8-9, as well as between geophones 9-10 from the reverse profile (backshot). This may indicate horizontal as well as vertical velocity variations in the p-wave profile. Intercept-Time Method (ITM) Interpretation by ITM assumes subsurface material zones where each zone has a uniform velocity. Velocities are interpreted by determining straight-line (or nearly so) slopes along the various portions of the time- distance plot. A minimum of three data points are needed to confidently interpret each slope. Velocities are calculated as distance traveled divided by time elapsed for each portion of the time-distance plot. Interfaces between different material zones are assumed to be planar, although dipping interfaces can be readily interpreted along the two-dimensional geophone array. However, true dip is a three-dimensional problem. Actual velocities for dipping layers are interpreted by using the harmonic mean of the interpreted forward and reverse profile velocities. Thin zones whose influence is less than three data points cannot have their velocities confidently interpreted, or typically even be detected. This fundamental limit is part of the blind zone problem, where the refraction method has a limit of resolution in the vicinity of seismic interfaces. Manual EM 1110-1- 1802 discusses the blind zone problem. Fundamental physics prevents straightforward detection and interpretation of lower velocity horizons underlying higher velocity horizons, because refraction occurs downward rather than upward towards the surface. A higher velocity horizon underlying a velocity reversal condition would be interpreted to be deeper than it's true depth. 0 10 20 30 40 50 60 0 20 40 60 80 100 120 Distance, feet Time, millisecDepth interpretation formulas for 2 layers can be found in ASTM D5777. Multi-layer formulas can be found in Manual EM 1110-1-1802 (ASTM, 1998) and in Mooney (1984). ITM interpretations of 12-channel seismograph data can readily discern up to 3 layers, if present within the depth of investigation. Multiple shotpoints along the geophone array permit interpretations of changing interface depths and layer velocities, especially in the shallower layers. Depth and velocity interpretations in the deeper portion of a profile may become more speculative when shallower interface depths and velocities are complex and changing across the profile. Figure 12. Interpretation of Figure 11 time-distance data using ITM and non-linear optimization methods. Note that, with sufficient shotpoints, both methods can indicate lateral velocity changes in the near surface. Both methods also indicate a deeper down-dipping higher velocity horizon. Depth increments are 5-foot intervals. The ITM depth of investigation is inferred, while the optimization interpretation is calculated. The time-distance plots presented in Figure 11 demonstrate capabilities and limits of ITM interpretation. An ITM interpretation of the Figure 11 data is presented in Figure 12. Five shotpoints permits breaking the overall interpretation into a series of smaller interpretations for the shallower part of the subsurface profile. Two major aspects of the subsurface profile are indicated. A low velocity surficial material horizon is missing near the profile center, and the (interpreted) weathered rock contact dips downward from the beginning towards the end of the seismic line. No depth interpretations can be made from the centerpoint (60 ft) profiles; the velocity drops or a roughly vertical offset occurs beyond two geophones from the center shotpoint. Depth interpretations can be made from the shotpoints at the ends and quarterpoints. Horizontal velocity change interpretations are made based on the results from the five shotpoints. Forward and reverse profiles obtained only from the array ends could only interpret one velocity per layer. Finally, the geophone at a 25-foot distance appears to be at a local surficial low velocity anomaly. Arrival times for that geophone are late for each of the other profiles. For the purposes of interpretation for the highway project, data from that geophone was ignored. This interpretation is essentially a manual process performed by the interpreter, with calculations and presentation of the interpretation, implemented on a PC spreadsheet. Velocity Contours -30 -25 -20 -15 -10 -5 0 5 10 15 20 25 30 35 40 45 0 10 20 30 40 50 60 70 80 90 100 110 120 130 Distance, ft Depth, ft 1500-1999 2000-2499 2500-2999 3000-3499 2000 ~3300 ~2800 1700 ~2400 1800 1000 1000 1000-1499 171704 iterations Non-Linear Optimization Interpretation ITM Interpretation 30 60 90 120 p-wave velocities are in ft/secNon-linear Optimization Computer programs capable of generating interpretations without operator participation (after entering input data or preparing input data files) are available. A commercially available non-linear optimization routine (utilizing a simulated annealing algorithm) using a finite-difference model mesh to calculate travel times represents a recent advance in ray-tracing interpretation similar to tomography. This interpretation includes two very important features frequently missing from other interpretations. The first is interpretation of gradational, gradual velocity changes. The second is an explicit interpretation of the zone of investigation for the seismic data. Input files with locations and elevations of sources and receivers and first arrival times are prepared for an optimization run. Topographic variations are accounted for in the finite-difference mesh. A combined surface and downhole source and geophone array could conceivably be interpreted. The optimization software performs between thousands and tens of thousands of iterations to obtain best matches between actual arrival times and distances between sources and receivers and modeled, calculated times based on ray paths through the finite- difference mesh. Layers with uniform velocities are not represented. Rather, each element which is part of a ray-path in the finite-difference grid has an optimized velocity. Elements which are not parts of ray-paths do not have any velocity; the extent of the investigated zone is the part of the finite-difference mesh with optimized velocities. Presentation of gradational, gradual vertical and lateral velocity changes through the model can be presented in color bar form, where different velocities are presented as different colors. Model results can also be output as grid coordinates and velocity for further analysis or presentation utilizing contouring software. Effective optimization interpretations may require more field data, especially shotpint sources, than simpler interpretation methods such as ITM. An optimization model interpretation of the Figure 11 time-distance plots is presented in Figure 12. For the Figure 12 presentation, model coordinates and velocities were imported into a spreadsheet, and certain velocity ranges assigned square data points to match the finite-difference mesh, to generate a simple black-and-white velocity pseudo-contour presentation. Similarities between the optimized velocity model and the ITM model are apparent. A down-dipping trend for higher, deeper velocity is interpreted in both models. Both interpretations show variations in near surface velocities, including higher very near surface velocity at the center of the array. However, the optimized model presents an explicit zone of investigation. Such a zone is inferred for the ITM model. Figure 13 presents interpreted velocity with depth profiles at selected distances across the Figure 12 optimized model. A velocity with depth profile for the ITM interpretation is also presented. Gradual optimized velocity changes with depth are apparent, while the ITM approach must break the profile into a few discrete zones. Such an interpretation may be especially useful in soil and weathered rock profiles where gradual velocity changes would be anticipated. A typical trend for velocity change of cohesionless soil with depth, based on changes in soil modulus which scale to the square root of the effective stress (Richart et al., 1970), is also presented in Figure 13. Below about 2 feet, the optimized vertical velocity trend at 90 feet is relatively close to the soil modulus trend to at least a depth below about 12 or more feet. Optimized vertical velocity trends at 30- and 60-feet, however, appear to be dissimilar to the soil modulus trend. Optimized vertical velocity profile trends at several highway embankments and in cohesionless sand deposits have been similar to the cohesionless soil modulus trend. Other geologic situations may consist of discrete horizons with very large velocity velocity contrasts at horizon boundaries, such as presented in Figure 9. Those situations may be better interpreted using layer- based methods such as ITM. Resolution limits of the finite-difference grid may limit the effectiveness of interpretations in the near-surface region. This could require very close source and geophone spacings for excavatability studies for utility trenching and other very shallow investigations.Figure 13. Vertical velocity profiles with depth for interpreted results in Figure 12. Note how the ITM interpretation of discrete horizons is interpreted as a relatively smooth velocity gradient by the non-linear optimization routine. STRENGTHS, WEAKNESS & COST EFFECTIVENESS As discussed throughout this paper, the seismic refraction technique provides a simplified characterization of large portions of the subsurface in two-dimensional profile. This information complements borings and test pits, which typically provide point or one-dimensional (vertical) information. Above the water table, p-wave velocity is a measure of soil or rock mass modulus at in-situ conditions. Thus, the measured parameters of p- wave velocity and depth can be used to help characterize both material mass geometry and behavior. Weaknesses of the technique include the limitations of velocity reversals and blind zones previously discussed, and the possibility of oversimplified subsurface geometries from the resolution limits inherent in the methods used or equipment deployed. Issues in Specific Geologic Settings Several specific situations can effectively demonstrate limitations of seismic refraction. An interpretation may not be able to distinguish whether a relatively high velocity horizon is a weathered rock contact or a water table. Where a water table is suspected, additional subsurface exploration might be needed to verify it's presence or absence. It should be noted that initial seismic refraction work might be considered subsurface reconnaissance that helps scope or direct further subsurface investigation. Cemented soil caps common in arid climates, sometimes referred to as caliche, can cause very strong velocity reversals. It is possible to have a 120-foot long geophone array, with an anticipated 30 foot depth of investigation, have an actual depth of investigation of only about four to ten feet. Non-traditional analyses such as non-linear optimization will provide explicit interpretations of depths of investigation. Alternatively, the deepest interpreted horizon interface might be conservatively considered to be the depth of investigation. Some sedimentary rock environments, where erosion exposed and been slowed by a relatively strong rock horizon, may have similar seismic refraction results. -35 -30 -25 -20 -15 -10 -5 0 1000 1500 2000 2500 3000 3500 4000 P-wave Velocity, ft/sec Depth, ft optimization at 30 ft optimization at 60 ft optimization at 90 ft soil modulus trend ITM profile at 60 ftWeathered granite rock masses may be highly weathered to decomposed along major fracture zones, yet have large slightly to unweathered corestones between these fracture zones. An interpreted p-wave velocity for such a rock mass will reflect an average for the low velocity highly weathered to decomposed material and the high velocity slightly to unweathered corestones. For rippability assessment, that average p-wave velocity may not be representative of the excavation difficulty which those large corestones represent. Rippability assessment in these conditions should also include rock core drilling to investigate the potential and condition of corestones. Results of coring may capture some corestone characteristics, and results of the seismic work may assist in extrapolating the coring information across the site. Basalt flows can consist of very hard rock particles with relatively open fracturing exacerbated by shrinkage during cooling. These open fractures can attenuate seismic signals very rapidly. It is possible to have a formation with a representative p-wave velocity in excess of 10,000 f/s, in which the seismic signal has effectively disappeared in distances of about 60 to 80 feet. Rippability studies may need to be based on relatively short geophone arrays in these environments. Additionally, basalt flows can also serve as a cap material over softer rock or soil deposits. Strong velocity reversals are possible in basalt flow settings. Finally, subsurface conditions can be extremely variable, even chaotic, in these geologic settings. Cost Effectiveness Seismic refraction data can be collected, interpreted and integrated into an overall geotechnical report for considerably less than $1,000.00 per line (year 2000 dollars), depending upon the access, field conditions and level of detail of the interpretations. Seismic refraction can be very cost effective at sites needing considerable characterization of rock, where core drilling can be very expensive. Core drilling may still be necessary, but the complementary seismic refraction method provides effective site coverage and maximizes effective use of the knowledge obtained from the core drilling. Investigating areas with difficult access for drill rigs are obvious places where the method is very effective. Seismic refraction may be the only practical alternative for subsurface investigation in areas where drilling or other subsurface disturbance is prohibited, such as unmitigated cultural resource sites, or when vehicular access is not possible. Cost impacts of seismic refraction data as part of geotechnical investigations for large excavation projects such as highways in mountainous, rocky terrain, are difficult to quantify. If good, representative p-wave velocity data at large roadway cuts in weathered, fractured rock is available to potential contractors during bidding, then those contractors will have information to effectively address their potential for high-cost excavation. The potential for project cost savings reflected in the bids, and in reduced potential for changed conditions claims, may be very significant. Small investigations in rock environments with inadequate complexity or budget to justify core drilling can be effectively served using seismic refraction to obtain useful rock mass parameters. The data in Figure 9 is from such a small investigation. SUMMARY Seismic refraction is a cost effective means to obtain generalized subsurface information for geotechnical or geological characterization over relatively large areas. The method can be used in areas inaccessible to vehicles or of a sensitive nature where vehicles are prohibited. Efficient equipment and procedures have been available to the geotechnical engineering profession to develop experience with implementation and use of the results for over two decades. Continued improvements in equipment and interpretation software will make the method more effective and useful for geotechnical exploration and design of transportation facilities. Straightforward application of seismic refraction methods can be performed by geotechnical or geological personnel with appropriate knowledge and experience. Applications of seismic refraction methods by geophysical specialists can provide essential information in extraordinary or critical subsurface exploration situations.SELECT BIBLIOGRAPHY American Society of Civil Engineers (ASCE), 1998, Geophysical Exploration for Engineering and Environmental Investigations, Technical Engineering and Design Guides as Adapted from the US Army Corps of Engineers, No. 23, ASCE Press, Reston, VA, 7-23. American Society for Testing and Materials (ASTM), 2000, Annual Book of ASTM Standards, Volumes 04.08 Soil and Rock (I) & 04.08 Soil and Rock (II), ASTM, West Conshohocken, PA. Mooney, H.M., 1984. Handbook of Engineering Geophysics, Volume 1: Seismic. Bison Instruments Inc., 5706 West 36 th Street, Minneapolis, MN, 55416. Redpath, B.B., 1973. ?Seismic Refraction Exploration for Engineering Site Investigations.? Technical Report E-73-4, U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS. OTHER REFERENCES Bieniawski, Z.T., 1989, Engineering Rock Mass Classifications: John Wiley & Sons, N.Y., 8. California Department of Transportation (Caltrans), 1978. Calculating Earthwork Factors Using Seismic Velocities: Report No. FHWA-CA-TL-78-23, August. Caterpillar Tractor Company (Caterpillar), 1984, Caterpillar Performance Handbook 15 th Edition, Peoria, Illinois. --------------------1993, Caterpillar Performance Handbook 24 th Edition, Peoria, Illinois. Richart, F.E., Hall, J.R. Jr. and Woods, R.D., 1970, Vibrations of Soils and Foundations, Prentice-Hall, Englewood Cliffs, New Jersey. Rucker, M.L., 1996, Integrating the refraction seismic method into stream crossing characterization for scour and excavation conditions, in Shackelford, C.D., Nelson P.P. and Roth, M.J.S. (eds.), Geotechnical Special Publication No. 58: ASCE, New York, 1163-1177. ------------------ 2000, Earthwork factors in weathered granites by geophysics, in Nazarian, S. and Diehl, J. (eds), Geotechnical Special Publication No. 108: ASCE, Reston, Virginia, 201-214. Rucker, M.L. and Keaton, J.R., 1998, Tracing an earth fissure using seismic refraction methods with physical verification, in J.W. Borchers (ed.), Land Subsidence Case Studies and Current Research, Proceedings of the Dr. Joseph F. Poland Symposium on Land Subsidence, Special Publication No. 8, Association of Engineering Geologists, Star Publishing Company, Belmont, California, 207-216. Transportation Research Board (TRB), 1996, Landslides, Investigation and Mitigation, Special Report No. 247, Turner, A.K., and Schuster, R.L. (eds), National Academy Press, Washington, D.C., 241. Van Heerden, W.L., 1987, General relations between static and dynamic moduli of rocks: International Journal Rock Mechanics & Geomechanics Abstracts, 24(6), 381-385.