Regional Seismic Discriminants Research Articles

Identification of Quarry Blasts Aided by Infrasound Data

We present our analysis of the detections from the Piszkés-tető, Hungary infrasound array that has been operational since May 2017. We introduce an automatic search method to associate infrasound detections with a preliminary list of explosions detected by the seismic network. Once infrasound detections are associated with seismic events, we relocate the events using both the seismic arrival time and infrasound azimuth observations. We validate our methodology on ground truth events, i.e., explosions confirmed by the mine operators. We demonstrate that seismo-acoustic locations are able to identify the source of explosions even for closely spaced quarries. The discriminative power of the infrasound azimuth is strongest at near-field distances, where local and regional seismic discrimination methods have the most difficulties.

Gradual Fault Weakening with Seismic Slip: Inferences from the Seismic Sequences of L’Aquila, 2009, and Northridge, 1994

We estimate seismological fracture energies from two subsets of events selected from the seismic sequences of L’Aquila (2009), and Northridge (1994): 57 and 16 selected events, respectively, including the main shocks. Following Abercrombie and Rice (Geophys J Int 162: 406–424, 2005), we postulate that fracture energy (G) represents the post-failure integral of the dynamic weakening curve, which is described by the evolution of shear traction as a function of slip. Following a direct-wave approach, we compute mainshock-/aftershock-source spectral ratios, and analyze them using the approach proposed by Malagnini et al. (Pure Appl. Geophys., this issue, 2014) to infer corner frequencies and seismic moment. Our estimates of source parameters (including fracture energies) are based on best-fit grid-searches performed over empirical source spectral ratios. We quantify the source scaling of spectra from small and large earthquakes by using the MDAC formulation of Walter and Taylor (A revised Magnitude and Distance Amplitude Correction (MDAC2) procedure for regional seismic discriminants, 2001). The source parameters presented in this paper must be considered as point-source estimates representing averages calculated over specific ruptured portions of the fault area. In order to constrain the scaling of fracture energy with coseismic slip, we investigate two different slip-weakening functions to model the shear traction as a function of slip: (i) a power law, as suggested by Abercrombie and Rice (Geophys J Int 162: 406–424, 2005), and (ii) an exponential decay. Our results show that the exponential decay of stress on the fault allows a good fit between measured and predicted fracture energies, both for the main events and for their aftershocks, regardless of the significant differences in the energy budgets between the large (main) and small earthquakes (aftershocks). Using the power-law slip-weakening function would lead us to a very different situation: in our two investigated sequences, if the aftershock scaling is extrapolated to events with large slips, a power law (a la Abercrombie and Rice) would predict unrealistically large stress drops for large, main earthquakes. We conclude that the exponential stress evolution law has the advantage of avoiding unrealistic stress drops and unbounded fracture energies at large slip values, while still describing the abrupt shear-stress degradation observed in high-velocity laboratory experiments (e.g., Di Toro et al., Fault lubrication during earthquakes, Nature 2011).

Scaling Transition in Earthquake Sources: A Possible Link Between Seismic and Laboratory Measurements

We estimate the corner frequencies of 20 crustal seismic events from mainshock–aftershock sequences in different tectonic environments (mainshocks 5.7 < M W < 7.6) using the well-established seismic coda ratio technique (Mayeda et al. in Geophys Res Lett 34:L11303, 2007; Mayeda and Malagnini in Geophys Res Lett, 2010), which provides optimal stability and does not require path or site corrections. For each sequence, we assumed the Brune source model and estimated all the events’ corner frequencies and associated apparent stresses following the MDAC spectral formulation of Walter and Taylor (A revised magnitude and distance amplitude correction (MDAC2) procedure for regional seismic discriminants, 2001), which allows for the possibility of non-self-similar source scaling. Within each sequence, we observe a systematic deviation from the self-similar $$ M_{0} \propto \mathop f\nolimits_{\text{c}}^{ - 3} $$ line, all data being rather compatible with $$ M_{0} \propto \mathop f\nolimits_{\text{c}}^{ - (3 + \varepsilon )} $$ , where e > 0 (Kanamori and Rivera in Bull Seismol Soc Am 94:314–319, 2004). The deviation from a strict self-similar behavior within each earthquake sequence of our collection is indicated by a systematic increase in the estimated average static stress drop and apparent stress with increasing seismic moment (moment magnitude). Our favored physical interpretation for the increased apparent stress with earthquake size is a progressive frictional weakening for increasing seismic slip, in agreement with recent results obtained in laboratory experiments performed on state-of-the-art apparatuses at slip rates of the order of 1 m/s or larger. At smaller magnitudes (M W < 5.5), the overall data set is characterized by a variability in apparent stress of almost three orders of magnitude, mostly from the scatter observed in strike-slip sequences. Larger events (M W > 5.5) show much less variability: about one order of magnitude. It appears that the apparent stress (and static stress drop) does not grow indefinitely at larger magnitudes: for example, in the case of the Chi–Chi sequence (the best sampled sequence between M W 5 and 6.5), some roughly constant stress parameters characterize earthquakes larger than M W ~ 5.5. A representative fault slip for M W 5.5 is a few tens of centimeters (e.g., Ide and Takeo in J Geophys Res 102:27379–27391, 1997), which corresponds to the slip amount at which effective lubrication is observed, according to recent laboratory friction experiments performed at seismic slip velocities (V ~ 1 m/s) and normal stresses representative of crustal depths (Di Toro et al. in Nature in press, 2011, and references therein). If the observed deviation from self-similar scaling is explained in terms of an asymptotic increase in apparent stress (Malagnini et al. in Pure Appl Geophys, 2014, this volume), which is directly related to dynamic stress drop on the fault, one interpretation is that for a seismic slip of a few tens of centimeters (M W ~ 5.5) or larger, a fully lubricated frictional state may be asymptotically approached.

Treatment of Correlation for Clustered Reference Data in Kriging Regional Seismic Discriminants

Abstract Substantial improvements to the performance of discriminating explosions from earthquakes have been demonstrated by correcting regional seismic phase amplitudes or their ratios (e.g., Pn / Sn and Pn / Lg ) for variations in propagation path and station effects. One of the most promising methods for estimating the corrections and their uncertainties is a form of Bayesian kriging. However, there are valid concerns regarding the robustness of the kriged amplitude corrections to aberrant calibration data, that is, that regional phase amplitudes from earthquakes with anomalous mechanisms, depths, and/or paths could significantly bias the amplitude correction grids. The methodology of Bottone et al. (2002) will properly account for such situations, unless the residual variance is correlated. To first demonstrate the problem, the approach of Bottone et al. (2002) is applied to explosions and clusters of earthquakes with distinct focal mechanisms at the Nevada Test Site (NTS). The results indicate that serious discrimination errors (i.e., miscategorizations of explosions as earthquakes) can occur. In the case of highly correlated clusters, the localized reference data are weighted too high in estimating the corrections and the posterior calibration uncertainties become too small. To address this issue, an extension of the kriging methodology, incorporating an additional correlation length to treat correlated focal mechanisms, is presented and tested on the NTS data sets and on large samples of explosions and earthquakes worldwide. Application to the worldwide data sets indicates that the power of the hypothesis test to reject earthquakes as explosions is marginally reduced, without miscategorizing any known explosions and providing a robust treatment of clustered data.

Seismic Discrimination of the May 11, 1998 Indian Nuclear Test with Short-period Regional Data from Station NIL (Nilore, Pakistan)

Regional seismic discriminants for the May 11, 1998 Indian underground nuclear test(s) and earthquakes recorded at station NIL (Nilore, Pakistan) provide new data to test strategies that can be used to monitor the Comprehensive Nuclear-Test-Ban Treaty (CTBT). Three categories of regional discriminants (ratios of P- and/or S- wave energy) were measured on short-period (0.5-6 Hz) seismograms: P/S amplitude ratios (phase ratios) measured in the same frequency band, P- and S -wave spectral ratios (i.e., low frequency to high frequency for the same phase) and P/S cross-spectral ratios (i.e., low frequency S- wave to high frequency P -wave). The P/S amplitude ratios show good separation of the Indian nuclear test and regional earthquakes for Pn/Lg and Pn/Sn, however Pg/Lg does not discriminate as well. Pn/Lg and Pn/Sn discriminate well at frequencies as low as 0.5-2 Hz, especially after accounting for path effects. This observation differs from previous studies that report poor separation of earthquakes and explosions at lower frequencies. The P/S amplitude ratios do not show any magnitude dependence, suggesting that forming the ratios in a fixed frequency band cancels the effects of source size-corner frequency scaling. Spatial variability of the observed discriminants arises from variations in crustal waveguide and/or attenuation structure (path propagation effects). Grouping amplitude ratios for earthquakes with paths similar to the Indian test greatly improves discrimination. Removing distance trends does not generally improve discrimination. Accounting for path effects with Bayesian kriging significantly improves discrimination. Spectral ratios (e.g., Pn [0.5-1 Hz]/Pn [4-6 Hz]) and cross-spectral ratios (e.g., Lg [1-2 Hz]/Pn [4–6 Hz]) show distance and magnitude dependence. We developed a technique for simultaneously removing the effects of distance and source size-corner frequency scaling on amplitude and spectral ratios. The technique uses a gridsearch to find several parameters that characterize the observed distance and magnitude dependence. Discrimination of the Indian test improved dramatically after the distance and magnitude trends were removed from the spectral and cross-spectral ratio data.

Path Corrections for Source Discriminants: A Case Study at Two International Seismic Monitoring Stations

Improving the performance of short-period regional seismic discriminants by applying propagation corrections is explored using observations from two seismic monitoring stations in Asia. Frequency-dependent regional phase amplitude ratio measurements at stations NIL and ZAL for earthquakes and underground nuclear explosions were obtained from the prototype-International Data Center (pIDC) that has been established for developing monitoring capabilities of the Comprehensive Nuclear-Test-Ban Treaty (CTBT). The pIDC discriminant measurements have large scatter, much of which is attributed to wave propagation effects in the heterogeneous crustal waveguide. Linear regressions indicate that the phase ratios are correlated with topographic characteristics along the individual paths, providing an empirical means for correcting for path effects beyond conventional distance corrections. Kriging, a spatial multiple regression algorithm, also reveals coherent spatial patterns in the data indicative of regional path effects. Using available high-resolution topography data, correction of regional P/S ratios for the best models obtained from multivariate regressions systematically reduces the data variance relative to distance corrections alone, as has been observed for other data sets. The reduced scatter in the measurements increases the separation between earthquake and explosion populations in most cases, enhancing the regional discriminant performance. The path-corrected discriminants isolate explosions better for NIL than for ZAL, even though some of the explosion sources are located in a common source area. Kriging achieves comparable or superior variance reduction for the discriminant measures, without requiring knowledge of the path structure, although this may not result in improved discriminant performance. While always desirable, corrections for heterogeneous path effects may prove inadequate in some cases, notably when phase blockage occurs or when strong attenuation eliminates the diagnostic high-frequency energy.

Discrimination information in phase amplitude thresholds with application to western China regional data

AbstractThis article develops a regional seismic discrimination method using information inherent in phase amplitudes that are unmeasurable due to small signal amplitudes and high noise levels. The method, quadratic negative evidence discrimination (QNED), is an enhancement to the teleseismic techniques proposed by Elvers (1974) and is extended to regional discrimination. The method presented in this article is developed for a single seismic station and makes use of the empirical evidence in the regional Pg versus Lg discriminant (see Pomeroy et al., 1983). We develop the equations necessary to compute the station-specific, missed-explosion, and false-alarm error rates. These error rates depend on the required minimum signal-to-noise ratio (S/N) and can be adjusted, within limits, to desired levels. We also show that these equations are an accurate assessment of the errors in seismic discrimination. We propose that many of the current approaches to assessing seismic discrimination errors are often overly optimistic. For some applications, this disparity can be significant. For an application to Western China regional data [Pg versus Lg (1.5 to 3 Hz)], a widely used estimate of the missed-explosion error rate is 20%, leading to the perception that explosions can be identified with an accuracy rate of 80%. A proper accounting of the missed-explosion error rate, using QNED, shows that explosions can be identified with accuracy rate of only 73%.

Statistical analysis of irregular wave-guide influences on regional seismic discriminants in China: Additional results forPn/Sn,Pn/Lg, andPg/Sn

AbstractReducing scatter in measurements of regional-phase amplitude ratios is desirable for seismic discrimination applications and also provides insight into crustal structure controls on energy partitioning of regional seismic waves. Our previous analysis (Fan and Lay, 1998) of the regional seismic discriminant, Pg/Lg, indicated that variations in crustal structure cause path-specific fluctuations of those amplitude ratios for earthquakes recorded at broadband station WMQ in western China. In this study, we extend our multivariate regression analysis to Pn/Sn, Pn/Lg, and Pg/Sn amplitude ratios recorded at WMQ in frequency bands of 0.75 to 1.5 Hz, 1.5 to 3.0 Hz, and 3.0 to 6.0 Hz for the suite of path-specific parameters: path length, mean path elevation, variance of topography along the path, rms topographic slope variations, mean crustal thickness, and mean sediment thickness. Optimal three- and four-parameter models all achieve reductions in variance of the measurements relative to conventional distance corrections. At low frequency, the improvements for ratios involving Pg can be more than a factor of 2. For all short-period amplitude ratios, mean path elevation seems to play an important role. While crustal thickness and sediment thickness affect ratios involving Pg, topographic variance and surface slope variations have more influence on ratios involving Pn. Strong crustal variations associated with the structure of the Tibetan Plateau are responsible for much of the amplitude variations.

Regionalized versus single-station wave-guide effects on seismic discriminants in western China

Abstract Regional seismic signals play an important role in identifying low-magnitude explosion and earthquake sources for monitoring the Comprehensive Test Ban Treaty (CTBT). To enhance the performance of seismic discriminants, regional phases need to be corrected for propagation effects of large- and intermediate-scale crustal wave-guide heterogeneity. Multivariate regression analysis using parametric path characteristics has shown promise for reducing the propagation-induced scatter in short-period regional phase amplitude ratios. While path length is an important parameter for most seismic phase amplitudes, other parametric features of crustal wave-guide structure are often significantly correlated with regional amplitude ratios of P-wave energy (Pn, Pg) to S-wave energy (Sn, Lg) as demonstrated by our prior work on station WMQ. A question remains as to whether these effects can be regionalized so that they apply to all observations in a given region. We explore the viability of regionalization of wave-guide effects on regional seismic discriminants for frequencies of 0.75 to 6 Hz in central Asia and southwest China by analyzing single- and multi-station data. Linear regressions for single-station data indicate that the significance of specific path parameters extracted from available crustal models varies from one station to another and from one subregion to another. In general, corrections for optimal combinations of wave-guide parameters obtained from multivariate regression reduce the data variance more than conventional corrections for path length alone for both single-station and multi-station data. However, in central Asia, several stations show path dependence on certain variables with opposite signs, so regionalization with multi-station data does not perform well. This is due to highly path-specific effects. In southwest China, there is greater commonality in behavior among stations, and fairly good regionalization can be achieved. However, the resulting variance reduction is still less than that for individual stations because some path effects appear to be direction dependent. Our results indicate that individual station calibration is the preferred procedure, but this should include corrections for path effects beyond simple distance correction.

Correcting regional seismic discriminants for path effects in western China

Abstract The effect of path on regional seismic wave propagation can be significant. In an effort to improve discriminant performance, we explore the effect of path upon Pg/Lg ratios. Our primary objective is to find path corrections that reduce scatter within earthquake and explosion ratio populations, while at the same time increasing the separation between the two populations. We emphasize the 1.5- to 3-Hz and 2- to 4-Hz bands, as Pg and Lg in these bands can often be observed at smaller magnitudes and greater distances than higher-frequency bands, which have previously been shown to be reliable discriminants. For data, we use 271 earthquakes from northwest China, 25 nuclear explosions from the Kazakh test site (KTS), and one nuclear explosion from the Lop Nor test site recorded at station WMQ. We also use 185 earthquakes from the same region and seven nuclear explosion from the Lop Nor test site recorded at station AAK. Event-station distances range from 200 to 1400 km, earthquake magnitudes range between mb 2.5 and 6.2, and explosion magnitudes range between mb 4.5 and 6.5. In addition to ratio-distance trends, we examine Pg/Lg ratio-parameter trends related to topography, basin thickness, and crustal thickness. The parameters we consider are mean, roughness, gradient mean, and gradient roughness of the topography, basin thickness, and crustal thickness along each event-station path. We also consider the same parameters after weighting by path length. Through linear regressions, we found path corrections that reduce scatter within event populations, and we also found path corrections that increase the separation between earthquakes and KTS explosions recorded at WMQ. We obtained the best improvement in discrimination performance at WMQ by removing the trends of topography roughness, mean topography, and the gradient of basin thickness after weighting the parameters by path length. For AAK, we found that removing the trends of mean topography and the basement roughness improved discrimination performance over the uncorrected case. However, unlike WMQ, weighting these parameters by path length degraded discriminant performance. Because we see no predictable or repeatable trends for “adjacent” central Asian stations and overlapping regions of interest, we recommend an even more empirical approach to correcting for the effect of path. Where earthquakes are abundant, such as the Tian Shan, contouring a grid of ratio residuals (for each band of interest) may be a simpler method of finding appropriate path corrections.

A procedure for estimation of source and propagation amplitude corrections for regional seismic discriminants

We outline a procedure for the estimation of frequency‐dependent source and propagation amplitude corrections for regional seismic discriminants (source path amplitude correction (SPAC)). For a given station and phase a number of well‐recorded earthquakes are inverted for source and path corrections. The method assumes a simple Brune [1970] earthquake‐source model and a simple propagation model consisting of a frequency‐independent geometrical spreading and frequency‐dependent power law Q. The inverted low‐frequency levels are then regressed against mb to derive a set of corrections that are a function of mb and distance. Once a set of corrections is derived, effects of source scaling and distance as a function of frequency are applied to amplitudes from new events prior to forming discrimination ratios. The resulting discriminants are normally distributed and amenable to multivariate feature selection, classification, and outlier techniques. To date, most discrimination studies have removed distance corrections once a particular amplitude ratio is formed (distance corrected ratio (DCR)). DCR generally works well for phase ratios taken in a particular frequency band. However, when different frequency bands are combined (for phase spectral ratios or cross spectral ratios), significant source‐scaling effects (e.g., corner‐frequency scaling) can remain, causing the discriminants to vary as a function of event size and to be nonnormally distributed. It is then often necessary to construct nonphysical transformations in an attempt to make the discriminants multivariate normal. The SPAC technique can be used to construct discriminants that are multivariate normal by using simple physical seismic source and propagation models. Moreover, phase amplitude residuals as a function of frequency can be spatially averaged and used as additional path‐specific corrections to correct for additional propagation effects such as phase blockages.

Statistical analysis of irregular wave-guide influences on regional seismic discriminants in China

Abstract Short-period regional phases play an important role in identifying low-magnitude seismic events in the context of monitoring the Comprehensive Test Ban Treaty. Amplitude ratios of regional phases comprised mainly of P-wave energy (Pn, Pg) to those comprised mainly of S-wave energy (Sn, Lg) effectively discriminate between explosions and earthquakes in many regions, particularly at frequencies higher than 3 Hz. At lower frequencies, discrimination is usually poor due to large scatter that causes overlapping of event populations. Scatter in regional discriminant measures such as Pg/Lg ratios is caused by both source and propagation effects, and reducing the scatter imparted by the latter is essential to improving the discriminant performance when events do not share identical paths. Regional phases experience distance-dependent amplitude variations due to effects such as critical angle amplification, geometric spreading, and attenuation. Discriminant measures are usually corrected for empirically determined distance trends for a given region, but large scatter persists after such corrections. This study seeks to develop more sophisticated empirical corrections for path properties in order to further reduce the scatter in regional discriminant measures caused by propagation effects. Broadband seismic waveforms recorded at station WMQ, in western China, demonstrate that regional Pg/Lg ratios show significant distance dependence for frequencies less than 6 Hz. However, variations in crustal structure cause additional path-specific amplitude fluctuations that are not accounted for by regionally averaged distance corrections. Blockage of Lg phases on paths traversing the margins of the Tibetan Plateau is one such effect. Regression analysis demonstrates that Pg/Lg ratios measured at WMQ display significant correlations with path-specific properties such as mean elevation, topographic roughness, basement depth, and crustal thickness. Multiple regressions using optimal combinations of parameters yield corrections that reduce variance in Pg/Lg measurements for frequencies less than 3 Hz by a factor of 2 or more relative to standard distance corrections. This should systematically enhance the performance of the Pg/Lg discriminant at low frequencies. The method presented here can be used for all regions and all short-period regional discriminants. It is likely that the extraordinary crustal heterogeneity in western China represents an extreme case of path-dependent effects.

A preliminary study of regional seismic discrimination in central Asia with emphasis on western China

AbstractIn support of a Comprehensive Test Ban Treaty, we have started to evaluate regional seismic-event discrimination capabilities for central Asia, emphasizing western China. We have measured noise and signal levels of over 260 earthquakes and 26 underground nuclear explosions recorded at the broadband, digital station WMQ in western China and over 120 earthquakes and five nuclear explosions at station AAK in Kyrgyzstan. The explosions are from the Kazakh Test Site (KTS) and Lop Nor, China. The earthquakes are mostly from northwest China. Event magnitudes (measured as either mb or ML) range between 2.5 and 6.5 and maximum event-station distance is about 1700 km. We formed phase, spectral, cross-spectral, short-period/long-period, and long-period ratios to test many possible event discriminants. All ratios were corrected for distance effects before forming ratio-versus-magnitude discrimination plots.The most consistent discriminants for separating earthquakes from explosions are P/S phase ratios for frequencies above 3 to 4 Hz. P-wave (f &amp;gt; 1 Hz) to Rayleigh-wave (0.03 Hz &amp;lt; f &amp;lt; 0.1 Hz) ratios are also effective, but surface-waves are difficult to measure for magnitudes below about 4.5. For the P/S ratios, separation between earthquakes and explosions increases with frequency, and for the P/R ratios, separation increases with increasing P frequency. The P (3 to 6 Hz)/S (0.75 to 1.5 Hz) cross-spectral ratios also separate earthquakes and explosions. These cross-spectral ratios may prove quite useful because, due to weak high-frequency Lg signals from explosions, the high-frequency P/Lg ratio sometimes cannot be evaluated, but the P (3 to 6 Hz)/Lg (0.75 to 1.5 Hz) can be evaluated successfully.P spectral ratios (3 to 6 Hz/0.75 to 1.5 Hz) also separate earthquakes and explosions but not quite as consistently as the P/S ratios. However, P spectral ratios may prove useful in cases where path effects block or highly attenuate Sn and Lg waves, rendering the P/S ratios unavailable. In contrast to Nevada Test Site (NTS) events, the Lg (3 to 6 Hz/0.75 to 1.5 Hz) spectral ratio does not separate earthquakes and nuclear explosions in central Asia. This observation may be a source effect, as Asian explosions are thought to be detonated in highly lithified rocks below the water table, while most of the smaller (mb &amp;lt; 4.8) NTS explosions have been detonated in poorly lithified rocks above the water table.For station AAK, the short-period spectral and cross-spectral ratios identify a few earthquakes (not recorded at WMQ) from the north Pamir region that plot with the Lop Nor explosions. The waveforms from these earthquakes lack surface waves and a distinct Lg phase. These events, which we interpret as having occurred in a subducted slab, do separate from the explosions on high-frequency (f &amp;gt; 4 Hz) P/S discrimination plots. For station WMQ, long-period R/L (Rayleigh-wave/Love-wave) ratios separate KTS explosions and Asian earthquakes. However, R/L ratios fail to identify the large Lop Nor explosions recorded at station AAK because the Love waves are as large as the Rayleigh waves, suggesting that strong tectonic release occurs with these explosions. Because we have examined only one path at a single station for these explosions, source-radiation pattern may be influencing this observation.

Variation of regional seismic discriminants with surface topographic roughness in the Western United States

Abstract Regional distance P and S waves behave differently with propagation distance due to complex elastic and anelastic effects in the crustal wave guide and uppermost mantle. It is well recognized that seismological discriminants based on P/S ratios or individual phase spectral ratios must be corrected for regionally dependent path effects for successful discrimination to be achieved. Such corrections usually involve only distance-dependent decay, parameterized to account for geometric spreading and attenuation factors averaged over some geographic region. However, large scatter persists in distance-corrected discriminants, which degrades their performance, and strategies are needed to further reduce the scatter by accounting for individual path variability. We demonstrate that characterizing some of the path variability by the readily measured surface topographic roughness along each path allows reduction of the scatter in some regional discriminants. Pg/Lg and Pn/Lg ratios and Pg, Pn, and Lg low-frequency/high-frequency spectral ratios for 80 earthquakes in the Western United States recorded at four stations of the Livermore NTS Network are analyzed for path effects. While all of these discriminants display distance dependence, log Pg/Lg is also found to be significantly correlated with average surface roughness along each path. The correlation is emphasized when the product of distance and roughness is used as an independent variable, and the data in the frequency range 1 to 6 Hz have the strongest trends. Corrections based on the correlations with distance and roughness can reduce the variance of the log Pg/Lg discriminant by 22 to 25% in the 2- to 4-Hz band. Log Pn/Lg and individual phase spectral ratios have stronger distance dependence than log Pg/Lg, but the correlation with surface roughness is weak for those measurements. It appears that interchange of Pg and Lg energy that reverberates in the crust accounts for the sensitivity to the gross wave-guide properties that are manifested in the surface roughness parameterization. The results support the concept of pursuing path corrections beyond purely range-dependent trends.