Yellowknife Seismic Array Research Articles

Seismic Evaluation of a Multipurpose Tower in Yellowknife, Northwest Territories, Canada

Locating scientific instruments in remote regions of the planet poses many common problems beyond just logistics, in particular when these instruments are to function autonomously or semiautonomously. In seismology, this is particularly prevalent for seismic stations located in the polar regions of the planet. Extremely cold, but also variable temperature, wind, and sunlight conditions as well as general inaccessibility hamper both operation and maintenance of equipment in such regions (Gill, 1974; Holcomb, 1982). The search for a reliable source of electrical power for the continuous operation of sensors and digitizing equipment continues to be one of these concerns. Many technologies have been investigated including radioisotope thermoelectric generation (e.g., Lamp, 1994), thermal electric generators (TEGs) (e.g., Holcomb, 1982), diesel engines, chemical batteries, wind and solar power generation (Anandakrishnan et al. , 2000), and more recently, fuel‐cell technology (e.g., Stehle et al. , 2011). Each technology has its own pros and cons, and in general, the technology that is chosen is based on which best suits the application, its requirements, and stresses of the station and its operators. For seismic instruments, this often requires that any noise generated by this choice does not adversely affect the operation or observations of ground motion the instrument is trying to measure. Such is the case at the Yellowknife teleseismic array in Canada’s Northwest Territories. Commissioned in 1962, the role of the Yellowknife seismic array (YKA) was to investigate the capabilities of distributed arrays of seismometers to monitor for the tell‐tale ground motions from explosions of nuclear weapons testing by the then nuclear‐capable countries of the United States, Great Britain, France, and the former Soviet Union. In one of the eventually five similar arrays constructed by the United Kingdom Atomic Energy Authority (UKAEA), the research was to determine if small numbers of seismic arrays could be …

Composition and variation of noise recorded at the Yellowknife Seismic Array, 1991–2007

We analyze seismic noise recorded on the 18 short‐period, vertical component seismometers of the Yellowknife Seismic Array (YKA). YKA has an aperture of 23 km and is sited on cratonic lithosphere in an area with low cultural noise. These properties make it ideal for studying natural seismic noise at periods of 1–3 s. We calculated frequency‐wave number spectra in this band for over 6,000 time windows that were extracted once per day for 17 years (1991–2007). Slowness analysis reveals a rich variety of seismic phases originating from distinct source regions: Rg waves from the Great Slave Lake; Lg waves from the Atlantic, Pacific, and Arctic Oceans; and teleseismic P waves from the north Pacific and equatorial mid‐Atlantic regions. The surface wave energy is generated along coastlines, while the body wave energy is generated at least in part in deep‐water, pelagic regions. Surface waves tend to dominate at the longer periods and, just as in earthquake seismograms, Lg is the most prominent arrival. Although the periods we study are slightly shorter than the classic double‐frequency microseismic band of 4–10 s, the noise at YKA has clear seasonal behavior that is consistent with the ocean wave climate in the Northern Hemisphere. The temporal variation of most of the noise sources can be well fit using just two Fourier components: yearly and biyearly terms that combine to give a fast rise in microseismic power from mid‐June through mid‐October, followed by a gradual decline. The exception is the Rg energy from the Great Slave Lake, which shows a sharp drop in noise power over a 2‐week period in November as the lake freezes. The Lg noise from the east has a small but statistically significant positive slope, perhaps implying increased ocean wave activity in the North Atlantic over the last 17 years.

Detection of a ULVZ at the base of the mantle beneath the northwest Pacific

We used the Yellowknife seismic array (YKA) to measure the slowness of 1,371 P and Pdiff waves from earthquakes occurring in the circum‐Pacific region. The corresponding anomalies in P‐velocity show a sharp reduction of up to 6% across a patch of the lowermost mantle beneath the Northwest Pacific with lateral dimensions of several hundred kilometers. The location of this ultra low velocity zone (ULVZ) correlates with a long‐wavelength compositional boundary revealed by probabilistic mantle tomography. We interpret the ULVZ as partial melt created by paleo‐slab material that is being swept laterally from northwestern Pacific subduction zones towards the large, chemically distinct province beneath the south‐central Pacific.

Earthquake in southern Sweden wakes up Denmark on 16 December 2008

A moderately strong earthquake struck southern Sweden 5 km south-west of the town of Sjöbo, 60 km east of Malmö, in the early morning at 6:20 a.m. local time on 16 December 2008. The epicentre was located in Skåne, a region that is known for its extremely low seismicity, and its location was determined to be 55.5°N and 13.6°E with an uncertainty of about 6 km. A depth of 9 km with an uncertainty of 3 km was obtained from teleseismic observations at the Yellowknife seismic array, USA. Since waveform data from the Swedish national seismic network are not yet available, depth estimation using local stations has so far not been attempted. During the period 1970–2008, only three small earthquakes were detected in the region; the largest measured 2.8 on the local Richter scale. To our knowledge none of these previous earthquakes were felt by people. The historical archives dating back to 1375 show that 14 other earthquakes have been felt in the area. The largest of these, recorded in 1894, was felt over an area of 7300 km2 and had an epicentre 50 km east of the 16 December 2008 earthquake (Scandinavian Earthquake Archive 2003). The activity in southern Sweden is similar to that of northern Sjælland and north-western Jylland, and confirms the low seismicity of the region (Gregersen et al. 1991).

Locating scatterers in the mantle using array analysis of PKP precursors from an earthquake doublet

Well separated individual PKP precursors observed at the Yellowknife seismic array (YK) for a high quality doublet of earthquakes provide an opportunity to study the location of the corresponding scatterers and assess the stability of the location estimation. Based on the comparison of the waveforms of stacked individual precursors and those of PKIKP phases, we are able to determine that most of these precursors originate from scattering of the PKPbc (rather than the PKPab) branch above the B caustic on the receiver side. This allows a reliable location of the scatterers in the lower mantle. Their depths range from 2890 km (the CMB) to 2270 km, scattering angles range from 45.8° to 16.0°, and surface projections range from southern Ontario to northern Saskatchewan in Canada. These locations are associated with transitions from slow to fast velocities in mantle tomographic models and follow the expected general dip direction of fossil slabs under north America. This suggests that the subducted slab remnants under north America have retained their compositional signature. The fact that we can essentially treat these scatterers as reflections from plane boundaries suggests that the remnant fragments of slab may be spatially extended, which should be confirmed using broadband data. Average differences in measured slowness and back-azimuth of the doublet precursors are as small as 0.08 s/deg and 1.4 deg, respectively. Our study indicates that it may be possible to locate such scatterers using single earthquakes and small aperture seismic arrays.

Short wavelength topography on the inner-core boundary

Constraining the topography of the inner-core boundary is important for studies of core-mantle coupling and the generation of the geodynamo. We present evidence for significant temporal variability in the amplitude of the inner core reflected phase PKiKP for an exceptionally high-quality earthquake doublet, observed postcritically at the short-period Yellowknife seismic array (YK), which occurred in the South Sandwich Islands within a 10-year interval (1993/2003). This observation, complemented by data from several other doublets, indicates the presence of topography at the inner-core boundary, with a horizontal wavelength on the order of 10 km. Such topography could be sustained by small-scale convection at the top of the inner core and is compatible with a rate of super rotation of the inner core of approximately 0.1-0.15 degrees per year. In the absence of inner-core rotation, decadal scale temporal changes in the inner-core boundary topography would provide an upper bound on the viscosity at the top of the inner core.

Application of Method of Spectral Component Ratio of Wavelet-Packets to Discrimination between Earthquakes and Nuclear Explosions

The spectral analysis method is widely used in discrimination between nuclear explosions and earthquakes. As a stationary method, it has low resolution due to its invariable time-frequency window, even if the Gabor transform is adopted. To improve its resolution, a wavelet transform is applied to the analysis of earthquakes and nuclear explosions, which were recorded at the Urumqi seismic station in the Xinjiang Autonomous region. And a criterion of spectral component ratio of wavelet-packets (SCRWP), which is used to identify the earthquakes and nuclear explosions, is proposed. Analysis of data of the India nuclear test recorded by the Yellow Knife Seismic Array in Canada has also confirmed that the SCRWP method is rather effective in discrimination between nuclear explosions and earthquakes. The result shows that the SCRWP value is generally higher than 1.0 for earthquakes, but usually less than 1.0 for nuclear explosions.

A reconnaissance teleseismic study of the upper mantle and transition zone beneath the Archean Slave craton in NW Canada

The objective of this study is to investigate upper mantle structure below the Archean Slave craton, site of the oldest known rocks on Earth and numerous diamondiferous kimberlites, and thence to gain an understanding of early craton formation and kimberlite genesis. To this aim, a temporary array, consisting of 13 sites equipped with broadband seismometers, recorded teleseisms between November 1996 and May 1998. This data set has been augmented with recordings from the Yellowknife seismic array. Our three most robust observations and their interpretation are: (1) P-wave travel-time tomography reveals the oldest part of the craton, the Central Slave Basement Complex, to be underlain by the fastest seismic velocities, indicating that this block remains distinct well into mantle depths; (2) receiver function analysis requires only the Moho as major S-wave velocity discontinuity and points to a fairly constant crustal thickness throughout the Slave province; and (3) SKS splitting analysis shows little variation in delay times and fast polarization directions across the array, leading us to conclude that the present-day plate motion of North America is the primary cause for mantle fabric beneath the entire array. Furthermore, the data let us rule out the possibility that the Mackenzie plume had any seismologically detectable effect on the Slave lithosphere. More speculative results of our investigation, namely a possible genetic link between a low seismic velocity anomaly at depth with the overlying Lac de Gras kimberlite field, and a possible Archean origin of two shallow low-velocity anomalies, will require further investigation.

Seismic waves converted from velocity gradient anomalies in the Earth’s upper mantle

Summary Modelling of elastic wave propagation in 1-D structures is frequently performed using reflectivity techniques in which the Earth’s velocity profile is approximated by stacks of homogeneous layers. The complete reflection/transmission (R/T) response of a zone with arbitrary 1-D depth variation (including both gradients and discontinuities in material properties) can, however, be calculated using invariant embedding techniques. Results from earlier studies are here extended to derive exact expressions for R/T matrices in arbitrary, 1-D anisotropic media using a form of Born approxi-mation valid for thin scatterers and which does not assume small perturbations in material properties. The R/T matrices are solutions to a system of non-linear, ordinary differential equations of Ricatti type and may be manipulated using standard R/T matrix algebra. In an equivalent description, the wavefield within the heterogeneous zone is considered in terms of depth-dependent contributions from up- and downgoing waves propagating within the embedding reference medium. This leads to efficient calculation of the internal wavefield using R/T matrices of the heterogeneous stratification and portions thereof at minor additional expense. Mode conversion of teleseismic P and S phases from velocity gradients is examined by way of examples and comparison with three-component data from broad-band stations of the Yellowknife seismic array. The frequency dependence of such wave interactions depends on the differences in vertical slowness between incident and scattered modes. It is shown that significant energy is converted from transition zones with extent L<λP/2, a broader interval than will generally produce intramode reflections. A layer structure identified from Ps conversions near 75 km depth below the Slave craton is shown to be compatible with a ~ 10 km thick gradient zone in which anisotropy increases from ambient levels to δVp= ± 5 per cent, δVs= ± 2.5 per cent at a discontinuous upper boundary. This characterization supports a previous interpretation as the upper strata of a former oceanic plate juxtaposed against overriding lithosphere during an ancient episode of shallow subduction.

Teleseismic P wave attenuation and nuclear explosion source functions inferred from Yellowknife Array data

We report here the results of a comprehensive seismic attenuation investigation along the paths connecting Canada's Yellowknife seismic array (YKA) with seven active nuclear explosion testing areas. The data consist of more than 600 explosion‐generated teleseismic P wave records. A dual time‐frequency averaging technique is used to take advantage of the array recording characteristics without the drawback of the conventional beam‐forming, excessive annihilation of high‐frequency signal energies. The dual averaging technique, deployed in conjunction with a multiwindow spectral analysis method, yields smooth amplitude spectra whose falloff at high frequencies suffers little from spectral leakage due to the familiar presence of a prominent low‐frequency plateau. Measured in terms of t*, the highest attenuation (0.66 s) is found along the path which originates from the Tuamotu test area; somewhat less attenuating are the two paths which depart from the Pahute Mesa (0.59 s) and Yucca Flat (0.50 s) nuclear test areas, both located within the U.S. Nevada Test Site. We find t* for these three paths to be substantially (up to 0.21 s) higher than recently published estimates (e.g., Der et al., 1985). We attribute these disparities largely to differences in spectral leakage control capability between the conventional single window and the improved multiwindow spectral analysis methods. The least attenuating paths all originate from the Soviet test areas: Novaya Zemlya (NZ), west Kazakhstan, Degelen Mountain (DM), and Shagan River (SR). The last two of these test areas, DM and SR, are both located in east Kazakhstan. The P wave signatures of the Soviet explosions are rich in high‐frequency (&gt;4.5 Hz) energies, and the YKA data (0.5–8.0 Hz) support a frequency‐dependent t* whose value at high frequencies (&gt;4.5 Hz) is as small as 0.17 s. To gain a grasp of the ramifications of the t* disparity between the multiple‐window and the single‐window results, we have compared explosion source time functions obtained by the multichannel deconvolution technique of Shumway and Der (1985) in order to assess their sensitivity to the input t* value. In our example involving the deconvolved source functions of five French Tuamotu explosions, we find that a 0.1‐s t* difference is large enough to cause clearly discernible signature differences, in terms of the signal frequency content as well as the extractability of a secondary arrival some 0.4 s behind the first P arrival. This secondary arrival is believed to be the depth phase pP, a seismic signature of importance in both yield estimation and earthquake/explosion source discrimination. The absorption band modeling (Minster, 1978a, b) of the French Tuamotu explosion data yields 1.08±0.05 and 0.079±0.008 s for t*0 and τm, respectively. The corresponding parameter estimates derived from the U.S. explosion data are somewhat smaller. The t*0 and τm estimates are the smallest along the paths which depart from the four Soviet test areas. For the NZ‐YKA path the t*0 and τm estimates are 0.56±0.08 and 0.061±0.013 s, respectively. Plagued by a strong trade‐off between the two model parameters, these estimates are not tightly constrained, however.

Meteoroid sonic shock-wave-generated seismic signals observed at a seismic array

At approximately 04:20 local time (10:20 UT) on 19 September 1986, a local resident observed a meteor in the sky above Yellow-knife, North-west Territory, Canada. We subsequently found seismic signals, recorded at the Yellowknife seismic array in Northwest Territory, Canada, that coincided with the visual observation of the meteor in the sky overhead. The signals cannot be explained by any earthquake, explosion, or impact-generating mechanism. They are shown to be consistent with their having been generated by the shock wave created by the associated meteoroid as it passed through the atmosphere overhead. The meteoroid's trajectory and velocity, as inferred from the array records, is consistent both with the observed direction of travel of the meteoroid and independent seismic observations recorded at two single stations in the Northern Canadian seismic network.

Regional differences in upper mantle heterogeneity fron coherency measurements on array data

When a ray-path passes through an homogeneous earth, the energy arriving at a seismic array should be relatively coherent with the signal complexity being dependent mainly on the source function. When the signal, however, encounters both lateral and vertical inhomogeneities along its path, much of its energy will be scattered such that the energy arriving at the array will be relatively incoherent. The signal complexity should therefore depend not only on the source function but also on the scattering characteristics of the ray-path. Measurements of the relative amounts of coherent and incoherent energy arriving at an array for different travel paths should therefore reveal regional differences in upper mantle heterogeneity. Approximately 300 earthquake recordings at the Gauribidanur medium aperture seismic array (GBA) in southern India (distance 15°–36°, azimuth 0°–360°): and over 65 earthquake recordings at the Yellowknife seismic array (YKA) in western Canada (distance 12°–30°, azimuth 180°–285°) were selected for this study. The analyses showed that there is relatively more heterogeneity for the continental upper mantle region to the north of GBA than for the corresponding oceanic structure to the south-west and southeast, with the structure above 400 km tending to be more complex than the structure below that depth. The YKA analyses showed that, in contrast to the GBA results, the upper mantle structure under the north-western part of the North American continent is progressively more heterogeneous with increasing depth between 400 and 750 km.

The identification and interpretation of upper mantle travel-time branches from measurements of dT/dΔ made on data recorded at the Yellowknife seismic array

There is broad agreement among various seismological studies that the upper mantle has two regions where very high positive velocity gradients or transition zones exist. In most cases, the presence of these zones implies that two major triplications are likely to exist in the body-wave travel-time curve for distances less than 30°. Because of the difficulties in observing and identifying later arrivals belonging to the various travel-time branches, the inversion of the seismic data is often very difficult. In this paper an adaptive processing technique was employed to examine the variations in slowness that occur along the first 36 s of the short-period P-wave trains recorded at the Yellowknife medium aperture seismic array. Over 100 earthquakes from the Alaska Peninsula and California regions were selected. From the California results we were able to clearly observe the 12–13 s/deg slowness branch as a later arrival out to distances as great as 26°. Other later arrival branches as well as cusps associated with the 400 and 650 km discontinuities were not well defined even though the cross-over point as determined from slowness measurements on first arrivals were clearly located. An inversion of the data showed that the '650 km' transition zone occurred at a much shallower depth west of the array compared to the corresponding region to the south.

Reduced false alarm rates in seismic array detection by non‐linear beamforming

It has been suggested that a non‐linear, multi‐channel summation process increases the signal‐to‐noise ratio of small coherent signals. Using the data base of the International Seismic Month, it is shown conclusively that a logarithmic beamforming process increases the detection probability for seismic events for a specified false alarm rate by at least a factor of two. A detector based on this process is working on‐line at the Yellowknife seismic array.

Seismology, dT/d and Deep Mantle Convection

Summary dT/dA and azimuth measurements at the Yellowknife Seismic Array (YKA) for P and pP phases that have penetrated to depths greater than 1850 km show systematic bias relative to those values calculated from spherically symmetric Earth models. Moreover, the corresponding arrival vectors for the phases PcP, ScP, PnKP and SnKP (n > 2) approaching YKA from within the same narrow azimuth ranges as the P andpP phases show related systematic errors. From an analysis of combinations of all these phases, we show that most of the observed bias for P and pP is produced along the upgoing portion of the ray path, and that any heterogeneities and velocity anomalies near the base of the mantle cannot be clearly elucidated by a dT/dA study of the P phase alone. P wave data must be supplemented by other phases that traverse similar paths within the upper mantle beneath arrays. Evidence for regional variations in structure at depths greater than 1850 km is presented. An abrupt drop in dT/dA between distances of 87 and 90 is reported for two different regions of the Earth and possibly a third; this drop represents a fairly rapid increase in velocity gradients at about 2700 km depth that may be a world-wide phenomenon. There are also strong lateral variations in the upper mantle to the north-east and east of YKA. Array observations of precursors to PP, 1'' P' and PKP indicate small-scale lateral variations in the crust and upper mantle and in the lowest 200 km of the mantle that give rise to extensive scattered P wave energy. We initially try to relate our data on lateral variations to regional variations in travel times and mantle convection. Our main concern is the problem of motion within the lower mantle, and how dT/dA measurements in combination with theoretical studies of the properties of the scattering regions might indicate flow. The combined P wave data on small-scale and large-scale lateral variations suggest that the mantle is most homogeneous between depths of 800 km and 2500 km; consequently, we propose that mantle convection is confined to the upper mantle and the lowest 400 km of the mantle, and that convection plumes, if they exist, do not originate at depths greater than 800 km.

Anomalous azimuths of P: evidence for lateral variations in the deep mantle

Certain anomalies of dT/dΔ and azimuth of P-waves arriving at the Yellowknife seismic array have long defied explanation by near-array structure. Similarly, the anomalies are not easily explained by source structure. A study of 43 Aleutian events now leads to a satisfactory explanation in terms of lateral variations of velocities near the vertex of the ray path near 800 km depth, which can readily be accommodated within the constraints of the known mantle structure.

Detection capabilities of the Yellowknife seismic array and regional seismicity

abstract Detection statistics from a search of 99 days of Yellowknife Array data have been obtained. The mean 50 and 90 per cent cumulative detection levels are found to be m 3.9 and m 4.2 for the third zone from Yellowknife. Regional variations in the detection levels have been found. The 50 per cent cumulative detection level ranges from m 3.8 for Asia to m 4.3 for the Aleutians. Regional values of the slopes of the recurrence curves for log N versus magnitude have been obtained, the mean slope for the third zone from Yellowknife being −0.99.

Calibration of the Yellowknife Seismic Array with First Zone Explosions

Summary Recordings from a crustal seismic experiment, which was conducted in the Yellowknife area in 1966, were used for calibration of the Yellowknife seismic array. In the immediate vicinity of the array the crust is found to be very uniform. A superficial layer with an intercept time of 0.172+0.012 s and unknown velocity is underlain by a crust with a P wave velocity of 6-04+0-01 km s-' near the top: assuming this velocity constant throughout the second layer, the total thickness of the crust is about 34+2 km. The Mohorovicic discontinuity is horizontal under the array within the resolution of this experiment and the apparent P, velocity is 8-15 km s-'. At a distance of a few tens of kilometres the crustal uniformity breaks down. The distances are such that, for most teleseismic signals, the effect of these inhomogeneities should be negligible.

Epicentral uncertainties and detection probabilities from the Yellowknife seismic array data

Abstract Analog recording tapes from the Yellowknife seismic array have been processed diqitally in Canada for over a year, with concentration on the automatic detection and epicenter location of events between 26° and 90° distance, using short period P-wave arrivals. For the purpose of detection, signals from the individual seismometers in the two arms of the cross array are analog band-pass filtered, digitized at 20 samples/s, multiplexed into a digital computer, velocity and azimuth filtered and correlated, using an exponentially weighted integration over time with an equivalent width of 1.6 s. The correlograms for up to 168 phased beams are scanned for values exceeding a preset trigger level and an event is recorded when the level is passed consistently a number of times. In late 1966, during a seismically quiet period and with the array fully operational, the 50 per cent automatic detection level achieved by this method for events in the Third Zone to Yellowknife was m4.0 ± 0.1, slightly better than the level of an analog trigger operated at the station which uses the correlogram method for a single unphased beam only. The 50 per cent detection level of the Yellowknife standard station is about m4.4 and thus the array-computer automatic detection method gives about Δm0.4 improvement, which is expected from the processing method used if the noise is largely uncorrelated. No significant variations in the detection level with azimuth have yet been observed. Approximate epicenter locations are determined from the best apparent arrival vector. The best vector is assumed to be given by the maxima of parabolas at constant azimuth and wave number interpolated between the highest values of the correlograms. USCGS P.D.E. information is used in conjunction with the J-B tables to calculate an expected apparent arrival vector. The difference between the expected and best interpolated arrival vectors has an average deviation of about 6 ms/km. Their distribution does not suggest a simple crustal or upper mantle cause under the array station.

Discrimination of Temporally Overlapping Seismic Events

ROUTINE digital processing of seismic events recorded on analogue magnetic tape at the Yellowknife seismic array, Canada, is being carried out in Ottawa with the aid of a computer. The array is of the UKAEA pattern and has been described by Manchee and Somers1. Processing methods have been described by Weichert et al.2. Although no attempt is being made to process every event that is received, a large enough sample is selected from each month's recording to enable a library of third zone events to be accumulated on digital tape. Sometimes, as in the present instance, a particular event is of more than passing interest.