Three-dimensional seismic structure of the upper mantle beneath the central part of the Eurasian continent

This work is a study of the upper-mantle seismic structure beneath the central part of the Eurasian continent, including the northern Mongolia, Altai and Sayan orogenic areas and the Baikal rift zone. Seismic velocity models are reconstructed using the inverse teleseismic scheme. This scheme uses information from earthquakes located within the study area recorded by the Worldwide Network. The seismic anomaly structure is obtained for different volumes in the study area that partially overlap one another. Special attention has been paid to the reliability of the results: several noise and resolution comparisons are made.
The main results are as follows. (1) A cell structure of anomalies is observed beneath the Altai–Sayan region: positive, cold anomalies correspond to regions of recent orogenesis, negative anomalies are located beneath the depression of the Great Lakes in Mongolia and Hubsugul Lake. (2) A large negative anomaly is observed beneath the Hangai dome in Mongolia. (3) Strong velocity variations are obtained in a zone around Baikal Lake. A large negative anomaly is traced beneath the southern margin of the Siberian craton down to a depth of 700 km. Contrasting positive anomalies (4–5 per cent) are observed at a depth of 100–300 km beneath the Baikal rift. Our geodynamical interpretation of the velocity structure obtained beneath central Asia involves the existence of two processes in the mantle: thermal convection with regular cells, and a narrow plume beneath the southern border of the Siberian plate.     

Amplification of ground motion and waveform complexity in fault zones: examples from the San Andreas and Calaveras Faults

Summary. P -wave seismograms at ranges less than 10 km are synthesized by asymptotic ray theory and by summation of Gaussian beams for point sources located in a low-velocity wedge surrounding a fault. The computations are performed using models of the wedge inferred from the analysis of reflection and refraction experiments across the San Andreas and Hayward-Calaveras faults. Calculations in these models show that the 10–20Hz vertical displacements of earthquakes located at 3–10km depth are amplified by up to an order of magnitude in a 1–2km wide region centred on the fault trace compared to displacements predicted by laterally homogeneous models of the crust. This amplification is not cancelled by high attentuation in the fault zone and compensates for the reduction in amplitudes directly above the source predicted from the radiation pattern of a strike-slip earthquake. Depending on the source depth of the earthquake and the structure and velocity contrast of the wedge, multiple triplications in the travel-time curve of direct P - and S -waves will occur at stations in the fault zone. A wedge model successfully predicts the triplications observed in the P waveforms of aftershocks of the Coyote Lake earthquake recorded in the fault zone, showing that body waves from microearthquakes can be used to determine the three-dimensional velocity structure of the fault zone. The amplification, waveform complexity, and distortion of ray paths introduced by the low- velocity wedge suggest that its effects should be included in the interpretation of strong ground motions and travel times observed in the fault zone. For realistic models of the wedge, asymptotically approximate methods of calculating the body waveforms are strictly valid for frequencies greater than 20Hz. Numerical methods may be necessary to calculate accurately the wavefield at lower frequencies.     

Re-evaluation of focal depths and source mechanisms of selected earthquakes in the Afar depression

We present a stepwise inversion procedure to assess the focal depth and model earthquake source complexity of seven moderate-sized earthquakes  (6.2 > M _w > 5.1)  that occurred in the Afar depression and the surrounding region. The Afar depression is a region of highly extended and intruded lithosphere, and zones of incipient seafloor spreading. A time-domain inversion of full moment tensor was performed to model direct P and SH waves of teleseismic data. Waveform inversion of the selected events estimated focal depths in the range of 17–22 km, deeper than previously published results. This suggests that the brittle–ductile transition zone beneath parts of the Afar depression extends more than 22 km. The effect of near-source velocity structure on the moment tensor elements was also investigated and was found to respond little to the models considered. Synthetic tests indicate that the size of the estimated, non-physical, non-isotropic source component is rather sensitive to incorrect depth estimation. The dominant double couple part of the moment tensor solutions for most of the events indicates that their occurrence is mainly due to shearing. Parameters associated with source directivity (rupture velocity and azimuth) were also investigated. Re-evaluation of the analysed events shows predominantly normal faulting consistent with the relative plate motions in the region.     

Accurate hypocentre determination in the seismogenic zone of the subducting Nazca Plate in northern Chile using a combined on-/offshore network

The coupled plate interface of subduction zones—commonly called the seismogenic zone—has been recognized as the origin of fatal earthquakes. A subset of the after-shock series of the great Antofagasta thrust-type event (1995 July 30; M _w = 8.0) has been used to study the extent of the seismogenic zone in northern Chile. To achieve reliable and precise hypocentre locations we applied the concept of the minimum 1-D model, which incorporates iterative simultaneous inversion of velocity and hypocentre parameters. The minimum 1-D model is complemented by station corrections which are influenced by near-surface velocity heterogeneity and by the individual station elevations. By relocating mine blasts, which were not included in the inversion, we obtain absolute location errors of 1  km in epicentre and 2  km in focal depth. A study of the resolution parameters ALE and DSPR documents the importance of offshore stations on location accuracy for offshore events. Based on precisely determined hypo-centres we calculate a depth of 46  km for the lower limit of the seismogenic zone, which is in good agreement with previous studies for this area. For the upper limit we found a depth of 20  km. Our results of an aseismic zone between the upper limit of the seismogenic zone and the surface correlates with a detachment zone proposed by other studies; the results are also in agreement with thermal studies for the Antofagasta forearc region.     

An estimate of hypocentre location accuracy in a large network: possible implications for tectonic studies in Italy

Data recorded by the Italian Telemetered Seismic Network (ITSN) of the Istituto Nazionale di Geofisica (ING) have been widely used in recent years to image slab structures and to find evidence for active processes along the Italian Peninsula. However, the use of seismic data for geostructural purposes may be affected by the well-known trade-off between earthquake location and seismic-velocity parameters. Furthermore, the confidence ellipse predicted by standard procedures may be inadequate for the representation of the probable error of a computed localization. This paper evaluates the probable errors on the hypocentre determinations of the seismic events recorded by the ITSN, using a Monte Carlo method.
We compute synthetic arrival times using a 1-D velocity model appropriate as an average for the Italian area. The hypocentres used are all those recorded by the ITSN during the period January 1992 to March 1994 (1972 events). Station locations are those of the current ITSN configuration. The synthetic arrival times are perturbed with a Gaussian distribution of errors and input to ING's standard hypocentral location procedure, but using crustal velocities differing by 10 per cent from those used to generate them. Each simulation is repeated at least 30 times. Average absolute shifts of hypocentres are assessed in grid cells of linear dimension 33 km covering the whole Italian region.
For regions within the ITSN, shifts are typically 5–10 km in location and up to 20 km in depth. However, for offshore and coastal regions, they are much greater: 50 km or more in both location and depth (far exceeding the equivalent uncertainties quoted by ING bulletins). Possible consequences of this are highlighted by producing a cross-section of subcrustal hypocentres from the Adriatic to the Tyrrhenian Sea, where the large uncertainty in depth precludes any confident interpretation of dipping tectonic features.     

Localized velocity anomalies in the lower mantle

Summary. Two localized regions of velocity heterogeneity in the lower mantle with scale lengths of 1000–2000 km and 2 per cent velocity contrasts are detected and isolated through comparison of S, ScS, P and PcP travel times and amplitudes from deep earthquakes in Peru, Bolivia, Argentina and the Sea of Okhotsk. Comparison of the relative patterns of ScS-S differential travel times and S travel-time residuals across North American WWSSN and CSN stations for the different source regions provides baselines for interpreting which phases have anomalous times. A region of low S and P velocities is located beneath Northern Brazil and Venezuela at depths of 1700–2700 km. This region produces S -wave delays of up to 4 s for signals from deep Argentine events recorded at eastern North American stations. The localized nature of the anomaly is indicated by the narrow bounds in azimuth (15°) and take-off angle (13°) of the arrivals affected by it. The long period S -waves encountering this anomaly generally show 30–100 per cent amplitude enhancement, while the short-period amplitudes show no obvious effect. The second anomaly is a high-velocity region beneath the Caribbean originally detected by Jordan and Lynn, who used travel times from deep Peruvian events. The data from Argentine and Bolivian events presented here constrain the location of the anomaly quite well, and indicate a possible short- and long-period S -wave amplitude diminution associated with it. When the travel-time data are corrected for the estimated effects of these two anomalies, a systematic regional variation in ScS-S station residuals is apparent between stations east of and west of the Rocky Mountains. One possible explanation of this is a long wavelength lateral variation in the shear velocity structure of the lower mantle at depths greater than 2000 km beneath North America.     

The upper mantle structure of the North Sea region from Rayleigh wave dispersion

Summary. The phase velocity dispersion of fundamental mode Rayleigh waves (period range 13–127 s) is determined by the interstation method for three profiles that traverse the North Sea region of northwest Europe. The resulting observations have been combined to produce a regional phase velocity curve with 95 per cent confidence intervals, which belongs to the aseismic continental platform category of Knopoff.
Inversions of the regional phase velocity curve by the'Hedgehog'method indicate that the North Sea region is characterized by an upper mantle low-velocity zone of S -wave velocity 4.35–4.45 km/s between depths of approximately 85–200 km.     

A wide-angle seismic traverse through the Variscan of southwest Ireland

A wide-angle seismic profile across the western peninsulas of SW Ireland was performed. This region corresponds to the northernmost Variscan thrust and fold deformation. The dense set of 13 shots and 109 stations along the 120  km long profile provides a detailed velocity model of the crust.
  The seismic velocity model, obtained by forward and inverse modelling, defines a five-layer crust. A sedimentary layer, 5–8  km thick, is underlain by an upper-crustal layer of variable thickness, with a base generally at a depth of 10–12  km. Two mid-crustal layers are defined, and a lower-crustal layer below 22  km. The Moho lies at a depth of 30–32  km. A low-velocity zone, which coincides with a well-defined gravity low, is observed in the central part of the region and is modelled as a Caledonian granite which intruded upper-crustal basement. The granite may have acted as a buffer to northward-directed Variscan thrusting. The Dingle–Dungarvan Line (DDL) marks a major change in sedimentary and crustal velocity and structure. It lies immediately to the north of the velocity and gravity low, and shows thickness and velocity differences in many of the underlying crustal layers and even in the Moho. This suggests a deep, pre-Variscan control of the structural development of this area. The model is compatible with thin-skinned tectonics, which terminated at the DDL and which incorporated thrusts involving the sedimentary and upper-crustal layers.     

Small-scale heterogeneities below the Gräfenberg array, Germany from seismic wavefield fluctuations of Hindu Kush events

Small-scale elastic heterogeneities (<5  km) are found in the upper lithosphere underneath the Gräfenberg array, southeast Germany. The results are based on the analysis of broadband recordings of 17 intermediate-depth (201–272  km) events from the Hindu Kush region. The wavefront of the first P arrival and the following 40  s coda are separated into coherent and incoherent (scattered) parts in the frequency range from 0.05 to 5  Hz. The frequency-dependent intensities of the mean and fluctuation wavefields are used to describe the scattering characteristics of the lithosphere underneath the receivers. It is possible to discriminate a weak-fluctuation regime of the wavefield in the frequency range below approximately 1.5–2.5  Hz and a strong-fluctuation regime starting at 2.0–2.5  Hz and continuing to higher frequencies. In order to explain the observed wavefield fluctuations, an approach with seismic scattering at random media-type structures is proposed. The preferred model contains heterogeneities with 3–7 per cent perturbations in seismic velocity and correlation lengths of 0.6–4.8  km in the crust. This is compatible with models from active seismic experiments. Scattering in the lithospheric mantle is not required, but cannot be excluded at weak velocity contrasts (<3 per cent).     

Seismotectonics and analysis of earthquakes from the Hindukush region

Summary. Over 80 earthquakes, exclusively from the Hindukush focal region, which were recorded at the Gauribidanur seismic array (GBA) have been used in this study. These events have similar epicentral distances and a narrow azimuthal range from GBA but varying focal depths from 10 to 240 km. A fault plane dipping steeply (75°) in the north-west direction and striking N 66° E has been investigated on the basis of the spatial distribution of earthquakes in two vertical planes through 68° E and 32° N. Short period P -wave recordings up to 30 s were processed using the adaptive cross-correlation filtering technique. Slowness and azimuthal anomalies were obtained for first arrivals. These anomalies show positive as well as negative bias and are attributed to a steep velocity gradient in the upper mantle between the 400–700 km depth range where the seismic rays have their maximum penetration. Relative time residuals between the stations of GBA owe their origin very near to the surface beneath the array. A search of the signals across the array revealed that most of the events occurring at shallower depths had complex signatures as compared to the deeper events. The structure near the source region, complicated source functions and the scattering confined to the crust—upper mantle near source are mainly responsible for the complexity of the Hindukush earthquakes as the transmission zone of the ray tubes from turning point to the recording station is practically the same.     

The crustal structure beneath the northwest fjords, Iceland, from receiver functions and surface waves

Five broad-band seismic stations were operated in the northwest fjords area of Iceland from 1996 to 1998 as part of the Iceland Hotspot project. The structures of the upper 35  km or so beneath these stations were determined by the modelling and joint inversion of receiver functions and regional surface wave phase velocities. More than 40 teleseismic events and a few regional events containing high-quality surface wave trains were used. Although the middle period passband of the seismograms is corrupted by oceanic microseismic noise, which hinders the interpretation of structural details, the inversions reveal the overall features. Many profiles obtained exhibit large velocity gradients in the upper 5  km or so, smaller zero gradients below this, and, at ~23  km depth, a zone 2–4  km thick with higher velocity gradients. The two shallower intervals are fairly consistent with the 'upper' and 'lower' crust, defined by Flovenz (1980 ). The deep zone of enhanced velocity gradient seems to correspond to the sharp reflector first reported by Bjarnason et al . (1993 ) and identified by them as the 'Moho'. However, this type of structure is not ubiquitous beneath the northwest fjords area. The distinctiveness of the three intervals is variable, and in some cases a structure with velocity gradient increasing smoothly with depth is observed. We term these two end-members structures of the first and second types respectively. Structures of the second type correlate with older areas. Substantial variation in fundamental structure is to be expected in Iceland because of the great geological heterogeneity there.     

Three-dimensional seismic structure beneath the Australasian region from refracted wave observations

The earthquakes in the seismicity belt extending through Indonesia, New Guinea, Vanuatu and Fiji to the Tonga–Kermadec subduction zone recorded at the 65 portable broad-band stations deployed during the Skippy experiment from 1993–1996 provide good coverage of the lithosphere and mantle under the Australian continent, Coral Sea and Tasman Sea.
The variation in structure in the upper part of the mantle is characterized by deter-mining a suite of 1-D structures from stacked record sections utilizing clear P and S arrivals, prepared for all propagation paths lying within a 10° azimuth band. The azimuth of these bands is rotated by 20° steps with four parallel corridors for each azimuth. This gives 26 separate azimuthal corridors for which 15 independent 1-D seismic velocity structures have been derived, which show significant variation in P and S structure.
The set of 1-D structures is combined to produce a 3-D representation by projecting the velocity values along the ray path using a turning point approximation and stacking into 3-D cells (5° by 50 km in depth). Even though this procedure will tend to underestimate wave-speed perturbations, S -velocity deviations from the ak135 reference model exceed 6 per cent in the lithosphere.
In the uppermost mantle the results display complex features and very high S -wave speeds beneath the Precambrian shields with a significant low-velocity zone beneath. High velocities are also found towards the base of the transition zone, with high S -wave speeds beneath the continent and high P -wave speeds beneath the ocean. The wave-speed patterns agree well with independent surface wave studies and delay time tomography studies in the zones of common coverage.     

Massif Central (France): new constraints on the geodynamical evolution from teleseismic tomography

The Massif Central, the most significant geomorphological unit of the Hercynian belt in France, is characterized by graben structures which are part of the European Cenozoic Rift System (ECRIS) and also by distinct volcanic episodes, the most recent dated at 20 Ma to 4000 years BP. In order to study the lithosphere-asthenosphere system beneath this volcanic area, we performed a teleseismic field experiment.
During a six-month period, a joint French-German team operated a network of 79 mobile short-period seismic stations in addition to the 14 permanent stations. Inversion of P -wave traveltime residuals of teleseismic events recorded by this dense array yielded a detailed image of the 3-D velocity structure beneath the Massif Central down to 180 km depth. The upper 60 km of the lithosphere displays strong lateral heterogeneities and shows a remarkable correlation between the volcanic provinces and the negative velocity perturbations. The 3-D model reveals two channels of low velocities, interpreted as the remaining thermal signature of magma ascent following large lithospheric fractures inherited from Hercynian time and reactivated during Oligocene times. The teleseismic inversion model yields no indication of a low-velocity zone in the mantle associated with the graben structures proper. The observation of smaller velocity perturbations and a change in the shape of the velocity pattern in the 60–100 km depth range indicates a smooth transition from the lithosphere to the asthenosphere, thus giving an idea of the lithosphere thickness. A broad volume of low velocities having a diameter of about 200 km from 100 km depth to the bottom of the model is present beneath the Massif Central. This body is likely to be the source responsible for the volcanism. It could be interpreted as the top of a plume-type structure which is now in its cooling phase.     

A test of the precursory accelerating moment release model on some recent New Zealand earthquakes

The proposal that the moment release rate increases in a systematic way in a large region around a forthcoming large earthquake is tested using three recent, large New Zealand events. The three events, 1993–1995, magnitudes 6.7–7.0, occurred in varied tectonic settings. For all three events, a circular precursory region can be found such that the moment release rate of the included seismicity is modelled significantly better by the proposed accelerating model than by a linear moment release model, although in one case the result is dubious. The 'best' such regions have radii from 122 to 167 km, roughly in accord with previous observations world-wide, but are offset by 50–60 km from the associated main shock epicentre. A grid-search procedure is used to test whether these three earthquakes could have been forecast using the accelerating moment release model. For two of the earthquakes the result is positive in terms of location, but the main shock times are only loosely constrained.     

State of stress in the Southern Tyrrhenian subduction zone from fault-plane solutions

In this paper we present revised locations and original focal mechanisms computed for intermediate and deep earthquakes that occurred within the Southern Tyrrhenian subduction zone between 1988 and 1994, in order to improve our knowledge of the state of stress for this compressional margin. In particular, we define the stress distribution within a large portion of the descending slab, between 40 and about 450 km depth. The seismicity distribution reveals a continuous 40–50 km thick slab that abruptly increases its dip from subhorizontal in the Ionian Sea to a constant 70° dip in the Tyrrhenian. We computed focal mechanisms for events with magnitudes ranging from 2.7 and 5.7, obtaining the distribution of P - and T -axes for many events for which centroid moment tensor (CMT) solutions are not available, thus enabling the sampling of a larger depth range compared to previous studies. We define three portions of the slab characterized by different distributions of P - and T -axes. A general down-dip compression is found between 165 and 370 km depth, whereas in the upper part of the slab (40–165 km depth) the fault-plane solutions are strongly heterogeneous. Below 370 km the P -axes of the few deep events located further to the north have a shallower dip and are not aligned with the 70° dipping slab, possibly suggesting that they belong to a separated piece of subducted lithosphere. There is a good correspondence between the depth range in which the P -axes plunge closer to the slab dip (∼ 70°) and the interval characterized by the highest seismic energy release (190–370 km).     

Determination of Lg-wave attenuation using single-station seismograms: a case study for Zimbabwe

In this paper we show how the quality factor Q may be calculated using a single seismograph station and a number of events recorded on analogue seismograms. We followed Nuttli's (1973) method and extended it to one seismograph station. Using the single station Bulawayo (BUL), we determined a mean Q value of 650 for Zimbabwe. Furthermore, we considered different propagation paths over Zimbabwe as the seismic waves travelled to BUL and found a low Q value of 350 for the Deka fault zone. the Q value of 650 obtained in this study agrees well within error with that of 603 reported by Chow et al. (1980) using the multistation, multi-event method, and lies within the range of Q values (454–759) reported by Xie & Mitchell (1990) using a 'back-project'method to image large-scale lateral variations of Lg coda Q . the results obtained are important as the Q value constitutes part of the input data in seismic hazard calculations. the method may be used to determine Q in regions where there may be only one station with reliable analogue seismogram data.     

Upper-mantle structure of the Baltic Shield below the Swedish National Seismological Network (SNSN) resolved by teleseismic tomography

Upper-mantle structure under the Baltic Shield is studied using non-linear high resolution teleseismic P -phase tomography. Observed relative arrival-time residuals from 52 teleseismic earthquakes recorded by the Swedish National Seismological Network (SNSN) are inverted to delineate the structure of the upper mantle. The network consists of 47 (currently working) three-component broad-band stations located in an area about 450 km wide and 1450 km long. In order to reduce complications due to possible significant three-dimensionality of Earth structure, events chosen for this study lay close to in-line with the long-axis of the array  (±30°)  . Results indicate P -wave velocity perturbations of ±3 per cent down to at least 470 km below the network. The size of the array allows inversion for structures even at greater depths, and lateral variations of velocity at depths of up to 680 km appear to be resolved. Below the central part of the array (60°–64° N), where ray coverage is best, the data reveals a large region of relatively low velocity at depths of over about 300 km. At depths less than about 250–300 km, the models include a number of features, including an apparent slab-like structure dipping gently towards the north.     

Shear wave anisotropy in the crust around the San Andreas fault near Parkfield: spatial and temporal analysis

We systematically analysed shear wave splitting (SWS) for seismic data observed at a temporary array and two permanent networks around the San Andreas Fault (SAF) Observatory at Depth. The purpose was to investigate the spatial distribution of crustal shear wave anisotropy around the SAF in this segment and its temporal behaviour in relation to the occurrence of the 2004 Parkfield M 6.0 earthquake. The dense coverage of the networks, the accurate locations of earthquakes and the high-resolution velocity model provide a unique opportunity to investigate anisotropy in detail around the SAF zone. The results show that the primary fast polarization directions (PDs) in the region including the SAF zone and the northeast side of the fault are NW–SE, nearly parallel or subparallel to the SAF strike. Some measurements on the southwest side of the fault are oriented to the NNE–SSW direction, approximately parallel to the direction of local maximum horizontal compressive stress. There are also a few areas in which the observed fast PDs do not fit into this general pattern. The strong spatial variations in both the measured fast PDs and time delays reveal the extreme complexity of shear wave anisotropy in the area. The top 2–3 km of the crust appears to contribute the most to the observed time delays; however substantial anisotropy could extend to as deep as 7–8 km in the region. The average time delay in the region is about 0.06 s. We also analysed temporal patterns of SWS parameters in a nearly 4-yr period around the 2004 Parkfield main shock based on similar events. The results show that there are no appreciable precursory, coseismic, or post-seismic temporal changes of SWS in a region near the rupture of an M 6.0 earthquake, about 15 km away from its epicentre.     

Upper mantle structure beneath the Mid-Atlantic Ridge north of the Azores based on observations of compressional wave

Summary. Four seismic refraction profiles have been interpreted which serve to indicate the structure of the lithosphere near the Mid-Atlantic Ridge close to the Azores. An east–west profile which crosses the ridge axis yields a crustal structure. Although energy is propagated across the ridge axis within the crust the axial region marks a clear barrier to propagation within the mantle. A profile parallel to the axis (4 my isochron) shows, below a 7.6 km/s layer, a low-velocity zone underlain by an 8.3 km/s refractor 9 km below the sea bed. On profies normal to the ridge axis higher velocities, which are observed on lines shot towards the ridge, can be attributed to this refractor if it has a dip of several degrees away from the ridge. On another profile parallel to the axis (9 my isochron) a velocity of about 8.3 km/s is only found to exist much deeper at about 30 km depth. These observations are interpreted in the light of seismic refraction results recently obtained by Lewis & Snydsman and of quantitative petrological models, such as that of Bottinga & Allègre. A velocity model based on Bottinga & Allègre's model allows us to understand our results qualitatively. In particular the two 8.3 km/s refractors at 9 and 30 km depth correspond to two different residual peridotite layers. The upper layer contains 1.5–2 wt per cent water and as the lithosphere moves away from the ridge axis the temperature in this layer becomes low enough to start hydration reactions. These cause the low-velocity zone observed at 4 my and the total disappearance of the shallow level refractor before 9 my.     

A transformational approach to geophysical inverse problems

A set of coordinate transformations is used to linearize a general geophysical inverse problem. Statistical and analytic techniques are employed to estimate the parameters of such linearization transformations. In the transformed space, techniques from linear inverse theory may be utilized. Consequently, important concepts, such as model parameter covariance, model parameter resolution and averaging kernels, may be carried over to non-linear inverse problems. I apply the approach to a set of seismic cross-borehole traveltimes gathered at the Conoco Borehole Test Facility. the seismic survey was conducted within the Fort Riley formation, a limestone with thin interbedded shales. Between the boreholes, the velocity structure of the Fort Riley formation consists of a high-velocity region overlying a section of lower velocity. It is found that model parameter resolution is poorest and spatial averaging lengths are greatest in the underlying low-velocity region.