Characterization of a heterogeneous landfill usingseismic and electrical resistivity data
Laura Amalia Konstantaki1, Ranajit Ghose2, Deyan Draganov2,Giovanni Diaferia3, and Timo Heimovaara1
ABSTRACT
Understanding the processes occurring inside a landfill is im-portant for improving the treatment of landfills. Irrigation andrecirculation of leachate are widely used in landfill treatments.Increasing the efficiency of such treatments requires a detailedunderstanding of the flow inside the landfill. The flow dependslargely on the heterogeneous distribution of density. It is, there-fore, of great practical interest to determine the density distri-bution affecting the flow paths inside a landfill. Studies inthe past have characterized landfill sites but have not led tohigh-resolution, detailed quantitative results. We performedan S-wave reflection survey, multichannel analysis of surfacewaves (MASW), and electrical resistivity survey to investigatethe possibility of delineating the heterogeneity distribution in
the body of a landfill. We found that the high-resolution S-wavereflection method offers the desired resolution. However, in thecase of a very heterogeneous landfill and a high noise level, theprocessing of high-resolution, shallow reflection data requiredspecial care. In comparison, MASW gave the general trend ofthe changes inside the landfill, whereas the electrical resistivity(ER) survey provides useful clues for interpretation of seismicreflection data. We found that it is possible to localize fine-scaleheterogeneities in the landfill using the S-wave reflectionmethod using a high-frequency vibratory source. Using empiri-cal relations specific to landfill sites, we then estimated the den-sity distribution inside the landfill, along with the associateduncertainty considering different methods. The final interpreta-tion was guided by supplementary information provided byMASW and ER tomography.
INTRODUCTION
Sustainable aftercare of sanitary landfills is a serious concern. Toreduce the aftercare period with specific treatment technologies, op-erators and researchers try to understand the processes occurringinside a landfill (Scharff, 2005; Van Vossen, 2010). There is an in-creasing need for understanding the heterogeneity inside the land-fills for optimizing the treatment technologies, especially regardingthe recirculation of leachate (Powrie and Beaven, 1999). Preferen-tial flow paths depend highly on the heterogeneous distribution ofdensity. It is critical to understand the flow of leachate and to modelthe behavior of the processes inside a landfill. There are models thatexplain the hydrobiomechanical behavior of landfills; however, they
lack detailed quantitative information of density distribution, whichcan greatly improve the accuracy of model predictions (White et al.,2004; McDougall and Fleming, 2013). Stoltz et al. (2012) illustratethat it is important to have a knowledge of the density distributionfor the estimation of the moisture-retention properties.In the past, geophysical methods have been used to image the
body of a landfill, but they have faced difficulties due to uncertain-ties, artifacts, and noise. Reflection and refraction seismic studies(Green et al., 1999) suffer from uncertainties mainly due to strongscattering events. Carpenter et al. (2013) image a landfill and studythe effect of leachate recirculation on Poisson’s ratio and shearmodulus; they state that their results can be improved further, incase independent density measurements are available. Electrical
Manuscript received by the Editor 3 June 2014; revised manuscript received 29 September 2014; published online 22 December 2014.1Delft University of Technology, Section of Geoengineering, Department of Geoscience and Engineering, Delft, The Netherlands. E-mail: l.a.konstantaki@
tudelft.nl; [email protected] University of Technology, Section of Applied Geophysics and Petrophysics, Department of Geoscience and Engineering, Delft, The Netherlands.
resistivity (ER) measurements have been performed at numerouslandfill sites, but they have provided mainly qualitative informationand have suffered from artifacts (e.g., Bernstone et al., 2000; Jollyet al., 2011). However, when ER is used in conjunction with othermethods, they are found to be quite useful (e.g., Leroux and Dahlin,2010; Dahlin, 2012).So far, seismic studies have attempted to characterize landfills
using surface waves (Kavazanjian and Matasovic, 1996; Hakeret al., 1997), but this approach generally incorporates substantialuncertainties. For instance, the heterogeneity is often not taken intoaccount in the inversion of surface wave dispersion data (van Wijkand Levshin, 2004). Cone penetration tests (CPTs) have also beenwidely used for obtaining the density information (Zhan et al.,2008), but they provide a 1D density profile and not the spatial dis-tribution. CPT is expensive, considering that it only provides 1Dinformation. Further, CPT is an invasive approach. Mantlik et al.(2009) show that gravity measurements can be useful in determin-ing the high-density contrasts in a landfill, but the interpretation ofgravity data relies on the availability of ER measurements anddownhole information for depth calculation.The present research is motivated by the growing need for a more
reliable definition of the heterogeneity in the landfill. Additionally,we wanted to explore the possible advantages of using the high-res-olution seismic reflection method in combination with ER tomog-raphy (ERT) in landfill studies. The objectives were to investigatethe following:
• the possibility to localize fine-scale heterogeneities in a land-fill using high-resolution S-wave seismics
• the prospect of sensible density estimation in a landfill fromS-wave velocities
• the supplementary information from multichannel analysisof surface waves (MASW) and the ER method that canbe useful when combined with S-wave seismic studies.
We present a strategy for imaging and characterization of a land-fill using a combination of seismic reflection, MASW, and ERmethods. The idea has been tested on field data from a very hetero-geneous municipal landfill site with a high background noise level.We have examined the uncertainty associated with the MASW
method in landfill application. Finally, we present results of imagingof the landfill heterogeneity and provide the estimated density dis-tribution in the body of the landfill.
FIELD EXPERIMENTS
We acquired seismic and ER data in the Wieringermeer landfill inthe summer of 2013. The Wieringermeer landfill is located in theprovince of North Holland (The Netherlands) and is operated byAfvalzorg (van Meeteren et al., 2009). We performed the geophysi-cal measurements on the eastern part of the landfill, on cell number6, which has a size of 2.6 hectares and a total volume of 281.083tons. The landfill is 22 years old, with 90% of its waste placed in1992–1994 and 10% in 1998. The depth of the landfill is known tobe approximately 12 to 15 m; it has a maximum elevation of 12 m(referenced to sea level) starting at 3 m below sea level. Cell 6 has abottom liner and a leachate- and gas-drainage system, but no topliner other than a soil cover of approximately 1–1.5-m thickness.The waste composition of cell 6 is shown in Table 1. We estimatethe percentage for the different materials based on information onthe specific volume of the material and the total volume of the wastefrom the report of van Meeteren et al. (2009), specifically for cell 6.Information for waste composition in north Europe from Pipatti andVieira (2006) is used for the subcategory of commercial, coarsehousehold, and shredded wastes.Figure 1a shows the photo of the Wieringermeer landfill site; cell
6 is indicated by the red rectangle. We acquired four different datasets: seismic reflection, surface-wave dispersion, ER with Wennergeometry, and ER with dipole-dipole geometry. We performed allmeasurements in two days, with a five-week period between thedays. The weather conditions on the days before the measurementsand during the measurements were similar on both days: 0.2 mm ofrain on the days before the measurements and no rain during themeasurements (L. Meijer, personal communication, 2013). Therewas a busy traffic rod and industry buildings close to the experimentsite (Figure 1a). This resulted in a low signal-to-noise ratio (S/N) inthe seismic data. In addition, operational gas-extraction pipes in thesubsurface of cell 6 added further noise to the data. There were no
Table 1. Waste composition of Wierengermeer landfill, cell 6 (Pipatti and Vieira, 2006; van Meeteren et al., 2009).
Material % Subcategory
Soil and soil decontamination residues 2.29 —Construction and demolition waste 3.17 —
Commercial, coarse household, and shredded waste 72.83 Material %
Food 23.8
Paper 30.6
Wood 10
Textiles 2
Plastic 13
Metal 7
Glass 8
Sludge and composting waste 21.71 —
EN14 Konstantaki et al.
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
truck movements on the eastern side of the landfill on the days ofthe measurements, which was to our advantage.Letters A, B, C, and D in Figure 1a indicate the approximate lo-
cations of the geophone array and the electric cable; this is ex-plained in Figure 1b. The seismic and the ER profiles werecoincident. For the seismic reflection measurements, we used ahigh-frequency electrodynamic, horizontal (S-wave) vibrator asthe source (Ghose et al., 1996; Brouwer et al., 1997, Ghose,2012) and horizontal 10 Hz single-component geophones as receiv-ers. We used S-waves because they are more sensitive to subtlechanges in the soil type (e.g., Ghose and Goudswaard, 2004)and their velocity is directly linked to the small-strain stiffness. Ad-ditionally because of the usually low velocity of S-waves in softsoils, the use of S-waves results in high resolution. For a very
heterogeneous subsurface such as in a landfill and for the very shal-low depths of interest, high-frequency vibrators are more suitablethan impulsive sources (e.g., Ghose et al., 1996, 1998).The acquisition parameters are summarized in Table 2. We used
48 geophones at 0.5 m intervals, resulting in a 23.5-m receiverspread. The receiver array was kept fixed, and the source wasmoved. The source spacing was 1 m. We started shooting 4 m be-hind the first geophone and continued to 4.5 m after the last geo-phone. There was a total of 33 shot points. The horizontal vibratorhad a linear sweep of 20–300 Hz, a time sampling interval of0.5 ms, and a sweep length of 3.2 s. The record length was4.2 s. Crosscorrelation and deterministic source-signature deconvo-lution were tested for compressing the raw vibrograms. The com-pressed vibroseis trace length was 1 s, the same as the trace length
for the impulsive P-wave data acquired at thissite. For every shot location, four sweeps wererecorded separately. Vertical stacking of the shotgathers was performed after vibrogram compres-sion to correct for any shot-to-shot variation(Ghose, 2002). For surface-wave data acquisi-tion, the geometry configuration follows theone of the S-wave reflection profile, with the dif-ference that we used a vertical hammer for the P-wave source, 10-Hz vertical geophones, and atime-sampling interval of 0.25 ms. To increasethe S/N, at each shot point, four to six shot gath-ers were recorded and subsequently stacked.For the ER measurements, we used two dif-
ferent geometries to obtain a good resolution.Wenner measurements provide a good verticalresolution (e.g., Ward, 1990) and a good S/N,whereas dipole-dipole geometry records best theresponse of large anomalies (Cardimona, 2002)and is more sensitive to lateral changes (e.g.,Ward, 1990). We used four electric cables to con-nect 64 electrodes at 1-m spacing, thus having a63-m-long profile. The connection to the acquis-ition system (MPT-DAS I) was in the middle ofthe lines at a 32-m lateral location. The injecteddirect current of the MPT-DAS I was set to2500 mA. The recording time for the Wennermeasurements was 30 min, whereas for the di-pole-dipole measurements, it was 60 min. Foreach position of the current electrodes, we per-
Figure 1. (a) A map of the Wieringermeer landfill with the red rectangle indicating thelocation of cell 6. Letters A, B, C, and D show the approximate location of the electricand seismic cables. (b) Geometry for the geophysical measurements.
Table 2 Acquisition parameters for different geophysical methods.
Seismic reflection MASW ER
Source Horizontal (S-wave) vibrator; sweep ¼ 20–300 Hz P-wave hammer Electric current (DC)
Source spacing 1 m 1 m —Receiver Horizontal 10 Hz geophones Vertical 10 Hz
geophonesElectrodes
Receiver spacing 0.5 m 0.5 m 1 m
Time sampling 0.50 ms 0.25 ms —Total time Record length ¼ 4.2 s ; sweep length ¼ 3.2 s; and
postcompression trace length ¼ 1 s1 s 30 min for the Wenner and 60 min for the
dipole-dipole geometry
Geophysical characterization of a landfill EN15
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
formed three measurements and stacked them subsequently to in-crease the S/N.
DATA PROCESSING AND IMAGING
S-wave seismic reflection
Unlike the usual layered subsurface, a landfill site is predomi-nantly made of many scatterers. This required careful processingof seismic data. Our experiment site was additionally very noisy.We therefore discuss the data processing steps in detail. An over-view of the processing steps is given in Figure 2.After geometry installation and correct header assignment, the
raw vibroseis data are compressed. This is done by crosscorrelationor deconvolution of the recorded data with the source monitor signal
(Ghose, 2002). For the vibrator that we used, a good source mon-itoring is possible. Figure 3 shows a typical raw shot gather aftercrosscorrelation and after deconvolution of the raw vibroseis datausing the source monitor signal. The same band-pass filter (4–10–160–200 Hz) is applied in Figure 3a and 3b, for comparison.Although the sweep of the vibrator is 20–300 Hz, due to the intrin-sic losses in the medium, frequencies lower than 20 Hz are alsosignificantly present in the recorded signals. Below 4 Hz, theground-roll noise is dominant, whereas above 160 Hz, mainlyhigh-frequency ambient noise is present. We have applied automaticgain control with a 30 ms window. The effective source wavelets inthe data obtained by crosscorrelation and deconvolution of thesource monitor with itself are also shown in Figure 3. The decon-volution clearly performs better, allowing for a good separation ofevents (reflections and diffractions), whereas the crosscorrelationburies the events in their dominant ringing characters. The side-lobeenergy is much less in the deconvolved data. It has been shown ear-lier that unless the vibrator monitor signal is sufficiently accurate,the advantage with vibroseis deconvolution is rather limited. How-ever, when the vibrator monitor signal is of high quality, vibroseissource-signature deconvolution performs better than crosscorrela-tion: The side-lobe energy in the effective source wavelet isreduced, resulting in improved resolution and better event separa-tion (Ghose, 2002). Deconvolution corrects for the phase and am-plitude changes. Because deconvolution flattens the amplitudespectrum over the frequency bandwidth representing a good S/N,the wavelet is sharp.In total, four sweeps, two for each opposite horizontal force di-
rection (equivalent to left and right strikes in case of a sledge-hammer S-wave source), are generated at each shot location forthe purpose of S-wave source stacking. The polarity of the vibratormonitor signal was reversed before raw vibrogram compression.Taking the difference between traces at each receiver location forS-wave sources with opposite force directions is expected to min-imize any source-generated P-wave and enhance the S-wave. How-ever, in the case of our data, we find that the signal is too dissimilarbetween S-wave sources with opposite force directions; therefore,Figure 2. Processing steps for the S-wave reflection data set.
Figure 3. Raw compressed vibroseis S-wave shot gather for the source location at 9 m. Raw vibrograms have been compressed by (a) cross-correlation and (b) deconvolution of the source signature. A reliable source monitor is available for this vibrator. The single trace shown on theleft of the shot gather in panels (a) and (b) is the autocorrelation (Klauder wavelet) and the autodeconvolution of the band-limited sourcemonitor.
EN16 Konstantaki et al.
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
we perform only stacking of data for the S-wave source with thesame force direction.A long train of surface waves, as well as direct and refracted
waves, suppress much of the diffracted body-wave arrivals. The sur-face waves appear to also come from other anthropogenic sources atthis site (operational gas pipes and work that was being performedin nearby buildings). Therefore, in the next step, we carefully re-move those unwanted events. Figure 4 shows two shot gathers(source location 9 and 22 m), before and after muting of theseevents. We illustrate only the first 0.5 s of the data because the latertimes are dominated by surface waves, as visible at times exceeding0.3 s. Because the frequency and the propagation velocity of thesurface-waves are very close to those of the S-wave reflections,it is difficult to remove the surface waves with a frequency or fre-quency-wavenumber (f-k) filter without losing a great deal of thetarget reflected wavefield. Surface waves created by the gas facili-ties on the landfill site and the generator we used for the S-wavevibrator have a frequency of approximately 50–60 Hz, which isa range similar to the frequency of the reflection signal. That iswhy we apply surgical muting in addition to top and bottom muting,keeping only the desired events and removing as much as possibleother noise. This is done for every shot separately, taking care tokeep the meaningful events that show up in the successive shotsand remove the rest (surface waves and refractions). Bad tracesare also removed. For such a heterogeneous and noisy landfill site,it is important to do this step carefully for eachshot. As will be explained later, we interpretevents based on the examination of shot gathers,the stacked image, and the migrated image. Weiteratively improve the muting processes so theevents are always evident in each of these threetrace displays. This makes the muting processobjective and consistent. The result is illustratedin Figure 4b and 4d.The trace-edited and muted shot gathers allow
us to identify the scattered body-wave arrivalsfrom the landfill. Having a predominant fre-quency of 80 Hz and a highest velocity of300 m∕s allows for a spatial resolution of at least1 m. The size of the scatterers we identify isaround 1–2 m. One of the main challenges inseismic imaging of landfills has been the resolu-tion of the heterogeneities inside the landfill.Many closely spaced diffractions tend to smearout in the final result (e.g., De Iaco et al.,2003). With the high-resolution S-wave vibroseisdata, it was possible to identify these diffractions,even in a noisy environment, based on a thoroughcross-check on shot gathers and stacked and mi-grated images. A few of these events are high-lighted in Figure 4b and 4d: Red hyperbolasand numbers indicate the diffractions, whereasgray-shaded areas and capital letters indicatethe reflections.A careful velocity analysis of the common
midpoint (CMP) gathers is performed next.The details of the velocity analysis are explainedin the following paragraphs. We use only CMPgathers with fold ≥6. Figure 5a shows the
stacked time section. We perform prestack depth migration to ac-count for the heterogeneous subsurface and to obtain a good imag-ing result in depth. We use a migration scheme that is based onoptimized space-frequency wavefield extrapolation operators(Thorbecke et al., 2004). For migration, we use a simplified,smoothly varying 1D velocity model (macromodel). This modelis obtained by smoothing the 2D stacking velocity field. The pre-stack depth-migrated section is shown in Figure 5b. The 1D velocitymacromodel is illustrated on the right in Figure 5. We apply thesame top and bottom mute to the stacked and migrated sections.We also apply a poststack band-pass filter (10–25–120–160).The interpreted time-stacked and depth-migrated sections areshown in Figure 5c and 5d, respectively.At this stage, we need to check if the diffraction patterns iden-
tified in the shot gathers also migrate to the appearance of localscatterers in the prestack depth image at the correct locations.We, therefore, critically review our interpretation made on shotgathers, the stacked time section, and the prestack depth-migratedsection (Figures 4b, 4d, 5c, and 5d). We observe that scatterers 3, 4,5, and 7 are present and readily interpretable in all three plots. Scat-terers 1, 2, and 6 are not readily interpretable, and the migration failsto image scatterer 6. Layers B through F are evident in all threeplots, whereas it is rather difficult to interpret layer A in the stackedtime section and in the migrated depth section. A feature that isvisible in the migrated depth image, after the diffraction events have
Figure 4. (a) Raw S-wave shot gather for the source located at 9 m, (b) processed S-wave shot gather shown in panel (a), (c) raw S-wave shot gather for the source located at22 m, and (d) processed S-wave shot gather shown in panel (c). The red hyperbolas andthe numbers indicate diffraction patterns; the gray-shaded areas and the capital lettersindicate reflections.
Geophysical characterization of a landfill EN17
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
been collapsed to their true positions, is the bottom of the landfill(green line in Figure 5d). The interpreted landfill bottom is at an 11-to 13-m depth, which agrees well with the expected depth (12–15 m). The landfill below the seismic line consists of some layeredstructures (especially at 20–25-m lateral distance) with interspersedlocal scatterers. As will be shown later on in the joint interpretationof ER and seismic data, these scatterers correspond to density
heterogeneities within the landfill body, which would act as obstruc-tions to leachate flow in the landfill.Errors in the velocity model will obviously result in errors in im-
aging and characterization of the heterogeneities (Zhu et al., 1998).It is important to pay special attention to the velocity analysis incase of a heterogeneous subsurface such as a landfill. Figure 6 ex-plains the steps that were taken to perform the velocity analysis. We
Figure 5. (a) Stacked time section, (b) prestack depth-migrated image, (c) interpreted stacked time section, and (d) interpreted prestack depth-migrated image. A representative 1D velocity field obtained from the stacking velocity is used in migration (shown on the right). The redhyperbolas and the numbers indicate the location of the diffractions, whereas the gray-shaded areas and the capital letters the location ofreflections. The green line in panel (d) indicates the bottom of the landfill. The marked events represent those that are visible in the twoshot gathers in Figure 4. There are more scatterers in these sections that are not marked.
Figure 6. (a) Steps for interactive velocity analysisof seismic reflection data on a very heterogeneouslandfill. (b) An example of step 2: Events E and 5are recognized in the CMP supergather, and theyare used in picking the velocity. (i) Before NMOcorrection, the events are weak in the stackedtraces, (ii) after NMO correction, with correctvelocities the energy is focused, scatterer 5 ap-pears and the reflection becomes much stronger.The purple line in the CVS panels indicates thechange in the rms velocity with TWT that givesthe best illumination for the scatterer.
EN18 Konstantaki et al.
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
create supergathers using five neighboring CMPs. We first carryout constant velocity stacking (CVS) for the velocity range20–400 m∕s, with a step in two-way time (TWT) and in velocityof 10 ms and 10 m∕s, respectively. These parameters should be dataset dependent. A step in the velocity analysis is to address thoseevents that are commonly identifiable in shot gathers and CMPsupergathers. Due to the high level of heterogeneity, we followthe events laterally within a time window. For example, we identifythe events in the supergathers in a 0–10 ms time window. Then, wepick the hyperbola corresponding to the maximum semblance value(Figure 6b); we ensure the goodness of the pick through examina-tion of the CVS panel and the stack section. We iterate these stepstill the event becomes clearly identifiable in the stacked section. Incase of the presence of a diffraction event in the time window, wesee no obvious clear hyperbola in the CMP gather and no flatteningof the event after normal moveout (NMO) correction, but local en-ergy concentration in the CMP gather and after correct velocity as-signment a clear focusing of energy locally in the stacked section(Figure 6b, red hyperbola). Then, we move on to the next 10-mstime window. Keeping in mind that diffractions from local scat-terers are not flattened after the NMO correction, we take greatcare not to mute those events in the shot and CMP gathers. Thisis a challenging task, and this explains why in some areas of thestacked time section the scatterers are imaged well, whereas inother areas they are not (e.g., compare event 1 in the stacked timesection in Figure 5c with that in the shot gather in Figure 4b).Velocity analysis is done at every supergather. The rms velocity isconverted to approximate interval velocity using the Dix equation.The S-wave interval-velocity section is shown in Figure 7a (Fig-
ure 7b and 7c will be explained in the following paragraphs). Wecalculate the approximate depth through time-to-depth conversionusing a smooth velocity model estimated from the stacking veloc-ity field. The heterogeneous velocity field is indicative of the dis-tribution of objects in the landfill body. Note that the areas withstrong velocity contrast generally correspond to locations of thediffraction patterns identified in the shot gathers, suggesting suc-cessful velocity picking for the diffraction events. For example,event 1 located approximately at lateral distance x ¼ 15 m,TWT ¼ 130 ms (Figure 5c) has a velocity of about 180 m∕s.Event 7 approximately at x ¼ 23 m, TWT ¼ 70 ms has a velocityof about 130 m∕s.
Multichannel analysis of surface waves
MASW is commonly used for estimation of S-wave velocity inthe near surface (e.g., Park et al., 1999; Xia et al., 1999; Socco andStrobbia, 2004; Ivanov et al., 2006; Xia and Miller, 2010; Konstan-taki et al., 2013a; Tokeshi et al., 2013).We use P-wave sources and vertical geophones to record surface
waves of the Rayleigh type. In Figure 8, we show typical shot gath-ers for sources located at 9 and 22 m. These shot gathers are domi-nated by Rayleigh-wave energy (Figure 8), whereas the S-wave shotgathers exhibit a weaker presence of Love waves (Figure 4). Lovewaves are generated from the constructive interference of supercriti-cally reflected SH-waves. For the supercritical reflection to occur, alow-velocity top layer must be present meaning that Love wavescan only occur when the surface layer has a lower velocity thanthe half-space (Aki and Richards, 2002; Lowrie, 2007). On the otherhand, the generation of fundamental-mode Rayleigh waves does
not necessarily require a low-velocity upper layer, and the genera-tion of Rayleigh wave depends on P- and SV-wave interferences.We calculate the dispersion curves by picking the maximum en-
ergy in the velocity-frequency plot. In several studies, joint inver-sion of fundamental and higher modes has been found to be superiorto inversion using the fundamental mode only (e.g., Luo et al.,2007). In our data, fundamental and higher modes are present. How-ever, the higher modes are easily identifiable in only a few shots,and they are generally not laterally continuous and lack the low-fre-quency, steeper part of the dispersion curve. As a result, when thesehigher modes are also used in inversion, the inversion becomes un-stable. Furthermore, it might happen that the higher modes do notoccur in order — a problem known as mode jump (e.g., Lu et al.,2006; Dal Moro, 2011), but that different modes are mixed togetherin the observed dispersion images. This is also likely to be the casein our data. Therefore, we use only the fundamental mode. We ap-ply a filter in the f-k domain to remove the higher modes. Invertingthe dispersion curve of the fundamental mode, we aim for a best fitbetween the modeled and the observed curves (Figure 9a). Having atemporal sampling interval of 0.25 ms results in a Nyquist fre-quency of 2 KHz. In Figure 9, the data points are picked at 1-Hz intervals. For the inversion, we use a priori information relevantto landfill deposits: Poisson’s ratio of 0.4 (Sharma et al., 1990; Zek-kos et al., 2008; Carpenter et al., 2013), density of 800 kg∕m3
(Beaven et al., 2008), and a half-space at 15-m depth. The schemeused is based on Occam’s inversion, where the maximum model
Figure 7. (a) S-wave interval velocity section obtained from the re-flection data, (b) S-wave velocity field obtained by MASW usingRayleigh-wave dispersion data, and (c) S-wave velocity field ob-tained by MASW using Love-wave dispersion data. The numbersin the sections indicate the velocities in meters per second. Thecircles and rectangles highlight similar velocities at similar loca-tions in all three sections.
Geophysical characterization of a landfill EN19
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
smoothness is maintained whereas the rms error between the exper-imental and the theoretical curve is minimized (Constable et al.,1987). We obtain 1D velocity profiles at 1.5-m intervals. For thispurpose, we use a window of traces, with the trace number increas-ing or decreasing by 6 traces. For instance, the first 6 traces of allshots provide us with a velocity profile at 5.25-m lateral distance;the first 12 traces result in a velocity profile at 6.75 m distance.Dispersion curves are calculated for each shot position for everywindow. Therefore, for every location, we have 33 dispersioncurves. We invert all dispersion curves at each location and takethe average of the individual best-fit inverted curves to arrive atthe final S-wave velocity profile for that location. To obtain the finalS-wave velocity section, we interpolate between the velocity pro-files (Figure 7b). Note that the MASW section is restricted inthe lateral direction and in depth by our choice of a minimum win-dow of 6 traces, which results in the first CMP lateral location(CMPX) at 5.25 m and the last location at 26.25 m, and a half-spacewith thickness of 15 m. Because we want to image the landfill,whose expected depth is between 12 and 15 m, the depth of pen-etration from MASW is sufficient.For a critical evaluation of the MASW result on the landfill, we
use also the Love waves as recorded in the S-wave reflection dataset and performMASW.Wemute reflections in the raw S-wave data(Figure 4a and 4c) and keep only the surface waves for MASW. Wefollow the same procedure as for the Rayleigh waves. A dispersioncurve for the Love wave is shown in Figure 9b. Comparing thedispersion curves between the Rayleigh and Love waves (Figure 9aand 9b), we see that although both show a relatively good fit, thereis a better fit between the model and the observation for the Lovewaves. Error analyses performed for Rayleigh- and Love-waveMASW support this fact: The rms error is relatively low forthe Love-wave inversion (Figure 10). In Figure 9, we see thatthe rms error after five iterations is 3.8% for the Love-wave inver-sion and 6% for the Rayleigh-wave inversion. Furthermore, theLove wave dispersion curve has a broader frequency bandwidth(Figure 9), which increases the resolution in depth. The lower
the frequencies, the greater the depth of penetra-tion (e.g., Socco and Strobbia, 2004). Figure 7cshows the S-wave velocity section derived by in-version of dispersive Love waves. We see quitesimilar trends in the S-wave velocity fields ob-tained by Rayleigh- and Love-wave inversions(Figure 7b and 7c), although the estimated valueof velocity slightly differs between them. It hasbeen shown in the past that S-wave velocity es-timated from Love and Rayleigh wave data is notthe same (e.g., Lowrie, 2007). In the case of ourdata, another reason for the mismatch betweenthe S-wave velocities estimated from dispersiveRayleigh and Love waves might be the differencein data quality due to the use of different seismicsources — a horizontal (S-wave) vibrator for theLove waves and a vertical hammer for the Ray-leigh waves. Also, anisotropy (for SH- and SV-waves) can play a role in creating the slight mis-match in the estimated velocities.The velocity field obtained from the S-wave
reflection data (Figure 7a) shows more detailscompared to the MASW results (Figure 7b
Figure 9. (a) Rayleigh-wave dispersion curve at CMPX ¼ 22 m and (b) Love-wavedispersion curve at CMPX ¼ 22 m.
Figure 8. Shot gather for P-wave source at (a) 9 m and (b) 22 mlateral distances.
EN20 Konstantaki et al.
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
and 7c). The MASW results assume a layered model, which in thecase of a heterogeneous landfill is not true, and this results in higheruncertainties (e.g., van Wijk and Levshin, 2004). However, the gen-eral trends are still similar. At approximately 100–150 ms, we see asimilar lateral velocity distribution in all three sections: Velocitiesstart at values of 120–130 m∕s on the left (excepting the high veloc-ity of 260 m∕s in the reflection data), reach 150–180 m∕s at ap-proximately 8-m lateral distance (black circle in Figure 7), thenthey decrease a bit, then again rise to 180–190 m∕s at approxi-mately 15-m lateral distance (black rectangle in Figure 7). Thereis also a drop in velocity at approximately 22 m and an increaseat approximately 25-m lateral distance (white circle in Figure 7),visible in reflection data and Rayleigh-wave dispersion data.Although they are less obvious in the MASW results, the velocityheterogeneities at x ¼ 8 m, TWT ¼ 150 ms, at x ¼ 15 m, TWT ¼150 ms and at x ¼ 25 m, TWT ¼ 150 ms that are visible in theS-wave reflection data, are also present in the MASW results.The results for the first 0–100 ms show similar values in all three
sections in Figure 7 and they are well constrained in depth. Veloc-ities at TWT greater than 200 ms start to diverge. Although thevelocity values are in a similar range, MASW is not able to identifythe lateral differences sufficiently. Love and Rayleigh waves show asimilar trend: lower velocities at 5–15 m and higher velocities at15–25-m lateral distances.
Electrical resistivity tomography
ERT is performed to obtain the apparent resistivity distribution inthe landfill. We use a 2D joint inversion for the Wenner and dipole-dipole measurements to increase the depth and the resolution (de laVega et al., 2003). For this purpose, the code RES2DINV (Geo-tomo, 2010) has been used. This code uses a nonlinear least-squaresoptimization technique (de Groot-Hedlin and Constable, 1990;Loke and Barker, 1996). The total number of data points for theinversion is 4292 (650 for the Wenner and 3642 for the dipole-dipole measurements). We perform no preprocessing of the data,other than removal of a few outliers. For the inversion, we use afinite element method with a varying trapezoidal mesh. The damp-
ing factor is set free to vary with depth: from 0.03 to 0.15. Conver-gence is reached at the 5th iteration with a total rms error of 3%.The joint inversion result is shown in Figure 11b; the interval
velocity field from the S-wave reflection data is plotted for compari-son in Figure 11a. ER is related to the presence of leachate (Cardi-mona, 2002). ERT, however, suffers from inversion artifacts andlimited depth resolution (Jolly et al., 2011). The resolution ofERT decreases with depth and depth uncertainty increases. There-fore, what might actually be 11–12-m deep seems deeper in theERT. This might be the reason for the discrepancy in depth in caseof the boundary marked by F between S-wave velocity field andERT section. A joint interpretation of ERTand seismic is beneficial.The ERT section can indicate the locations of wet or dry pockets(lower or higher ER, respectively), whereas the seismic velocitysection shows the location of stiffness/density variations. Lookingat the two properties together, we mark that at the termination ofareas interpreted as a wet or dry pocket, there is also a changein the S-wave velocity. The capital letters and lines indicate thoselocations. In the ERT section above line A, there is a lower resis-tivity area, whereas in the seismic data below location A, there ishigher S-wave velocity. We interpreted this as a wet pocket createddue to the obstruction of leachate due to an underlying stiffer anddenser zone.
LANDFILL CHARACTERIZATION: DENSITYDISTRIBUTION
If we can roughly quantify the density distribution inside a land-fill, then that will enable us to distinguish zones or boundaries thatact as a barrier or obstruction to the flow of leachate. To accomplish
Figure 11. (a) Seismic interval-velocity section obtained from S-wave reflection data and (b) Wenner and dipole-dipole ER joint in-version. The letters and the lines indicate the lower limits of wet ordry pockets and the upper limits for high-stiffness/density zones.
Figure 10. Error analysis for inversion of (a) Rayleigh-wavedispersion and (b) Love-wave dispersion. The rms errors areplotted.
Geophysical characterization of a landfill EN21
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
this, we use the empirical relationship between S-wave velocity(VS) and unit weight (γwaste) as derived by Choudhury and Savoikar(2009) based on a database of measurements on landfills:
VS ¼ 1
0.0174 − 0.000978γwaste: (1)
Equation 1 is based on published results of more than 30 surveysthat involved independent estimation of γwaste and VS values in land-fills. In these surveys, VS is obtained mainly from surface-wavemethods and borehole seismics at landfill sites, which differ inage and composition. We use this equation to obtain γwaste, sepa-rately for the different data sets — data sets obtained from S-wavereflection analysis, from MASW using Rayleigh waves, and fromMASW using Love waves — for up to a 12-m depth where weexpect the bottom of the landfill. We then calculate bulk density asfollows:
ρ ¼ γwasteg
; (2)
where g is the acceleration due to gravity (9.81 m∕s2).Figure 12 shows the relation between γwaste and VS from the
study of Choudhury and Savoikar (2009) and the three seismic stud-ies that we present here. Note the good agreement among all fourcurves. As for the three S-wave velocity fields derived in this study,we use the relationship given by Choudhury and Savoikar (2009) toobtain γwaste. The similarity in trend between these three curves andthat of Choudhury and Savoikar (2009) is obvious. However, theestimation of VS and the curve fitting are done independent of eachother. The very small difference between these best-fit curvessuggests that the estimated value of γwaste will be little affected ir-respective of which approach is used for VS estimation. The twoMASW-derived values show almost no difference at all; whereasthe difference in the estimated γwaste values between reflectionand MASW methods is 1% or less for the VS range 100–160 m∕s.The main distinction between the MASW methods and the S-
wave reflection method is in the detail of heterogeneity mapping.This is primarily due to the higher resolution with the reflectionmethod. The reflection method agrees well with the ERT resultand seems to be more reliable for imaging a heterogeneous landfill.We use the S-wave reflection results for determining the density
distribution in the landfill (Figure 13). Materials with a wide rangeof densities ranging from 100 kg∕m3 up to 1400 kg∕m3 are presenthere. This is in agreement with the density values generally found inmunicipal landfills (EPA, 2008; Leonard et al., 2000; WRAP,2009). However, the very low density at a relatively great depth(x ¼ 21 m, TWT ¼ 200 ms) is rather unrealistic, and it is probablya result of using the empirical relationship below the landfill bot-tom, where the relationship is no more valid. In addition, velocitieslower than 61 m∕s are excluded here because they would corre-spond to very low density values that are not realistic. As seen alsoin Figure 12, the uncertainty of using the relationship of Choudhuryand Savoikar (2009) increases for these very low velocities. Note inFigure 13 that the high densities of the order of 1200–1400 kg∕m3
are present at approximately x ¼ 15 m, TWT ¼ 120 ms. This zonemay act as a barrier or obstruction to leachate flow. Such high-den-sity and/or high-stiffness zones probably correspond to greater com-paction and/or agglomeration of high-density wastes such asconstruction materials. Softer, lower density zones should corre-spond to relatively loose and porous materials where an easier trans-port of fluid is possible.
DISCUSSION
For a heterogeneous subsurface such as a landfill, MASW is notexpected to offer the desired high resolution. A comparison betweenthe MASW and S-wave reflection method is not quite justified be-cause these two methods are based on different principles. However,from a practical point of view, the high-resolution S-wave reflectionmethod is clearly better suited for heterogeneity mapping inside alandfill. Because of the low velocity of S-waves in soft soils, thewavelength is generally short. In addition, in our investigation,the use of a high-frequency electrodynamic vibrator has helpedin enhancing the resolution further. In the future, use of suchhigh-frequency sources might become crucial to image and charac-terize the small-scale heterogeneities, such as those in a municipallandfill.In the context of heterogeneous landfill characterization, MASW
has a few shortcomings: (1) nonuniqueness of the inversionresult impeding reliable delineations of short-wavelength velocityvariations, (2) errors introduced due to the small spatial windows(not enough data), (3) errors due to average values taken for thea priori information, and (4) the assumption of a layered subsurfacewith no scatterer, which is generally not the case in a landfill. Never-theless, MASW can image several primary features in the velocityfield, which can be useful for a firsthand check of the velocity fieldobtained from S-wave reflection data. MASW provides a similar
Figure 12. The relationship between unit weight (γwaste) and S-wave velocity (VS) from the study of Choudhury and Savoikar(2009) based on landfill database and for the VS values obtainedin this study using three different seismic methods. A hyperbolicfitting has been performed to data from all four studies. The regres-sion coefficients (R) are marked.
Figure 13. Density distribution inside the landfill obtained from theVS field derived from S-wave reflection data.
EN22 Konstantaki et al.
Dow
nloa
ded
01/0
5/15
to 1
31.1
80.1
30.1
87. R
edis
trib
utio
n su
bjec
t to
SEG
lice
nse
or c
opyr
ight
; see
Ter
ms
of U
se a
t http
://lib
rary
.seg
.org
/
range of velocities as seismic reflection, thus giving confidence tothe general interpretation.MASW could provide better results, if the data acquisition was
especially adapted for the purpose. For example, by using roll-alongdata acquisition, we would maintain wide spatial windows, thus im-proving the results. While calculating the 1D S-wave velocity pro-files through MASW, we have used gathers of different sourceoffsets and different receiver-spread lengths. This might have af-fected the accuracy of estimation of the dispersion curve — Some-thing that could be avoided by using a roll-along acquisition.Further, a more sophisticated inversion scheme that would take intoaccount the heterogeneity in the subsurface could improve theMASW results. For instance, by using an elastic, homogeneous,isotropic, layered half-space as the forward kernel in the MASWinversion, we implicitly disregard scattering as being effective insurface wave dispersion curve modeling. A full-waveform model-ing should possibly be more appropriate in this case. Finally, it hasbeen shown earlier that a joint inversion of Rayleigh- and Love-wave dispersion curves could be beneficial (Hamimu et al.,2011). However, we have found that only when the modal identitiesof the observed higher modes are clear, such joint inversion can leadto a result that is superior to that of the conventional approach usingthe fundamental mode only. Otherwise, the joint inversion is practi-cally very much limited in its effectiveness. In the case of our data,the higher modes of surface waves are not laterally continuous,causing difficulty in their identification. This has resulted in unsta-ble results in case of joint inversion of fundamental modes togetherwith the higher modes. Also, for a fine-scale delineation of the sub-surface heterogeneity, it is possibly not advisable to ignore the factthat Rayleigh- and Love-wave velocities are generally slightlydifferent.The acquisition and processing of the seismic reflection data
might be considered expensive by landfill operators. However,the higher resolution, the greater reliability, and the significantlymore information content that one may expect from the high-res-olution S-wave reflection data should sufficiently justify the extracost. In this vein, the use of high-frequency seismic sources, such asthe electrodynamic, horizontal vibrator that we have used in ourstudies, is a good option to increase cost/production efficiencyand quality. Although high-resolution S-wave reflections can gen-erally image the heterogeneities better than the MASW method,there are challenges in processing the reflection data from suchhighly heterogeneous sites. To image the short-wavelength varia-tions in the landfill and to identify unambiguously the diffractionsamid reflections and noises, careful data handling and cross-checksamong the shot gather, CMP gather, and stacked section are neces-sary. We have used the focusing of energy in the CMP supergathersand in the stacked section to assign correct velocity to the local dif-fractors. This has been done in an iterative manner.For illuminating the localized shallow scatterers present in a land-
fill, a high spatial density of the source distribution is beneficial.Recently, it has been found, based on synthetic tests, that interfero-metric reconstruction of extra shot gathers, with the source locatedat places where, in reality, there is no active source present, doesgenerally improve the imaging resolution of the very shallow scat-terers in a landfill (Konstantaki et al., 2013b). This approach, if ap-plied to field data presented here, is expected to improve the S-wavereflection results further. Also, full-waveform inversion has beenused successfully in imaging the heterogeneities in the shallow sub-
surface; this may help improve the quality of the derived velocityfield and hence the result of prestack depth migration (e.g., Adamc-zyk et al., 2014).The empirical relationship that we have used to translate S-wave
velocity field to unit weight and then to density distribution is de-rived from independent field studies in more than 30 landfill sites.This is a nonlinear relationship. Therefore, strong contrasts in thevelocity values will result in moderate changes in the density. Inother words, only large velocity contrasts can reliably be translatedto density contrast values. In case of municipal landfills containingdifferent kinds of objects, such large density contrasts are commonand they manifest as diffractors in seismic data. The used empiricalrelationship appears to be quite reliable and useful for such high-density contrast areas, which also define the leachate flow paths.The importance of acquiring high-quality seismic data is para-
mount. We had to mute a large part of the recorded seismic wave-field because of the presence of strong surface waves. There weredifficulties in removing the surface waves, due to the similarity oftheir velocity and frequencies to those of the reflections and scat-tered events. The results will be much better if the amount of surfacewaves and other noises can be reduced through more careful dataacquisition. To acquire data at times when the anthropogenic noiseis minimal is an option. Use of roll-along data acquisition shouldalso be useful in improving the data quality.
CONCLUSIONS
We have presented an approach to localize the heterogeneities ina municipal solid-waste landfill and to determine the distributionand approximate values of density inside the landfill, using com-bined seismic reflection and ER methods. This has been possibleusing specific steps of data processing and iterative velocity analysisfor high-resolution S-wave reflection data, and translation of theseismic velocity field to approximate values of density using an em-pirical relation that is especially valid for municipal solid-wastelandfills. We have obtained a high-resolution image of the subsur-face. Interpreting this image together with the ERT result, we havesucceeded to locate possible wet and dry pockets inside the landfill.Density distribution that we have obtained will be useful in under-standing the pathways of leachate flow inside the landfill. This isimportant for localizing the biochemical behavior of the landfill anddesigning accordingly the treatment procedure, i.e., recirculatingleachate in such a way that it flows homogeneously in all areasof the landfill. From a practical point of view, we have found thatthe seismic reflection method is superior to the MASW method forimaging and characterizing a heterogeneous landfill.
Adamczyk, A., M. Malinowski, and A. Malehmir, 2014, High-resolutionnear surface velocity model building using full-waveform inversion —A case study from southwest Sweden: Geophysics Journal International,197, 1693–1704, doi: 10.1093/gji/ggu070.
Aki, K., and P. G. Richards, 2002, Quantitative seismology: University Sci-ence Books.
Beaven, R. P., W. Powrie, and K. Zardava, 2008, Hydraulic properties ofMSW, in D. Zekkos, ed., Geotechnical characterization, field measure-ments, and laboratory testing of municipal solid waste: Proceedings ofthe International Symposium on Waste Mechanics, ASCE, 1–43.
Bernstone, C., T. Dahlin, T. Ohlsson, and H. Hogland, 2000, DC-resistivitymapping of internal landfill structures: Two pre-excavation surveys: Envi-ronmental Geology, 39, 360–371, doi: 10.1007/s002540050015.
Brouwer, J., R. Ghose, K. Helbig, and V. Nijhof, 1997, The improvement ofgeotechnical subsurface models through the application of S-wave reflec-tion seismic exploration: 3rd Annual Meeting, EEGS, Extended Ab-stracts, 103–106.
Cardimona, S., 2002, Electrical resistivity techniques for subsurface inves-tigation, http://www.dot.ca.gov/hq/esc/geotech/geo_support/geophysics_geology/documents/geophysics_2002/061cardimona_resistivity_overview.pdf, accessed 7 March 2014.
Carpenter, P. J., K. R. Reddy, and M. D. Thompson, 2013, Seismic imagingof a leachate-recirculation landfill: Spacial changes in dynamic propertiesof municipal solid waste: Journal of Hazardous, Toxic, and RadioactiveWaste, 17, 331–341, doi: 10.1061/(ASCE)HZ.2153-5515.0000175.
Choudhury, D., and P. Savoikar, 2009, Simplified method to characterizemunicipal solid waste properties under seismic conditions: Waste Man-agement, 29, 924–933, doi: 10.1016/j.wasman.2008.05.008.
Constable, S. C., R. L. Parker, and G. G. Constable, 1987, Occam’sinversion: A practical algorithm for generating smooth models fromelectromagnetic sounding data: Geophysics, 52, 289–300, doi: 10.1190/1.1442303.
Dahlin, T., 2012, Application of resistivity-IP to mapping of groundwatercontamination and buried waste: Presented at Geophysical Associationof Ireland Seminar on Environmental Geophysics.
Dal Moro, G., 2011, Some aspects about surface wave and HVSR analyses:A short overview and a case study: Bollettino Geofisica Teorica ed Ap-plicata, 52, 241–259, doi: 10.4430/bgta0007.
de Groot-Hedlin, C., and S. Constable, 1990, Occam’s inversion to generatesmooth, two dimensional models from magnetotelluric data: Geophysics,55, 1613–1624, doi: 10.1190/1.1442813.
De Iaco, R., A. G. Green, H. R. Maurer, and H. Horstmeyer, 2003, A com-bined seismic reflection and refraction study of a landfill and its host sedi-ments: Journal of Applied Geophysics, 52, 139–156, doi: 10.1016/S0926-9851(02)00255-0.
de la Vega, M., A. Osella, and E. Lascano, 2003, Joint inversion of Wennerand dipole-dipole data to study a gasoline-contaminated soil: Journal ofApplied Geophysics, 54, 97–109, doi: 10.1016/j.jappgeo.2003.08.020.
EPA, 2008, Waste materials — Density data, http://www.epa.vic.gov.au/business-and-industry/lower-your-impact/~/media/Files/bus/EREP/docs/wastematerials-densities-data.pdf, accessed 10 January 2014.
Geotomo, 2010, Rapid 2-D resistivity & IP inversion using the least-squaresmethod, http://web.gps.caltech.edu/classes/ge111/2010Data/Resistivity/Res2dinv%202.pdf, accessed 10 January 2014.
Ghose, R., 2002, High-frequency shear wave reflections from shallow sub-soil layers using a vibrator source: Sweep cross-correlation versus decon-volution with groundforce derivative: 72nd Annual International Meeting,SEG, Expanded Abstracts, 1408–1411.
Ghose, R., 2012, A microelectromechanical system digital 3C array seismiccone penetrometer: Geophysics, 77, no. 3, WA99–WA107, doi: 10.1190/geo2011-0266.1.
Ghose, R., J. Brouwer, and V. Nijhof, 1996, A portable S-wave vibratorfor high-resolution imaging of the shallow subsurface: 58th AnnualInternational Conference and Exhibition, EAGE, Extended Abstracts,M037.
Ghose, R., and J. Goudswaard, 2004, Integrating S-wave seismic reflectiondata and cone-penetration-test data using a multiangle multiscale ap-proach: Geophysics, 69, 440–459, doi: 10.1190/1.1707064.
Ghose, R., V. Nijhof, J. Brouwer, Y. Matsubara, Y. Kaida, and T. Takahashi,1998, Shallow to very shallow, high-resolution reflection seismic using aportable vibrator system: Geophysics, 63, 1295–1309, doi: 10.1190/1.1444431.
Green, A. G., E. Lanz, and H. Maurer, 1999, A template for geophysicalinvestigations for small landfills: The Leading Edge, 18, 248–254, doi:10.1190/1.1438264.
Haker, C. D., G. J. Rix, and C. G. Lai, 1997, Dynamic properties of munici-pal solid waste landfills from surface wave tests: Presented at 10th EEGSSymposium on the Application of Geophysics to Engineering and Envi-ronmental Problems.
Hamimu, L., J. Safani, and M. Nawawi, 2011, Improving the accurateassessment of a shear-wave velocity reversal profile using joint inversionof the effective Rayleigh wave and multimode Love wave dispersioncurves: Near Surface Geophysics, 9, 1–14, doi: 10.3997/1873-0604.2010029.
Ivanov, J., R. D. Miller, P. Lacombe, C. D. Johnson, and J. W. J. Lane, 2006,Delineating a shallow fault zone and dipping bedrock strata using multi-channal analysis of surface waves with a land streamer: Geophysics, 71,A39–A42, doi: 10.1190/1.2227521.
Jolly, J. M., R. D. Barker, and R. P. Beaven, 2011, Resolution of electricalimaging of fluid movement in landfills: Proceedings of the ICE —Waste and Resource Management, 164, 79–96, doi: 10.1680/warm.2011.164.2.79.
Kavazanjian, E. J., and N. Matasovic, 1996, In situ shear wave velocity ofsolid waste from surface wave measurements: Environmental Geotech-nics, 1, 97–102.
Konstantaki, L. A., S. Carpentier, F. Garofalo, P. Bergamo, and L. V. Socco,2013a, Determining hydrological and soil mechanical parameters frommultichannel surface-wave analysis across the Alpine Fault at Inchbonnie,New Zealand: Near Surface Geophysics, 11, 435–448, doi: 10.3997/1873-0604.2013019.
Konstantaki, L. A., D. Draganov, T. Heimovaara, and R. Ghose, 2013b, Im-aging scatterers in landfills using seismic interferometry: Geophysics, 78,no. 6, EN107–EN116, doi: 10.1190/geo2013-0099.1.
Leonard, S. M., J. K. Floom, and S. Brown, 2000, Estimating method anduse of landfill settlement: Environmental Geotechnics, 1–15, doi: 10.1061/40519(293)1.
Leroux, V., and T. Dahlin, 2010, Time-domain IP and resistivity sectionsmeasured at four landfills with different contents: Presented at 16th Euro-pean Meeting of Environmental and Engineering Geophysics, P09.
Loke, M. H., and R. D. Barker, 1996, Rapid least-squares inversion of ap-parent resistivity pseudosections by a quasi-Newton method: GeophysicalProspecting, 44, 131–152, doi: 10.1111/j.1365-2478.1996.tb00142.x.
Lowrie, W., 2007, Fundamentals of geophysics: Cambridge UniversityPress.
Lu, L. Y., B. X. Zhang, and C. H. Wang, 2006, Experiment and inversionstudies on Rayleigh wave considering higher modes: Chinese Journal ofGeophysics, 49, 974–985, doi: 10.1002/cjg2.919.
Luo, Y., J. Xia, J. Liu, Q. Liu, and S. Xu, 2007, Joint inversion of high-frequency surface waves with fundamental and higher modes: Journalof Applied Geophysics, 62, 375–384, doi: 10.1016/j.jappgeo.2007.02.004.
Mantlik, F., M. Matias, J. Lourenco, C. Grangeia, and H. Tareco, 2009, Theuse of gravity methods in the internal characterization of landfills — Acase study: Journal of Geophysics and Engineering, 6, 357–364, doi: 10.1088/1742-2132/6/4/003.
McDougall, J. R., and I. R. Fleming, 2013, Hydro-bio-mechanical model-ling of filling and post closure landfill behaviour — A case study: 5thInternational Workshop on Hydro-Physico-Mechanics of Landfills, 77–88.
Park, C. B., R. D. Miller, and J. Xia, 1999, Multichannel analysis of surfacewaves: Geophysics, 64, 800–808, doi: 10.1190/1.1444590.
Pipatti, R., and S. M. M. Vieira, 2006, 2006 IPPC guidelines for nationalgreenhouse gas inventories: Waste, vol. 5..
Powrie, W., and R. P. Beaven, 1999, Hydraulic properties of householdwaste and implications for landfills: Proceedings of the ICE — Geotech-nical Engineering, 137, 235–247, doi: 10.1680/geng.1997.137.4.235.
Scharff, H., 2005, Ideas and intentions of the Dutch sustainable landfill re-search programme: Presented at International Seminar and Workshop TheSustainable Landfilling.
Sharma, H. D., M. T. Dukes, and D. M. Olsen, 1990, Field measurements ofdynamic moduli and Poisson’s ratios of refuse and underlying soils at alandfill site: Presented at Geotechnics of Waste Fill — Theory andPractise.
Socco, L. V., and C. Strobbia, 2004, Surface-wave method for near-surfacecharacterization: A tutorial: Near Surface Geophysics, 2, 165–185, doi: 10.3997/1873-0604.2004015.
Stoltz, G., A. J. Tinet, M. J. Staub, L. Oxarango, and J. P. Gourc, 2012,Moisture retention properties of municipal solid waste in relation to com-pression: Journal of Geotechnical and Geoenvironmental Engineering,138, 535–543, doi: 10.1061/(ASCE)GT.1943-5606.0000616.
Thorbecke, J., K. Wapenaar, and G. Swinnen, 2004, Design of one-waywavefield extrapolation operators, using smooth functions in WLSQ op-timization: Geophysics, 69, 1037–1045, doi: 10.1190/1.1778246.
Tokeshi, K., P. Harutoonian, C. J. Leo, and S. Liyanapathirana, 2013, Use ofsurface waves for geotechnical engineering applications in Western Syd-ney: Advances in Geosciences, 35, 37–44, doi: 10.5194/adgeo-35-37-2013.
van Meeteren, M., D. Timens, and K. U. Heyer, 2009, Feasibility study sus-tainable emission reduction at the existing landfills Kragge and Wierin-germeer in the Netherlands: Current status of landfill Wierengermeer,http://www.duurzaamstorten.nl/webfiles/DuurzaamStortenNL/files/R00001_Final_generic_report.pdf, accessed 10 January 2014.
van Vossen, W. J., 2010, Sustainable landfilling in the Netherlands: Develop-ments, methodologies and experiences: Osterreichische Wasser und Ab-fallwirtschaft, 62, 141–148.
van Wijk, K., and A. L. Levshin, 2004, Surface wave dispersion from smallvertical scatterers: Geophysical Research Letters, 31, L20602, doi: 10.1029/2004GL021007.
Ward, S. H., 1990, Resistivity and induced polarisation methods, in S. H.Ward, ed., Geotechnical and environmental geophysics: SEG, 147–190.
White, J. K., J. Robinson, and Q. Ren, 2004, Modelling the biochemicaldegradation of solid waste in landfills: Waste Management, 24, 227–240, doi: 10.1016/j.wasman.2003.11.009.
WRAP, 2009, Summary report — Material bulk densities, resource futures.http://www2.wrap.org.uk/downloads/Bulk_Density_Summary_Report_-_Jan2010.fc38a386.8525.pdf, accessed 10 January 2014.
Xia, J., and R. D. Miller, 2010, Estimation of near surface velocity and qual-ity factor by inversion of high-frequency Rayleigh waves, in R. D. Miller,
J. H. Bradford, and K. Holliger, eds., Advances in near-surface seismol-ogy and ground penetrating radar: SEG, 17–36.
Xia, J., R. D. Miller, and C. B. Park, 1999, Estimation of near-surface shear-wave velocity by inversion of Rayleigh wave: Geophysics, 64, 691–700,doi: 10.1190/1.1444578.
Zekkos, D., N. Matasovic, R. El-Sherbiny, A. Athanasopoulos-Zekkos, I.Towhata, and M. Maugeri, 2008, Dynamic properties of municipal solidwaste, in D. Zekkos, ed., Geotechnical characterization, field measure-ments, and laboratory testing of municipal solid waste: Proceedings ofthe International Symposium on Waste Mechanics: ASCE, 112–134.
Zhan, T. L. T., Y. M. Chen, and W. A. Ling, 2008, Shear strength charac-terization of municipal solid waste at the Suzhou landfill, China: Engi-neering Geology, 97, 97–111, doi: 10.1016/j.enggeo.2007.11.006.
Zhu, J., L. R. Lines, and S. H. Gray, 1998, Smiles and frowns in migration/velocity analysis: Geophysics, 63, 1200–1209, doi: 10.1190/1.1444420.
