NEAR-SURFACE CAVITY DETECTION BY HIGH-RESOLUTION SEISMIC REFLECTION METHODS USING SHORT-SPACING TYPE LAND STREAMER

Tomio INAZAKI, Geological Survey of Japan / Public Works Research Institute, Tsukuba, JAPAN, Shigeki KAWAMURA, Nippon Geophysical Prospecting, Co. Ltd., Tokyo, Japan,

Oshie TAZAWA, Kawasaki Geological Engineering, Co. Ltd., Tokyo Japan, Yoshihiro YAMANAKA, Suncoh Consultants, Co. Ltd., Tokyo, Japan, Naomi KANO, Geological Survey of Japan, Tsukuba, JAPAN,

Abstract

High-resolution seismic reflection surveying by means of short-spacing type Land Streamers was conducted to detect near-surface cavities. The target cavities were inferred to have high potentiality of collapsing because they had been excavated about 60 years ago as air-raid shelter tunnels or drifts of lignite mines. Until now, several surface geophysical methods have been applied to the cavity detection, however, they only provided blurry images for cavities due to the poorness of spatial resolution or insufficiency in obtaining information at the target depths. In contrast, our short-spacing type Land Streamer tools, which were originally developed by the first author, can provide high-resolution images up to 30 m in depth.

We applied three types of Land Streamer tools to the cavity detection: a horizontal geophone array at 30 cm spacing, the same array but 20 cm spacing, and an accelerometer array at 20 cm spacing. Each array has 48 channel sensors mounted on a woven belt which is easy to tow by hand. The targets of this study were abandoned drifts of lignite mine at Nagoya, central Japan, and air-raid shelter tunnels excavated in pyroclastic flow deposits at Kanoya, southern Kyushu.

As a result, distinct diffraction anomalies were imaged in unmigrated stacked sections at the just points where actual abandoned cavities were checked by sounding or drilling from the surface. Some anomalies occurred at depths shallower than the inferred horizons of shelter tunnels or lignite mines, which indicates roof falling and upward growing of cavities. It is concluded that our Land Streamer tools have high capability of detecting cavities smaller than 2 m in diameter located 5 to 10 m in depth.

Introduction

Detection and characterization of underground cavities is one of the major issues for environmental and engineering geology. Although the surface collapse caused by natural cavity such as limestone cave or lava tube is exceptional in Japan, highway road collapse due to erosion beneath the pavement is quite common and GPR (ground penetrating radar) surveying for the concealed void detection has been routinely conducted along highway. Recently, artificial cavities such as underground quarry, old lignite mine, and abandoned air-raid shelter have been arising collapse incidents. These cavities have deteriorated and recently have high potentiality for collapsing because they had been dug during World War II, about 60 years ago, with no thought of safety or in disregard of regulations. After the War, these underground facilities were left away and most of the written references on the facilities have been lost. So it would be safe to say that these cavities are the negative legacy of the war just like as abandoned landmines to be cleared.

Geophysical techniques are expected to contribute to delineate cavities at the near surface in the

area of interest. While GPR has been widely utilized for void detection underneath pavement (e.g. Lewis et al., 2002), the effective depth range of GPR is too shallow to delineate such cavities as old lignite mine and abandoned air-raid shelter. Gravity survey is one of the most inexpensive geophysical methods to cover an area and to find out relatively large cavern (e.g. Butler, 1984; Murata, et al., 1987). However, resolution of the conventional gravity survey is insufficient for detecting small cavities. Resistivity method has also been applied for cavity detection. It is capable to image relatively large cavities as high or low resistivity anomalies if the subsurface structure is not so electrically heterogeneous. The seismic method has played important role in cavity detection owing to its high resolution and superior capability. Pioneer works on application of the high-resolution seismic reflection method to cavity detection were conducted in late 1980s to early 1990s in the United States (e.g. Branham and Steeples, 1988; Miller and Steeples, 1991). They interpreted the zones which lack in reflection energy on the profile as the voids in a 1 meter thick coal seam at the depth of 9 m, or chaotic zones represented as a loss of reflection coherency or an decrease in the dominant frequency on it as the voids in a 0.6 m thick and 7 m deep coal seam. However there still remained ambiguity in their interpretation of cavities. It is required for much higher-resolution seismic reflection to delineate a cavity as a clear or apparent image like as diffraction anomaly in a GPR profile. One of the most effective ways for detecting cavities clearly is to make use of short wavelengths comparable to the GPR (Bachrach and Nur, 1998). As easily imagined, short wavelengths can be achieved by utilizing high frequency wavelets or taking advantage of low velocity as S waves in time domain. However, the higher resolution data we intend to acquire, the smaller heterogeneity tends to affect to the data. In other words, the signals originated from cavities would be disrupted by the subsurface irregularity, even though the cavity is the strongest one.

The high-resolution reflection seismics suffers from two major heterogeneous structures having wide velocity variation. The one is the surficial weathering layer which causes irregular time shifts and deteriorates waveforms reflected from cavities. The other is cavity and its surrounding zone, with high impedance and irregular shaped boundary, but it is the target to be imaged. Despite static corrections are useful for conventional reflection seismics, they are often inappropriate for the high-resolution reflection seismics because time shift and wave deformation is too high to correct it. If such irregular structure of the surficial layer is replaced physically with homogeneous medium, we could have the similar effect equivalent to the static correction. This is not altogether impossible when we set survey line onto the pavement where the surface is composed of homogeneous layers. The pavement acts just like as “water glass” to view subsurface through the rippling weathered surface for the high-resolution seismic surveying (Inazaki, 2002). However special care should be paid to planting geophones on the pavement surface to accomplish good coupling. In response to this, the first author designed and produced a new tool; an integrated geophone array which could be easily towed on the pavement, just like a marine streamer, and named it the “Land Streamer” (Inazaki, 1992). The Land Streamer is composed of a towing belt, geophone units, and a CDP cable. The main feature of the Land Streamer tool is in the non-stretch woven belts on which geophone units are mounted to form a multichannel geophone array. The woven belt, which is originally used for slinging cargos, takes charge of the full tension, and keeps the fixed spacing between geophone units on it when pulled. Each geophone unit houses dual elements horizontally but opposed, and is fastened onto the belt. Whereas each geophone unit is placed on the paved surface through a metallic baseplate instead of firm planting to the ground, the Land Streamer tool provides us comparatively clean data even on the pavement resistant to the traffic noises. After continual refinement, short-spacing types were manufactured and applied for detailed imaging of near-surface structures. The first author was also developed an accelerometer type Land Streamer and showed its efficiency to the delineation of detailed structure of near-surface (Inazaki and Lei, 2003).

Then we applied them to the near-surface cavity detection at two sites. The first field surveying was conducted to delineate near-surface voids originated from abandoned lignite mine at Nagoya, central Japan in the fall of 2002. The survey demonstrated that repetitive profiling of the same line using

different tool or with other geophysical method enables us to identify anomalies more accurately (Inazaki et al., 2003). We next conducted this repeated profiling with different types of Land Streamers for detecting abandoned air-raid shelter tunnels at Kanoya, southern Japan (Inazaki et al., 2004). As a result, cavities about 2 m in diameter located 5 to 10 m in depth were identified as the distinct diffraction anomalies in un-migrated stacked sections. The locations and depths of some anomalies were in concordant with those of the actual abandoned shelters traceable from the surface.

Case Study at Nagoya

Background and Geologic Setting Nagoya is located at the central part of Japan (Fig.1). With its population of more than 2 million,

it has been the core city of the Chukyo (Central Japan) Metropolitan Area, the third large industrial sector of Japan. Nowadays the area is producing automobiles (TOYOTA, MITSUBISHI), metal products (NIPPON STEEL), ceramics (NORITAKE), and many kinds of electric appliances. During two decades including the World War II, the area had forced to use lignite, low-grade but abundant in the area, instead of coal as the energy resources because the interregional transportation network had been entirely destroyed by air raid. The area was far from the coal fields in Japan, mostly located Hokkaido and Kyushu. A large number of small lignite mines were disorderly developed in and around the area with no thought of safety or in disregard of regulations even underneath habitation area. Then many mines were abandoned by 1960 without any backfill measures. After long-term abandonment along with surface development, these old mines have had high potentiality of collapse.

The test site is located at Naka-shidami district, northeastern part of Nagoya. The area is topographically characterized as a small terrace bounded by incised valleys at two sides and a scarp at northern side, and was lignite mining field. The district is now under redevelopment as habitation renewal area. The mined lignite layer at the survey site is very thin about 60 cm, 5 to 20 m deep overlain by terrace deposits. Check drilling conducted for the redevelopment planning has revealed that more than 40 % of the site was inferred to be mined and abandoned without reclamation. Abandoned lignite mines at shallow depths tend to cause surface collapse, which should be circumvented in the habitation renewal area. One of the most effective measures for preventing collapse incident is reclamation, which premises exact positioning of abandoned mine galleries. To do this, such geophysical methods as gravity and resistivity were applied at the site. However they could not delineate subsurface cavities due to the poorness of spatial resolution in comparison with small scaled cavities. Then we applied high-resolution seismic reflection survey using short-spacing Land Streamers combined with GPR and surface wave survey.

Figure 1. Location map of two test sites.

Outline of Field Survey Figure 2 shows two survey lines among four lines set at Naka-shidami district. We located the

lines on the paved roads sloping up to the terrace surface. Check drilling nearby at 130 m from the northern end of A Line had detected a void at 16 m in depth, and at 23 m in the hole at 225 m. The drilling at 45 m on D Line met a void 12 m deep. We interpreted these voids as the part of room after mining because they were as small as 60 cm and occurred at the horizon of lignite layer. Based on the check drilling, the lignite layer dips gently towards southwest, and the depths to the layer lowers from 5 m at the northeastern part to 30 m at the southern A room-and-pillar method was thought to be adopted around here, so delineation of this room-and-pillar structure at the horizon was the target of our survey using short spacing type Land Streamers. Table 1 summarizes the specifications of three types of Land Streamers used in the survey. SH_LS30 type of the Land Streamer is composed of a towing belt, 48 geophone units mounted on the belt at 30 cm spacing, and one CDP cable (Inazaki, 2002). Total weight is about 110kg. This weight is considerably heavy but draggable by hand. ACC_LS20 type Land Streamer uses accelerometer type sensors having high frequency response up to 8 kHz (Inazaki & Lei, 2003). The tool comprises 48 piezoelectric type accelerometer sensors, a short-spacing cable and an interface box, which supplies DC power to each sensor through a pair line, receives seismic signals through the same line and sends them to a 48-channel digital seismograph. Many types of accelerometer sensors can be powered by a DC source using a signal line. In such case, a weak DC of 0.3 to 20 mA is supplied, but the output signal also has DC offset too. When the DC offset in the output is removed, we can utilize the existing seismograph connecting with accelerometers. The interface box supplies the DC power to the sensors and receives signals through the same line in a cable and sends them to a seismograph after removal of DC component. We have made by hand this cable having 48 take-outs at 20 cm spacing by means of an off-the-shelf twist cable. SW_LS2 type Land Streamer is newly designed for multichannel surface wave survey (Hayashi et at., 2003). It consists of 24 geophones

Figure 2. Site map showing survey lines set at Naka-shidami district, northeastern part of Nagoya. More than 40 % of the site was inferred to be mined and abandoned without reclamation from check drilling.

Table 1. Specifications of Land Streamers used for the seismic surveys at Naka-shidami site, Nagoya.

TYPE SH_LS30 ACC_LS20 SW_LS2

ITEM No. of Channels 48 48 24

Channel Spacing 30 cm 20 cm 2 m Sensor Geo/H Acc/V Geo/V Natural Frequency 28 Hz 3Hz-8 kHz 4.5 Hz No. of Elements 2 1 1

Active Section Length 14.1 m 9.4 m 46 m Total Length 23.5 m 18.0 m 50 m Sensor Unit Detachable Detachable Detachable Total Weight Ca. 110 kg Ca. 40 kg Ca. 30 kg

mounted on baseplate. The geophones are clamped to towing ropes at 2 m spacing. The seismic source for SH_LS30 consists of an 8 kg wooden hammer manually struck

horizontally on one side of a wooden plank placed perpendicular to the measurement line on the pavement, weighted down with a rock block. The signal was enhanced through vertical stack (4-8) of the plank hammering. Vertical hitting the pavement surface with a wooden plank was adopted as the source for the measurement using ACC_LS20 tool. Although it is thought that P-waves are predominant in case of the vertical hitting, low frequency S-wave phases are mainly observed in the data acquired using ACC_LS20. This is one of the characteristic features of this tool when it used on the pavement. Comparative tests have revealed the tool mainly records SV waves as explained in Figure 3. Namely, the averaged shot gather obtained using ACC_LS20 tool is quite similar to that of SH_LS20 obtained under the same geometry at the same line but for different source, vertical hitting for ACC_LS20 compared with horizontal striking for SH_LS20.

The survey parameters for each tool and of each line are listed in Table 2. The shot station interval was set as twice as each receiver station interval. Resultantly, we had 24 as the maximum CMP fold at the same CMP bin size for SH_LS30 and ACC_LS20. A stepwise measurement technique similar to reflection seismics was applied SW_LS2. Along with the tools above three Land Streamers, we utilized the same accelerometer array but placed it on the pavement surface (ACC_PL20), and a GPR (SFCW_GPR). In ACC_PL20, accelerometer sensors were stuck on the pavement surface using clay pad at 20 cm spacing, and walkaway shooting along the fixed sensor array was repeated 5 times in A Line, and 3 times in D Line. However the lines were partly covered with this method because it took too much time. As well known, conventional GPR system does not have an enough probing depth range for delineating such cavities located at depths about 5 to 10 m. On the contrary, the GPR system adopted this survey enables to deepen the imaging depths because the system uses frequency modulated continuous waves just like as seismic sweep source (Suzuki et al., 1999). The spacing between transmitter antenna and the receiver was fixed to 3 m, and data were recorded at each 25 cm to form a single coverage profile.

We used two 48 channel seismographs for data acquisition; ABEM MK6+ and StrataView. The field survey was conducted in the fall 2002 as a collaboration research project among Public Works Research Institute (PWRI), National Institute of Advanced Industrial Science and Technology (AIST), and five private geophysical survey companies.

Table 2. Field parameters for each tool.

TOOL SH_LS30 ACC_LS20 ACC_PL20 SW_LS2 SFCW_GPRPARAMETER Shot Interval 0.6 m 0.2/0.4 m 0.2 m 4 m 0.25 m Minimum Offset 3.0 m 2.0 m 0.1 m 1 m 3.0 m Sample Rate 0.5 ms 0.25 ms 0.25 ms 1 ms 13.3 ns Record Length 1.0 sec 0.5 sec 0.5 sec 1 sec 2.67 µs CMP Bin Size 30 cm 20 cm 10 cm 2 m 25 cm A Line 201m/322 100m/406 58m/482 225m/113 201m/804D Line 66m/93 ------ 38m/280 70m/36 66m/274 (Line Length/

Number of Shots)

Figure 3. Comparison of averaged shot gather obtained under the same geometry using ACC_LS20 (left) and SH_LS20 tool (right).

Interpretations Figure 4 compares stacked depth sections along A Line using three types of the Land Streamers

with GPR profile (SFCW_GPR). We applied amplitude-preserving migration to ACC_PL20 data, and additional AGC after migration to SH_LS30 profile. Other two profiles were kept unmigrated to enhance diffraction anomalies.

A profile of SH_LS30 clearly delineates a reflection event which gently dips towards south (left to right). The reflection event is interpreted as the target lignite layer in correlation with check drilling which identified a thin lignite layer intercalated in clay beds at the horizon. SW_LS2 was not listed here because it only imaged up to 5 m deep due to high shear wave velocity in surficial layer at the site. The boundary between the lignite layer and overlying clay layer generally shows negative impedance, because the S-wave velocity of lignite is lower than that of clay layer. So, we interpreted the negative event colored with pale blue as the top of lignite layer. The event is continuously traced through the profile, which means room-and pillar structure was not imaged regretfully. Note that burrs occur on the reflection event at several portions marked by arrows. For instance, small but clear upsurges are

Figure 4. Comparison of seismic sections and GPR profile (top) along Line A. Clear diffractions

are identified in the middle section (ACC_LS20). These diffractions are imaged roughly at the same points in the GPR profile and SH_LS30 section (bottom).

identified at 6, 20, and 138 m in distance, and broad rises at 44 and 66 m. In addition, we could identify similar anomalies above the horizon at 114, 120, and 128 m. These anomalies can be inferred as remnants of diffraction waves migrated at the crest or mixtures of them and a horizontal event from there specific shape. Indeed, even if a small cavity which could not generate reflected waves, would yield diffraction waves as a point source. Furthermore, diffraction waves would appear more clearly on unmigrated stacked section when it is isolated. These anomalies on the reflection event are more obvious in the middle profile, unmigrated stacked section for ACC_LS20 and migrated ACC_PL20 stacked section. Several diffraction anomalies crest just on the reflection event in the ACC_LS20 section. Especially a strong anomaly appears at 16 m, 44 m in altitude on the ACC_PL20 section. In addition, diffraction anomalies are recognized above the horizon in ACC_PL20 at portions similar to the SH_LS30 section. Diffraction anomalies, relatively faint but accompanied by strong ringing waves as the after phase, are also marked out in the SFCW_GPR profile at the similar portions to below two seismic sections.

We interpreted that these anomalies were originated from the real geologic structures, not as artifacts, because they occurred commonly at similar portions in several profiles for different types of wavefields or survey methods. Namely the anomalies which occurred at the lignite horizon were correlated with cavities such as drifts, vertical shafts and enlarged part of galleries. For instance, the anomaly which appeared commonly at 16 m, 44 m in altitude on three profiles was attributed to an existing vertical shaft close to the line. As mentioned, some anomalies are identified above the lignite horizon. These are too deep for artificially buried objects from the surface and too shallow for mined cavities. It is probable to interpret these anomalies as small cylindrical cavities which have been growing upward from the existing drifts or mined rooms. Actually, check drilling detected cavities above the mined lignite horizon. We therefore assessed these anomalies as shallow hidden cavities which have high potential to cause surface collapsing. We then characterized the zones between 0 to 20 m and 110 to 130 m in distance as high potential zone of surface collapse.

Case Study at Kanoya

Background and Geologic Setting Abandoned air-raid shelter tunnels

excavated in pyroclastic flow deposits at Kanoya, southern Kyushu, were the other targets of this study. We conducted the high-resolution seismic reflection survey using Land Streamers at three sites located at the southern fringe of Kasano-hara Plateau, the surface of which is mainly composed of thick pyroclastic flow deposits named “Ito Pyroclastics” erupted about 26 ky. B.P. (Fig. 5). The flow deposits are soft, weakly cemented, and easily excavated even by hand. The Ito Pyroclastics is overlain by reworked pyroclastic sediment about 3 m thick. The surface of the plateau is covered with tephric loess.

During World War II, a number of military air bases for “Kamikaze” (suicide attack) had been constructed in and around

Figure 5. Location map of study areas. The survey lines were set at the southern fringe of Kasano-hara Plateau, Kanoya.

Kanoya. To protect facilities and airplanes from air-raid, a lot of shelter tunnels were dug adjacent to the air bases, especially at fringe part where the pyroclastics forms steep cliffs about 20 m in height. After 60 years abandonment, they have deteriorated and recently have high potentiality of collapse. More than 700 mouths of such shelters were discovered by the local government office. Unfortunately, no useful information has remained on the extension of shelters beneath the plateau.

Until now, several surface geophysical methods have been applied to the cavity detection at Kanoya City. However, they only provided blurry images for cavities due to the poorness of spatial resolution or insufficiency in obtaining information at the target depths (e.g. Handa, et al., 2003). We therefore applied the high-resolution seismic reflection surveying using three types of Land Streamers to the cavity detection at the site.

Outline of Field Survey

The parameters of the survey conducted at Kanoya are listed in Table 3. They were basically the same as those adopted at Nagoya. We applied three short spacing types Land Streamers to 4 lines. SH_LS20 type Land Streamer was designed to reduce manufacturing cost and improve the field performance. 48 geophone units are attached on a woven belt at 20 cm spacing. Because of its all-in-one configuration, only we have to do at survey site is to spread the 15 m long streamer and connect it with a seismograph through a cable. It takes only 5 minutes to do this, which is one of the remarkable features of the Land Streamer.

The field survey was conducted in the midsummer 2003 as the collaboration research project too. It took four days for two crews to accomplish the fieldwork on these three survey lines.

Figure 6 shows the location map of seismic survey line, S_1. An abandoned shelter tunnel about 2 m wide and 3 m high still opened its mouth at the foot of the cliff and was estimated to pass under the S_1 line at 30 m in distance. Another tunnel, mouth of which had been already shut, was inferred to extend into the terrace and intersect the line at about 75 m. These two shelter tunnels were the targets for imaging with the high-resolution seismic reflection. S_2 line was also set

Table 3. Field parameters at Kanoya.

TOOL SH_LS30 SH_LS20 ACC_LS20 PARAMETER Shot Interval 0.6 m 0.4 m 0.4 m Minimum Offset 3.0 m 2.0 m 2.0 m Sample Rate 0.5 ms 0.25 ms 0.25 ms Record Length 1.0 sec 0.5 sec 0.5 sec CMP Bin Size 30 cm 20 cm 20 cm S_1 105m/175 96m/240 98m/245 S_2 180m/300 122m/305 172m/430 S_3 EW 180m/300 ------ 100m/250 S_3 NS 180m/300 100m/250 100m/250 (Line Length/

Number of Shots)

Figure 6. Site map showing survey line S_1 set on a paved narrow road which goes along a rim part of terrace (top). Abandoned shelter tunnels are inferred to pass under the line at 30 and 75 m in distance, about 6 m deep (bottom).

on a paved road which runs on the plateau but along an incised gully (Fig. 7). A collapse incident took place 5 years ago at about 100 m south to the southern end of the line. It was caused by the failure of an abandoned shelter tunnel located just beneath the road. Heavy rain was believed to have had triggered the collapsing. Whereas no trace of the shelter tunnel had been identified at the slope beside the S_2 line, there would remain several tunnels extending into the plateau. The goal of this survey was to make it clear whether the abandoned tunnels exist underneath the road.

Interpretations

Figure 8 shows the stacked time sections along S_1 line using three types of the Land Streamers. The data are unmigrated to display diffraction patterns more obviously. The known shelter tunnel, which passes about 6 m below the line at the distance of 30 m, should be imaged as an anomaly with diffraction in the time section.

The corresponding diffraction anomalies are identified in all sections. However, they are not so clear. Note that two diffraction hyperbolas

Figure 7. Map showing survey line S_ 2 set on a paved road which goes along a rim part of plateau. A collapse incident occurred 5 years ago at about 100 m south to the southern end of the line. Resistivity survey had been conducted parallel to the seismic line, and imaged high-resistivity anomalies at the marked points. A number of Mini-Ram soundings were carried out after the seismic survey.

Figure 8. Time sections along S_1 line obtained using ACC_LS20 (top), SH_LS20, (middle), and SH_LS30 tool (bottom). Distinct diffraction patterns appear at about 75 m in top section. The diffraction originated from the existing shelter tunnel which intersects the line at 30 m is more visible in the middle panel. Due to migration effect, diffraction patterns are faint in the bottom panel.

appear closely each other at the inferred portion. As shown in Figure 6, the shelter tunnel branches just beneath the line. It is attributed to this branching just beneath the line that splitting of diffraction apexes and faint reconstruction. Estimated depths to the diffraction anomalies from averaged stacking velocity (80 m/s) and two way time (150 ms) were concordant with those of the shelter tunnel. Clear diffractions are delineated with apexes of which are at the distances of 72 m and 77 m especially in the top section using ACC_LS20 tool. This splitting may indicate branching of the tunnel too.

The stacked time sections for S_2 line are lined up in Figure 9. The sections are also unmigrated like as S_1 line sections. The top section for ACC_LS20 tool is featured by appearance of many diffraction patterns. On the contrary, only few anomaly patterns are identified in middle and bottom sections. Note that high amplitude reflectors are found at 70 m and 125 m in distance. A reflector can be traced laterally in the top section at about 100 ms in two way time, and correlated to the boundary between tephric loess and reworked pyroclastic sediments at 4.5 m deep. Deeper horizons can be also traced in the lower two sections through 100 to 200 ms in two way time. It is supposed that the strong reflectors above mentioned are interpreted as relatively large cavities greater than the Fresnel zone at that depth. The diffraction at 130 m in distance is identified both in the top and bottom sections, and it is characteristic that abrupt distortion in reflectors is recognized at the portion marked with up directing arrows close to this diffraction. This disturbed zone implies collapsing of cavities. The horizon of diffractions and strong reflectors about 100 ms in two way time seems to be too shallow as the original shelter tunnels, depths of which are presumed to be about 10 m from the surface. This indicates roof fall of original tunnel and upward growing of cavities. Actually, Mini-Ram sounding at 128A point detected a 3 m cavity from 4.3 m to 7.5 m in depth. In correlation with Mini-Ram sounding data, it is concluded

Figure 9. Time sections along S_2 line obtained using ACC_LS20 (top), SH_LS20, (middle), and SH_LS30 tool (bottom). Whereas many diffraction patterns appear in top section, only few anomaly patterns are identified in middle and bottom sections. Note that high amplitude reflectors occur at 70 m and 125m in distance. The diffraction at 130 m is identified both in top and bottom panel, accompanied with abrupt distortion at the portion marked with up directing arrows. Another strong diffraction appears at 40 m in top panel.

that diffraction anomalies which occur about 100 ms are from relatively small cavities, and strong reflections at 100 ms from laterally spreading large cavities.

Conclusions

The high-resolution seismic reflection surveys using the short receiver spacing type Land Streamer were tested for underground cavity detection at Nagoya and Kanoya. The target cavities were abandoned lignite mine and air-raid shelter tunnel both excavated 60 years ago. The Land Streamer surveying showed high field performance and provided high quality data. The survey also demonstrated that repetitive profiling of the same line using different tool or with other geophysical method enables us to identify anomalies more accurately.

As a result, cavities about 2 m in diameter located 5 to 10 m in depth were identified as the distinct diffraction anomalies in un-migrated stacked sections. The locations and depths of some anomalies were in concordant with those of the actual abandoned cavities traceable from the surface. Particularly, the surveys could delineate cavities at shallower depths than the target abandoned mine or air-raid shelter tunnels. Such cavities were inferred to have been growing upward from existing cavities and have high potentiality of surface collapse.

However, it still remains ambiguity on the interpretation of diffraction anomalies in correlation with underground cavities. It is required to prove the survey results or interpretation by means of other measurements such as check drilling. Other important point for cavity detection is to have an appropriate geological model which enables to explain seismic survey results with regard to deterioration process of underground cavity. The more actual model we do have, the more realistic interpretation we could have for detected anomalies.

Acknowledgements

The authors thank Mr. Tom Tanaka of Sofih Corporation, Tokyo, for the production of the Land Streamers. Additional thanks are expressed to Dr. Xinglin Lei of AIST for their field supports at developing stage of the tools. We also express special thanks to Mr. Yukio Funabashi and Mr. Yoshihiro Ogawa of Nagoya Urban Development Corporation, Mr. Suguru Motoshiromizu and Mr. Hiroshi Nagatomo of Kanoya City, who helped us to conduct field survey.

References

Bachrach, R., and Nur, A., 1998, Same wavelength GPR and ultra shallow seismic reflection on a river

point bar: Sand stratigraphy and water table complexity, Expanded Abs of the 68th Annual Meeting of SEG, 4p.

Branham K.L., and Steeples, D.W., 1988, Cavity detection using high-resolution seismic reflection methods, Min. Engin., 40, 115-119.

Brouwer, J., and Helbig, K., 1998, Shallow high-resolution reflection seismics, Elsevier, 391p. Butler, D. K., 1984, Microgravimetric and gravity gradient techniques for the detection of subsurface

cavities, Geophysics, 49, 1084-1096. Handa, S., Kitamura, R., Sakai, H., and Yoshida, Y., 2003, Survey of Underground Cavities by

HFCSMT Method, Proc. SEGJ 108th Conference, 315-318. (in Japanese) Hayashi, K., Okada S., Suzuki H., and Inazaki T. 2003, A surface wave method using Land Streamers,

Proc. SEGJ 108th Conference, 298-301. (in Japanese) Inazaki, T., 1992, Development of subsurface survey methods: in Final Report of Research &

Development of Utilization of Underground Space, vol. 3, 2–26, Ministry of Construction.

Inazaki, T., 2002, Ultra-shallow reflection surveying using short-spacing S-wave Land Streamers: Ann. Symp. Environ. Eng. Geophys. Soc. (SAGEEP2002) Expanded Abstracts, Paper 12GAP6, 9p.

Inazaki, T., and Lel, X., 2003, High-frequency seismic measurements using a piezoelectric type accelerometer array, Ann. Symp. Environ. Eng. Geophys. Soc. (SAGEEP2003), Expanded Abstracts, 1253-1262.

Inazaki, T., Kawamura, S., and Kano, N., 2003, Making of an accelerometer type Land Streamer and its utilization to near-surface cavity detection, Proc. SEGJ 109th Conference, 80-83. (in Japanese with English Abstract)

Inazaki T., Yamanaka Y., Kawamura S., and Tazawa O. (2004): High-resolution seismic reflection survey using Land Streamers for near-surface cavity detection, Proceedings of 7th SEGJ International Symposium, 475-480.

Lewis, J.S., Owen, W. P., and Narwold, C., 2002, GPR as a tool for detecting problems in highway-related construction and maintenance, Proceedings of 2nd Annual Conference on the Application of Geophysics and NDT Methodologies to Transportation Facilities and Infrastructure (Geophysics2002), 11p.

Miller, R.D., and Steeples, D.W., 1991, Detecting voids in a 0.6-m coal seam, 7 m deep, using seismic reflection, Geoexploration, 28, 109-119.

Murata, Y., Hiroshima, T., and Komazawa, M., 1987, On the gravimetric survey around a cave at Oya in Utsunomiya City, Proc. SEGJ 76th Conference, 281-282. (in Japanese)