Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
Contents lists available at SciVerse ScienceDirect
Journal of Volcanology and Geothermal Research
j ourna l homepage: www.e lsev ie r .com/ locate / jvo lgeores
Large-scale internal structure of the Sanbongi Fan–Towada Volcano, Japan: Puttingthe theory to the test, using GPR on volcaniclastic deposits
Christopher Gomez a,⁎, Kyoko S. Kataoka b, Kenji Tanaka c
a University of Canterbury, Natural Hazards Research Centre, WATERWAYS: Centre for Freshwater Management, College of Sciences, Department of Geography,Private Bag 4800, Christchurch 8140, New Zealandb Research Institute for Natural Hazards and Disaster Recovery, Niigata University, Ikarashi 2-cho 8050, Nishi-ku, Niigata 950-2181, Japanc Tanaka Geological Corporation, Kunitaka 2-324-7, Echizen City, Fukui 915-0082, Japan
⁎ Corresponding author at: University of CanterbuCentre, New Zealand.
The Towada Caldera Volcano is located between the prefectures of Aomori and Akita — Northeast Honshu Is-land, Japan. The caldera, today filled by a lake, has produced 15,000 years ago a complex eruption emplacingan ignimbrite topped by the lake outburst flood deposit, through which the present Oirase River erodes. Thisflood deposit has shaped the geomorphologic feature named Sanbongi Fan, on which Towada City extends.This flood event hypothesis is mainly based on sedimentological and geomorphological evidences of floodsmainly from outcrops retrieved from the Sanbongi Fan area. Because of the lack of extended outcrops – typ-ical of the Japanese environment – the present paper has therefore put the theory to the test using GPR(Ground Penetrating Radar) radargram extending along a 640 m length. The GPR used for the survey was aPulse-ekko-Pro with 50 MHz antennas, and the software Reflex was used to process the data. The radargramshave displayed a sole unit, which the GPR could not penetrate. It can be interpreted as the ignimbrite. On topof this deposit a series of subhorizontal layers, with the alternation between a backset and a foreset extendsbetween 5 m and 3 m depth. Above 3 m, the units are regular and subhorizontal. The deposit is also charac-terized by the extensive presence of boulders, which are located along three bands: (1) on top of the ignim-brite; (2) in the units deposited by the outburst flood, between 3 and 5 m depths; (3) and in the units close tothe surface, although part of these punctual elements are most certainly anthropogenic. Compared with theoutcrops, the present research confirms that the material located above the ignimbrite material have beendeposited by the outburst flood, creating large-sheet patterns, which have transported boulders. These sheetsdisplay backset and foreset patterns, depending on the position of the deposit, indicating flow pulsation orsurges under a ‘high-energy-flow’ condition.
Towada Volcano is located in between the prefectures of Akita andAomori, Northeast-Honshu Island Japan (Fig. 1). Its summit reaches915 m ASL (above sea level), but the lake filling the summit calderaonly reaches 400 m ASL. The caldera lake is 11 km in diameter andhas a surface of 61 km2. Within the 11 km caldera, the 6 ka ChuseriPlinian eruption has produced the 2 km wide Nakanoumi caldera.On the rim of this second caldera, the Ogurayama dome has grownand can be seen emerging from the caldera lake (Hayakawa, 1985;Kudo, 2010). This lake of ~4.2 km3
finds an exit in the Oirase River,flowing in a general eastward direction to the Pacific Ocean, 70 kmfrom the outlet of the caldera lake (Kataoka, 2011).
Towada is an active volcano, with at least three major eruptivephases that created the present caldera, with the Okuse eruption
ry, Natural Hazards Research
z (C. Gomez).
rights reserved.
55 ka, the Ofudo eruption 30 ka and the Hachinohe eruption 15 ka,which emplaced a large volume of ignimbrite (Hayakawa, 1985;Kudo, 2005) incised by the Oirase river. The latest eruption of TowadaVolcano occurred in AD915; Towada— an eruption episode emplacedPlinian deposits and ignimbrites. The Ofudo and Hachinohe ignim-brites form the base of the Sanbongi Fan distributed in the down-stream area of the Oirase River, where the city of Towada presentlyextends. The Sanbongi Fan (Nitobe, 1972), known as Towada Terrace(Kudo, 2005), is a geomorphologic feature of 17 km long and 7 kmwide for a present area of 49 km2, with an average gentle slope of0.23°. According to Kataoka (2011), the low-gradient angle and thestacking patterns of the sediments of the Sanbongi Fan suggest thatthe sediments were emplaced by a watery, diluted flow close tohyperconcentrated flows, which lasted for a long period of time.These hyperconcentrated sheet-flows would have acted as conveyerbelt, eventually carrying large boulders to the fan. Indeed, Kataoka(2011) has evidenced from outcrops the presence of large bouldersin the Sanbongi Fan. The size and the large amount of these bouldersdo not correspond with the river's present competence, neither with
Fig. 1. Location of the study area in northeast Japan, Aomori Prefecture, between Towada Lake (impounded in the caldera of the Towada Volcano) and Towada City. The 5 radar-grams used in this study are located on the true left-bank of the Oirase River, perpendicularly (radargram in Fig. 2) and parallel (radargrams in Figs. 3, 4, 5, 6) to the Oirase Rivervalley.
45C. Gomez et al. / Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
the river's estimated maximum paleo-competence and neither to thefan gradient, therefore the author has concluded that the bouldershave been emplaced by a large breakout flood from the caldera lakewith an estimated peak discharge of >2×104 to 3×105 m3/s.
Usually, outcrop information obtained by traditional geological andgeomorphological fieldworks is limited to the terraced edge or the dis-sected body of the fan deposits, or if artificial, road cuts and largequarries. Therefore, the information ismostly concentrated on themar-gins of the deposits and along the present Oirase River and one still lacksinsights in the central part of the Sanbongi Fan, around Towada City.However, there is a possibility that such a gigantic flood can leavemore large-scale landforms and bedforms resulting in sedimentary(internal) structures (e.g., Missoula flood (Bretz, 1923; Baker, 1973);Yenisei River (Komatsu et al., 2009))whichmaynot be detected bynor-mal outcrop scales (even thoughwith large outcrops). In addition, beingin a forestry area and/or urban development under a temperate climaticcondition, the scarcity of natural outcrops in the Japanese environmentand modification of landforms by human activities sometimes hamperthe direct access to sedimentary evidences intact.
Keeping aside difficulties on the extensive survey of such sedi-ments in Japan, the usage of GPR (Ground Penetrating Radar) inEarth-Sciences has been spreading widely in areas as diverse ashydrostratigraphy (e.g. Kostic et al., 2005), landslide assessments(e.g. Grandjean et al., 2006), fault identification (e.g. Rashed et al.,2003), and others from the 1990s. In volcanic and volcanic-derivedterrains, one of the first studies, by Paillou et al. (2001), tested the ap-plicability of GPR in a volcanic dry terrain, however the results werespatially reduced to a ‘borehole-like image’ of the subsurface. Thefirst extensive test of radargrams was conducted by Gomez-Ortiz etal. (2007) at Teide Volcano (Grand-Canaria); the authors have
compared GPR radargrams against various material outcrops: pyro-clastic fall deposits, massive-, heterogeneous lava, and flow dyke in-trusions. This comparison has proven that the GPR is a suitable toolfor the auscultation of volcanic subsurface, and more especially forthe imaging of internal layers. This characterization calibrated againstoutcrops has also been accompanied by a contribution comparingElectric Resistivity and GPR derived data (Gomez-Ortiz et al., 2006).In Indonesia, historic (1815 AD) pyroclastic fall and flow depositshave been studied at Tambora Volcano (Abrams and Sigurdsson,2007) and at Merapi Volcano for the 2006 eruption (Gomez et al.,2008, 2009). Although for older historic deposits, the study of the in-ternal structure of pyroclastic-flow deposits was difficult to read onradargrams, this limitation has not been encountered on pristineblock-and-ash flow deposits. Preliminary studies on lahar depositsat Mt. Semeru (East Java) have also been successful for the examina-tion of internal structures down to 2 m depth, where the signal wasperturbed by the water-table (Gomez and Lavigne, 2010). Carrivicket al. (2007) have also successfully imaged the sedimentary architec-ture of outburst flood deposits in Iceland with the aim of reconstruct-ing the depositional regime.
The present study aims to understand (1) large-scale sedimentarystructures, bedform morphology underneath the Sanbongi Fan sur-face, and (2) detailed boulder distribution resulted from the giganticflood event, for both using GPR applying with large areas on the San-bongi Fan.
2. Methods
The dataset has been acquired on the Sanbongi Fan (Fig. 1), re-cording radargrams of a total length of 640 m. The GPR data has
46 C. Gomez et al. / Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
been recorded using a GPR Pulse-ekko-pro (owned by Tanaka Geo-logical Corporation) mounted with 50 MHz antennas, which have awindow of ~13 m depth (theoretical optimum) and a vertical resolu-tion of 0.5 m.
The data have been encoded based on the distance recorded froma coding-wheel on flat horizontal surfaces in order to ease the dataprocessing. This encoding has been cross-correlated using the GPSGeoX 3.5 G from Trimble (precision 10 cm in x,y during the acquisi-tion with 9 to 12 satellites). The GPS data have been exported as sha-pefiles in the GIS solution of ESRI: ArcGIS 10®, in order to work on thedistribution of the GPR data. The latter has been processed with thesoftware Reflex®. The generic processing is as follows: (1) removalof surface echo; (2) correction of energy decay; (3) ‘dewow’ eliminat-ing the mean on trace; (4) correction of the AGC (Automatic GainControl) gain; and (5) measure of the real velocities from hyperbolas– generated by boulders and other punctual objects – and correctionof the velocity accordingly (0.04 to 0.075 m.ns-1 depending on the lo-cation) — since the study areas were flat, no further topographicalcorrections were necessary.
Once the processing of the radargrams was completed, internallayers and punctual elements, such as blocks/boulders, wereextracted, and the corresponding data were exported into GIS. Itresulted in a series of radargrams of a total extension of 640 m, offer-ing a wide window into the subsurface.
3. Results
3.1. Layering dominated by long sub-horizontal units
In this contribution, we have defined the terms ‘units’ and ‘layers’to identify units separated by linear reflectors, and we have also usedthe terms foresets and backsets in order to ease the explanation of theradargrams. These terms only describe the geometry on radargrambut they do not contain any definitions of stratigraphic and sedimen-tologic realms.
The studied area displays different patterns in the way the unitsare set together. The internal layering is dominated by a subhorizon-tal layering with foresets, backsets, and lenticular units.
On the radargram oriented transversally to the Oirase River valley(Fig. 2), the dominating pattern is a subhorizontal layering extendingin a very regular manner, with little thickness variations within the
Fig. 2. A 200 m long radargram of a transversal section of the southern extension of the Sapunctual elements (A) are boulders or other punctual elements located in the substratum aders which are part of the deposits; (C) are perturbation in the signal created by reflectors carea with a large amount of punctual elements, which create a complex field of hyperbolas
layers. In Fig. 2, we can distinguish at least 5 different layers lyingon a sole layer located between 5 and 7 m below the surface. Thissole layer has very different dielectric characteristics, as the GPR sig-nal does not penetrate it at all. The horizontal lines appearing withinthis lowest unit are echoes of the limits located above. This differencebetween the sole layer and the 5 other layers on top is characterizedby a strong signal reflection.
The subhorizontal top layers are all at least 200 m long. The thick-ness of the layers is between 1 m and 3 m. They are perturbed by (c)in (Fig. 2) which are the results of metallic elements located close tothe surface, blocking the signal. During the data collection, these ele-ments have been recognized as water pipe-covers.
This horizontal layering has also been evidenced on the otherradargrams (Figs. 3 to 6), which are perpendicular to the first radar-gram, offering an image of the longitudinal structure of the deposit.
The 4 last radargrams present however more complexity than inFig. 1. In Figs. 3 and 4, the two first layers located below 5 m depthare regular sub-horizontal units, which extend for at least 100 mlength each. These layers could not be clearly accounted for on theradargram of Figs. 5 and 6.
Lying on top of these two layers, there is an alternation of backsets(dipping upstream) and foresets (dipping downstream) between ~3and ~5 m depths. The foreset series is observed in Fig. 3 between0 and 60 m, and in Fig. 4, between 0 and 40 m. The second series ofunits is the backset, dipping upstream-ward. This backset bed domi-nates the radargrams in Figs. 3, 4 and 6. The backset is continuousfrom 60 m distance (Fig. 3) to the end of radargram in Fig. 4(~100 m length). On the radargram, Fig. 5, there is a short foreset in-tercalated between 0 and 30 m, before the bed displays a backset pat-tern until 120 m. The backset units are not uniform; there are twomain types: (1) short units of length b10 m, with the all layer dippingin one direction (Fig. 4, between 50 and 75 m between 3 and 5 mdepths; in Fig. 5 between 40 and 80 m distance and between 3 and5 m depths); and (2) long units of 10 m and more (e.g. Fig. 5 between60 and 80 m).
These backsets are disturbed by elements such as lens units andshort foresets of units on less than 10 m length (Fig. 3 dist.80–90 m, depths: 3–4 m; Fig. 4 dist. ~20 m and dist. ~40 m, depths:2–3 m).
Located above the foreset and backset, one can find a series of longunits, mostly subhorizontal forming an underdeveloped backset. In
nbongi Fan (location in Fig. 1, Radargram 1). Located between 3 and 5 m depths, thebove 2 m depth, and that could be the results of anthropogenic activities; (B) are boul-lose to the surface; some contain iron and disturb all the imagery below; and (D) is an.
Fig. 3. A 95 m long radargram (location: Fig. 1; radargram 2). The radargram is dominated by subhorizontal layers and a series of shorter layers between 80 and 90 m. The mark (A)around 60 m is created by the hole below a ‘bridge’, covered itself by soil of a few tens of centimeters thick. There are very few punctual elements on this radargram.
47C. Gomez et al. / Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
this series of units located above 2 or 3 m depth, the irregularitiesrecorded in the foreset below do not find any repercussion, and allthe units are very regular.
3.2. Examination of punctual reflectors in the deposit
Punctual elements >50 cm diameter – minimum size detectableby the set of antenna used for the survey – are present on every radar-gram. The punctual elements are concentrated in 3 horizontal bands,which are constant on the 5 radargrams: (1) the first band is locatedbelow 5 m depth, at the upper limit of the lower unit, through whichthe GPR can't ‘see’; (2) the second band is between 5 m and >2 mdepths; (3) the third band is above 2 m depth, with most of the punc-tual reflectors above 1 m depth. We have recognized that the latterpunctual elements could have partly been emplaced by human activ-ity, hence we did not include them in the interpretation.
The punctual reflectors in the band (1) are all isolated elements,and they are distributed quite evenly along the radargrams. Thesepunctual reflectors seem to sit on the lower unit, as they are all lo-cated precisely on the limit of the lower unit. Punctual reflectors ofthe band (2), located between 3 and 5 m depths, are more spreadvertically and they can be isolated (e.g. boulders/blocks B of
Fig. 4. A 95 m long radargram (location: Fig. 1 radargram Fig. 4). The relatively complex layeBs'). Blocks between 1 m depth and the surface can't be confidently attributed to any ‘natu
Fig. 6) or in cluster (e.g. boulders/blocks B of Fig. 5 between 50and 100 m).
4. Interpretations and discussion
The radargrams have revealed that a layered deposit of 5 to 7 mwas lying over a unit of electromagnetically impenetrable material.The layers are generally subhorizontal, with regular transverse (per-pendicular to the valley axis) extents that exceed 200 m. The 4 radar-grams, extending along the present valley, display more complexity;with a series of foresets and backsets and lensed-units located be-tween 3 and 5 m depths. On top of this series and below, one canfind long subhorizontal layers.
Within the deposits, there is also a high number of punctual re-flectors, which are large boulders >50 cm diameter characterizingthe Sanbongi Fan deposits (Kataoka, 2011) located into two depth-bands. The first set of boulders is located at the junction betweenthe lowest unit – through which the GPR can't see through – andwithin the central units, between 3 and 5 m. Punctual elementsabove 2 m aren't scarce either.
The units evidenced by GPR are by comparison with Kataoka(2011), as follows: a first layer of ignimbrite unit located at the
ring is mixed with what appears to be at least two layers with blocks/boulders (As' andral’ process, as they can have been emplaced by anthropogenic activity.
Fig. 5. Longitudinal transect of the Sanbongi Fan (location in Fig. 1, transect Fig. 5). (A) Punctual elements – most certainly boulders – located at the limit between the bottom unitand the other layers on top. (B) Boulders located in units between >2 and 5 m depth. (C) Blocks and punctual elements, which may have been emplaced by human activity.
48 C. Gomez et al. / Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
bottom, through which the radar could not see. From 5 to 7 m depths,on top of the ignimbrite, a series of units more or less regular has beenemplaced by hyperconcentrated-flow sheets. Above ~2 m depth, thesoil layers haven't been part of the present study (GPR work at thisdepth requires a different set of antennas), and therefore won't bediscussed here. The presence of punctual elements in the radargramscan also be assimilated to boulders, as per the description of the San-bongi Fan (Kataoka, 2011).
Despite an excellent correspondence between the GPR data andthe lithological, stratigraphical and sedimentological descriptions ofvisual evidences given by Kataoka (2011), the units located between5–7 m and 2–3 m depths have displayed complexities, which requirefurther discussion.
Compared to other volcanic flow deposits deposited in confined val-leys, the GPR signal reveals a very different structure (Gomez et al.,2008, 2009) to block-and-ash flow deposits, which are characterizedby backsets of short layers, symbol of the progradation of the flowunits during emplacement. The foresets observed in Sanbongi Fan ischaracterized by layers, which are closer to the horizontal and longer.This pattern can be linked to the low-gradient topography on whichthe layers are emplaced and to the open-field setting, which allows a
Fig. 6. Longitudinal transect of the Sanbongi Fan (location in Fig. 1, transect Fig. 6). (A) Puncand the other layers on top. (B) Boulders located in units between >2 and 5 m depths. (C)Perturbation in the signal created by a linear (or planar in 3D) object emplaced at 2 m dep
spread of the layers. The present data is closer to the lahar deposits sur-veyed by GPR at Semeru Volcano (Gomez and Lavigne, 2010). Indeedthe radargram Fig. 2 is (at a different scale) similar to the longitudinaltransect at Semeru Volcano, and the other transects (Figs. 3, 4, 5 and6) have similarities with the transversal radargrams of Semeru Volcano.The Sanbongi Fan being anunconfined environment, thiswould suggestthat the sheet-flows, described by Kataoka (2011), would have beencharacterized by tongues of higher velocity and energy draping uneven-ly the area. These differences would be reduced when reaching theedges of the fan, explaining the very regular sandy units. Such flowwas only one part of theflow,which emplaced the backsets and foresetslocated between 5 m depth and 3 m depth. Pulsations or surges in theflood event with fluctuations in flow energy and sediment concentra-tion would have emplaced the long uniform units, which one can findbelow and above the previously depicted units.
The study of the sedimentary architecture of outburst flood es-kers by GPR (Burke et al., 2010) has provided an internal architec-ture, which seems close to the one we have evidenced at Towada,with antidune cross-strata, subhorizontal plane beds, backset beds,foreset beds and boulder clusters. Therefore, the similarity in for-mation processes of the Sanbongi Fan implies that the possible
tual elements – most certainly boulders – located at the limit between the bottom unitBlocks and punctual elements, which may have been emplaced by human activity. (D)th.
49C. Gomez et al. / Journal of Volcanology and Geothermal Research 229–230 (2012) 44–49
gigantic flood events in volcanically affected areas can be evidencedby further GPR surveys.
5. Conclusions
The present contribution has confirmed the phenomenology of theaftermath of the eruption of the 15 ka ago eruption of the Towada Vol-cano and the rapid construction of the Sanbongi Fan. GPR imaging fromseveral very long transects (longitudinal and transverse) has shownthat it is possible to differentiate ignimbrite, flood sediments and boul-ders. It also confirmed the hints gathered from outcrops that the fanwas built by a succession of sheet-flows. It however, also proved thatdespite the general homogeneity suggested by the outcrops and thepresentmorphology, the outburstflowhas certainly experienced differ-ent phases with an increase and decrease in energy and sediment con-centration, creating large-scale foresets and backsets in the fan depositswith numerous boulders distributed following different patterns.
Acknowledgment
The Niigata University (Japan) has funded 75% of the research, andthe University of Canterbury (New Zealand) has funded 25%. The au-thors also appreciate the help of the Tanaka Geological Corporation,which provided the GPR material. The authors are also indebted totwo anonymous reviewers, who have improved the present manu-script and shared their ideas and recommendation.
References
Abrams, L.-J., Sigurdsson, H., 2007. Characterization of pyroclastic fall and flow depositsfrom the 1815 eruption of Tambora Volcano, Indonesia using ground-penetratingradar. Journal of Volcanology and Geothermal Research 161, 352–361.
Baker, V.R., 1973. Paleohydrology and sedimentology of Lake Missoula flooding in East-ern Washington. Geological Society of America, Special Paper, 144. 79 pp.
Bretz, J.H., 1923. The channeled scablands of the Columbia Plateau. Journal of Geology31, 617–649.
Burke, M.J., Woodward, J., Russell, A.J., Fleisher, P.J., Bailey, P.K., 2010. The sedimentaryarchitecture of outburst flood eskers: a comparison of ground-penetrating radardata from Bering Glacier, Alaska and Skeidararjokull, Iceland. GSA Bulletin 122,1637–1645.
Carrivick, J.L., Pringle, J.K., Russell, A.J., Cassidy, N.J., 2007. GPR-derived sedimentary ar-chitecture and stratigraphy of outrburst flood sedimentation within a bedrock val-
ley system, Hraundalur, Iceland. Journal of Environmental and EngineeringGeophysics 12, 127–143.
Gomez, C., Lavigne, F., 2010. Transverse architecture of lahar terraces, inferred fromradargrams: preliminary results from Semeru Volcano, Indonesia. Earth SurfaceProcesses and Landforms 35, 1116–1121.
Gomez, C., Lavigne, F., Lespinasse, N., Hadmoko, D.S., Wassmer, P., 2008. Longitudi-nal structure of pyroclastic-flow deposits revealed by GPR survey at MerapiVolcano, Java, Indonesia. Journal of Volcanology and Geothermal Research176, 439–447.
Gomez, C., Lavigne, F., Hadmoko, D.S., Lespinasse, P., Wassmer, 2009. Block-and-ashflow deposition: a conceptual model from a GPR survey on pyroclastic-flow de-posits at Merapi Volcano, Indonesia. Geomorphology 110, 118–127.
Gomez-Ortiz, D., Martin-Velazquez, S., Martin-Crespo, T., Marquez, A., Lillo, J., Lopez, I.,Carreno, F., 2006. Characterization of volcanic materials using ground penetratingradar: a case study at Teide Volcano (Canary Islands, Spain). Journal of AppliedGeophysics 59, 63–78.
Gomez-Ortiz, D., Martin-Velazquez, S., Martin-Crespo, T., Marquez, A., Lillo, J., Lopez, I.,Carreno, F., Martin-Gonzalez, F., Herrera, R., De Pablo, M.A., 2007. Joint applicationof ground penetrating radar and electrical resistivity imaging to investigate volca-nic materials and structures in Tenerife (Canary Islands, Spain). Journal of AppliedGeophysics 62, 287–300.
Grandjean, G., Pennetier, C., Bitri, A., Meric, O., Malet, J.-P., 2006. Caracterisation de lastructure interne et de l'etat hydrique de glissements argilo-marneux par tomogra-phie geophysique: l'exemple du glissement-coulee de Super-Sauze (Alpes du Sud,France). Comptes Rendus de Geoscience 338, 587–597.
Hayakawa, Y., 1985. Pyroclastic geology of Towada Volcano. Bulletin of the EarthquakeResearch Institute, University of Tokyo 60, 507–592.
Kataoka, K.S., 2011. Geomorphic and sedimentary evidence of a gigantic outburst floodfrom Towada caldera after the 15 ka Towada–Hachinohe ignimbrite eruption,northeast Japan. Geomorphology 125, 11–25.
Komatsu, G., Arzhannikov, S.G., Gillespie, A.R., Burke, R.M., Miyamoto, H., Baker, V.R.,2009. Quaternary paleolake formation and cataclysmic flooding along the upperYenisei River. Geomorphology 104, 143–164.
Kostic, B., Becht, A., Aigner, T., 2005. A 3-D sedimentary architecture of a Quaternarygravel delta (SW-Germany): implications for hydrostratigraphy. Sedimentary Ge-ology 181, 143–171.
Kudo, T., 2005. Geology of the Towada District. Quadrangle series, scale 1:50,000, Geo-logical Survey of Japan, AIST, 79 p. (in Japanese with English abstract).
Kudo, T., 2010. Eruption age and sequence of Ougurayama lava dome at Towada Volca-no, Northeast Japan Arc. Bulletin of the Volcanological Society of Japan 55, 89–107(in Japanese with English abstract).
Nitobe, K., 1972. Geomorphological study of the terraces in the middle and lowerreaches of the Oirase River. Annals of the Tohoku Geographical Association 24,77–85 (in Japanese with English abstract).
Paillou, Ph., Grandjean, G., Malezieux, J.-M., Ruffie, G., Heggy, E., Piponnier, D., Dubois,P., Achache, J., 2001. Performances of ground penetrating radars in arid volcanic re-gions: consequences for Mars subsurface exploration. Geophysical Research Let-ters 28–5, 911–914.
Rashed, M., Kawamura, D., Nemoto, H., Miyata, T., Nakagawa, K., 2003. Ground pene-trating radar investigations across the Uemachi fault, Osaka, Japan. Journal of Ap-plied Geophysics 53, 63–75.
