Hidden calderas evidenced by multisource geophysical data; example of Cappadocian Calderas, Central Anatolia

Journal of Volcanology and Geothermal Research 185 Ž1998. 99–128

Hidden calderas evidenced by multisource geophysical data; example of Cappadocian Calderas, Central Anatolia J.-L. Froger a,) , J.-F. Lenat ´ a, J. Chorowicz b, J.-L. Le Pennec a, J.-L. Bourdier c , d O. Kose , A. Gourgaud a ¨ d, O. Zimitoglu d, N.M. Gundogdu ¨ a

UniÕersite´ Blaise Pascal, OPGC (Centre de Recherches Volcanologiques), UMR 6524, 5 rue Kessler, 63038 Clermont-Ferrand, France b UniÕersite´ de Paris VI, Departement de Geotectonique, 4 place Jussieu, BP 129, 75252 Paris Cedex 05, France ´ ´ c UniÕersite´ d’Orleans, URA 1366 and CRV, GDR 969 CNRS, BP 6759, 45067 Orleans ´ ´ Cedex 2, France d Hacettepe UniÕersitesi, Muhendisligi Fakultesi, Beytepe Kampusu, ¨ ¨ ¨ ¨ 06532 Ankara, Turkey

Abstract The Cappadocian volcanic field in central Anatolia ŽTurkey. is characterised by a sequence of 10 Neogene ignimbrites. The associated calderas have been partly dismantled and buried by subsequent tectonic and sedimentary processes and, therefore, cannot be readily recognized in the field. Recent progress in the understanding of the stratigraphic correlations and flow patterns has identified two main probable source areas for the ignimbrites. Detailed study of these areas, based on gravity surveys, remote sensing data ŽSPOT and ERS1 images. and digital elevation models ŽDEM., has provided evidence for two major caldera complexes and their relationship to old stratovolcanoes and Neogene tectonics. The older Nevsehir– Acigol ¨ caldera complex, located between the towns of Acigol, ¨ Nevsehir and Cardak, is inferred to be the source of the Kavak and Zelve ignimbrites. The Nevsehir–Acigol ¨ caldera complex is defined mainly by a y35 mGal circular gravimetry anomaly about 15 km in diameter. The boundaries of this, now buried, caldera complex are shown by high gradients on the Bouguer gravity anomaly map. The younger Derinkuyu caldera complex, located between the Erdas stratovolcano and the Ciftlik basin, is inferred to be the source of the Sarimaden, Cemilkoy, and Kizilkaya ignimbrites. It is well-defined ¨ Gordeles ¨ by a rectangular Ž35 = 23 km. gravity low Žy30 mGal. with a positive high Žq20 mGal. in the center. Gravity, remote sensing data and the DEM provide evidence that the Erdas stratovolcano, on the northern margin of the Derinkuyu caldera complex, represents the remnants of a large stratovolcano partly cut by one or more caldera collapses. The positive anomaly within the Derinkuyu caldera complex is centered on the 15-km-wide Sahin Kalesi volcanic massif. Field evidence and structural features inferred from the DEM and remote sensing data strongly suggest that this massif is a resurgent doming associated with the Gordeles ignimbrite eruption. High-resolution ERS1, SPOT and DEM images reveal that the transtensive ¨ regime, active at least since the Miocene, influenced the location of eruptive centers and caldera complexes in Cappadocia. The two caldera complexes are located in transtensive grabens. The subsidence of these grabens, continuing after the caldera collapse events, most likely resulted in the burying of the calderas and could explain the difficulties in identifying them in the field. q 1998 Elsevier Science B.V. All rights reserved. Keywords: caldera; ignimbrite; gravity; remote sensing; digital elevation model; Cappadocia

)

Corresponding author. Tel.: q33-4-73-34-67-29; Fax: q33-4-73-34-67-44; E-mail: [email protected]

0377-0273r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 7 - 0 2 7 3 Ž 9 8 . 0 0 0 5 2 - 3

100

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

1. Introduction It is well-known that large ignimbrite eruptions are associated with caldera collapse Že.g., J.G.R. Calderas Special Issue, 1984.. In central Anatolia, the Nevsehir Plateau volcanic field is characterized by an impressive Miocene ignimbrite sequence covering an area of about 20,000 km2 . Considering the volume of the six major ignimbrites Žbetween 80 and 300 km3 each; total volume ) 870 km3 . ŽLe Pennec et al., 1994., it is likely that each of the eruptions resulted in the collapse of a caldera. Calderas, however, have not yet been unambiguously recognized in the Nevsehir Plateau. Previous geological studies proposed the existence of calderas, but none have been linked to a particular ignimbrite ŽEkingen, 1982; Pasquare et al., 1988.. Difficulties in identifying the ignimbrite sources stem here from the lack of obvious surface expression of ancient, eroded calderas structures. In this paper, we use multisource geophysical data to help identify caldera structures. Our approach consisted of a careful investigation of the source areas proposed by Le Pennec et al. Ž1994. for the major ignimbrite of the Nevsehir Plateau. On the basis of gravity data, remote sensing data and digital elevation models ŽDEM., we identified two major caldera complexes, which at present are almost totally concealed. For clarity, the expression ‘source area’ is used in this paper with the same meaning as used by Le Pennec et al. Ž1994., i.e., area where the vents are most likely to be located. The expression ‘caldera complex’ refers to a large depression formed by overlapping andror nested calderas. The first caldera complex, the Nevsehir–Acigol ¨ caldera complex, is defined mostly by gravity data. It correlates with the poorly documented Acigol ¨ and ¨ ¨ Ž1979. and Nevsehir calderas of Yildirim and Ozgur Ekingen Ž1982.. It consists of partly overlapping structures that we consider to be the sources of the Kavak Ž8.6–11.2 Ma. and Zelve ignimbrites, the oldest of the Nevsehir Plateau ignimbrites. In addition, 2.5D gravity models provide constraints on the geometry and location of the Nevsehir–Acigol ¨ caldera complex and are in reasonable agreement with estimates of extracaldera ignimbrite volumes. The second caldera complex, the Derinkuyu caldera

complex, has not been previously recognized. It is defined by both gravity data and structural features. It probably consists of a complex of at least four calderas that we consider to be the source of the Ž7.8 Ma. Sarimaden Ž8.5 Ma., Cemilkoy, ¨ Gordeles ¨ and Kizilkaya Ž5.0 Ma. ignimbrites. It shows, within its central part, evidence of large scale resurgent doming. Gravity data, remote sensing data and DEM also provide new information on the important role of regional tectonics in the formation and location of the two caldera complexes. Finally we propose a tentative model for the evolution of the caldera complexes.

2. Geological setting Since the Miocene, the convergence of the AfroArabian and Eurasian plates produced a complex set of subduction zones whose location and times life were a function of the geometry of continental plate margins ŽSengor ¨ and Yilmaz, 1981.. These subduction zones produced a volcanic belt between Greece and Iran, forming several provinces of different age and composition ŽInnocenti et al., 1982b.. In Anatolia, the volcanism is distributed within three main provinces ŽFig. 1.: Ž1. In the West Anatolian Volcanic Province, lavas were mostly calc-alkaline from the Eocene to Upper Miocene, then subsequently mostly alkaline ŽKeller, 1983.. This change in lavas composition is related to a transition in tectonic regime with time, from N–S compression to N–S extension ŽYilmaz, 1990.. Ž2. The East Anatolian Volcanic Province, located between the Caucasus and eastern Taurus, has lain in a strongly compressive context since the Upper Miocene Žthickness of continental crust estimated at f 53 km by Dewey et al., 1986.. East Anatolian magmatism is both alkaline and calc-alkaline ŽInnocenti et al., 1976, 1982a; Pearce et al., 1990; Yilmaz, 1990.. Ž3. The Central Anatolian Volcanic Province has developed since the Late Miocene over an Oligocene

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

101

Fig. 1. Structural sketch map of Turkey Žmodified from Le Pennec et al., 1994.. Large arrows indicate motion of Arabia and Anatolia relative to Eurasia. WAVP, West Anatolia Volcanic Province; CAVP, Central Anatolia Volcanic Province; EAVP, East Anatolia Volcanic Province; NAF, North Anatolian Fault; ZMRF, Zagros Main Recent Fault; EAF, East Anatolian Fault; EF, Ecemis Fault; DSF, Dead Sea Fault; TGF, Tuz Golu ¨ ¨ Fault; SF, Sungurlu Fault; AR, Amanos Range. Rectangle indicates location of Fig. 2.

foredeep basin located between the central Taurus Range and the Kirsehir crystalline massif ŽFig. 2. ŽInnocenti et al., 1982b; Pasquare et al., 1988.. The origin of the magmatism of this province, which is dominated by calc-alkaline products ŽInnocenti et al., 1982b; Gourgaud et al., 1992; Temel, 1992., is not clearly understood. The general NE–SW volcanic trend, between Kayseri and Karaman ŽFig. 2., is considered to be the result of a subducted small remnant wedge of oceanic crust at the northern corner of the African plate ŽInnocenti et al., 1975, 1982b.. The lack of deep earthquakes in the northeastern part of the Mediterranean Sea ŽJackson and McKenzie, 1988. may indicate that this inferred subduction is now inactive. 2.1. NeÕsehir Plateau Cappadocia is located in the central part of Central Anatolian Volcanic Province. Since Neogene

times, it has been the site of an intensive volcanic activity. It contains an impressive set of rhyodacitic to rhyolitic ignimbrites emplaced between 11.2 and 2.8 Ma ŽInnocenti et al., 1975; Besang et al., 1977; Temel, 1992.. These ignimbrites are interleaved with continental deposits. The volcano-sedimentary sequence infilled a large foredeep basin between the Kirsehir Massif to the north and the Taurus Range to the south ŽFig. 2.. Mio-Pliocene tectonism disrupted this volcanosedimentary infill, isolating a 10,000-km2 plateau between two large transtensive basins, the Tuz Golu ¨¨ Basin to the west and the Sultan Saz Basin to the east ŽInnocenti et al., 1975.. In this paper, following Le Pennec et al. Ž1994., we refer to this plateau as the Nevsehir Plateau. Erosion and tectonism have deeply incised the volcano-sedimentary sequence of the Nevsehir plateau, exposing in places the ophiolitic basement of the Anatolides–Taurides suture.

102

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Fig. 2. Geological sketch map of Central Anatolia, compiled mainly from the 1:2,000,000 geological map of Turkey ŽMaden ve Tetkik Arama Enstitusu, ¨ ¨ MTA, 1989.. Ž1. Kirsehir units; Ž2. Tauride units; Ž3. Miocene stratovolcanoes; Ž4. Quaternary stratovolcanoes; Ž5. Quaternary basic monogenic volcanoes; Ž6. Quaternary acidic monogenic volcanoes; Ž7. Miocene ignimbritic units; Ž8. Recent alluvial deposits; Ž9. Faults. KD, Kara Dag stratovolcano; KcD, Karaca Dag stratovolcano; HD, Hasan Dag stratovolcano; MD, Melendiz Dag stratovolcano; ED, Erciyes Dag stratovolcano; KoD, Koc¸ Dag stratovolcano; DD, Develi Dag stratovolcano; TGF, Tuz Golu ¨ ¨ Fault; EF, Ecemis Fault; DF, Derinkuyu Fault. SPOT coverage is outlined by solid boxes and ERS1 coverage by dashed boxes. Coordinates are in km, ŽUniversal Transverse Mercator, 36th zone..

Cappadocian volcanic activity also constructed the large Miocene stratovolcanoes of Erdas Dag, Melendiz Dag, Koc¸ Dag and Develi Dag ŽFigs. 2 and 3.. Since the Pleistocene, volcanic activity has been essentially represented by two stratovolcanoes ŽHassan Dag and Erciyes Dag. and by a diffuse set of monogenic centers. The monogenic centers are distributed in three main areas ŽFig. 3.. A first complex of acidic domes, the 1-Ma-old Golu ¨ ¨ Dag Acidic

Complex ŽBigazzi et al., 1993. is located northeast of Melendiz Dag stratovolcano. To the north of the Erdas Dag massif another complex of acidic domes and maars is located, the Acigol ¨ Acidic Complex, formed between f 150 ka and 15 ka ŽBigazzi et al., 1993.. Between the Golu ¨ ¨ Dag Acidic Complex and the Erdas Dag massif, there occurs a field of strombolian cinder cones. With the exception of diffuse geothermal activity

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

103

Fig. 3. Location of the places referred to in the paper. DB, Derinkuyu Basin; CD, Ciftlik Depression; MSC, Melendiz Suyu Canyon. AAC, Acigol ¨ Acidic Complex; GDAC, Golu ¨ ¨ Dag Acidic Complex; SC, strombolian cones. Interval of topographic contours is 50 m. Coordinates are in km, ŽUniversal Transverse Mercator, 36th zone..

ŽVincent, 1993., the Cappadocian volcanic field is not presently active. However, some documents suggest prehistoric, possibly historic, activity for the two

youngest stratovolcanoes Hassan Dag and Erciyes Dag ŽErcan, 1985.. The youngest products of Hassan Dag are less than 6000 years old ŽAydar, 1992..

104

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Rhyolitic domes as young as 15 ka have been dated in the Acigol ¨ Acidic Complex ŽBigazzi et al., 1993.. 2.2. NeÕsehir Plateau ignimbrites and the problem of associated calderas The Cappadocian ignimbrites ŽTable 1. are distributed over the Nevsehir Plateau, between the Kirsehir massif, the Taurus range, and the Tuz Golu ¨¨ and Sultan Saz basins. The ignimbrite sequence starts with the 11.2–8.6-Ma-old Kavak ignimbrites ŽInnocenti et al., 1975; Temel, 1992.. These are followed by six Mio-Pliocene ignimbrites: Ž1. the Zelve ignimbrite ŽLe Pennec, 1991.; Ž2. the 8.5-Ma-old Sarimaden ignimbrite ŽInnocenti et al., 1975.; Ž3. the Cemilkoy ¨ ignimbrite; Ž4. the Tahar ignimbrite; Ž5. the 7.8-Ma-old Gordeles ignimbrite ŽInnocenti et al., ¨ 1975.; Ž6. the 5.0-Ma-old Kizilkaya ignimbrite ŽInnocenti et al., 1975; Besang et al., 1977. and the moderately sized Quaternary Acigol ¨ tuff ŽLe Pennec et al., 1991; Druitt et al., 1995. which forms the top of the ignimbritic sequence. This stratigraphic sequence was initially proposed by Pasquare Ž1966, 1968. and later refined by Innocenti et al. Ž1975. and Pasquare et al. Ž1988.. In our opinion, subsequent modifications introduced by Shumacher et al. Ž1990. are not consistent with field data. The recent work of Le Pennec et al. Ž1994., based on comprehensive field observations, probably provides the most convincing stratigraphic framework.

Table 1 General characteristics of Cappadocian ignimbrites of the Nevsehir plateau Žsimplified from Le Pennec et al., 1994. Name of units in this study

Age ŽMa.

Surface Žkm2 .

Volume Žkm3 .

Kizilkaya Sofular Gordeles ¨ Tahar Cemilkoy ¨ Sarimaden Zelve Kavak ignimbrites

5.0 6.8 7.8

10600 100 3600 1000 8600 3900 4200 2600

180 1 110 25 300 80 120 80

8.5 8.6–11.2

Surfaces and bulk volumes are extrapolated from a probable initial extent and are minimal estimates.

Taking into account that some of the ignimbrites extend far beyond the margins of the Nevsehir Plateau, we calculate that the total surface area originally covered by the ignimbrites is close to 20,000 km2 . The total volume for the outflow tuff sheet of the six largest ignimbrites is ) 870 km3. Such large ignimbrite emissions would almost certainly have been associated with caldera formation ŽWilliams, 1942; Smith, 1979.. On the basis of gravity, magnetic, and geoelectric anomalies, Ekingen Ž1982. suggested the presence of a caldera, referred to as the Nevsehir caldera, south of Nevsehir. Geophysical data suggested the presence of a second smaller caldera on its southwestern flank ŽEkingen and Guven, 1978; Ekingen, 1982,.. ¨ This second caldera had been previously identified on the basis of geological and geomorphological data, and termed the Acigol ¨ caldera by Yildirim and ¨ ¨ Ž1979.. However, the volcanological signifiOzgur cance of the calderas was not appraised by these authors and they did not correlate any ignimbrites with the Nevsehir and Acigol ¨ calderas. Pasquare et al. Ž1988. and Vezzoli and Pasquare Ž1994. speculated that most of the Cappadocian ignimbrites issued from a caldera, 15 km in diameter, situated in the Ciftlik depression, on the northern side of Melendiz Dag stratocone Žsee Fig. 3 for localities.. They did not, however, provide any evidence to support this hypothesis. Shumacher Ž1994. proposed that the eruptive center of the Kizilkaya ignimbrite was located in the Misli plain Žsee Fig. 3 for localities.. He bases this on an isopach map of the Kizilkaya plinian fallout layer and on lateral variations of welding in the Kizilkaya ignimbrite. However, the criteria used by Shumacher cannot be straightforwardly applied to the Kizilkaya ignimbrite. The Kizilkaya basal plinian layer is often eroded by its own ignimbrite in proximal areas ŽLe Pennec et al., 1994.. Further, degree of welding is not a simple function of distance from source in the Kizilkaya ignimbrite. Based on a reliable stratigraphic framework, Le Pennec et al. Ž1994. identified two source areas for the six major ignimbrites of Nevsehir Plateau ŽFig. 4.. They used field criteria including lateral grain size variations, distribution pattern and facies variations of ignimbrites and related plinian deposits, grain fabrics and other current indicators within the

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

105

Fig. 4. Location of the two main source areas of Cappadocian ignimbrites as interpreted from field relations by Le Pennec et al. Ž1994., and distribution of gravity survey stations. Žsquares. 1977 General Directorate of Mineral Research and Exploration of Turkey ŽMTA.; Žcircles. 1992 french Center of Volcanological Research ŽCRV.; Žtriangles. 1994 CRV; Žstars. bases of the two last surveys. K, Kaymakli; D, ¨ Derinkuyu; Go, Il, Ilhara. KA-ZE, source area of the Kavak ignimbrites and the Zelve ignimbrite; SA-CE-GO-KI, source area of ¨ Golcuk; ¨ the Sarimaden, Cemilkoy, and Kizilkaya ignimbrites. Coordinates are in UTM Ž36th zone.. ¨ Gordeles ¨

106

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

ignimbrites. To the north, the area bounded by the towns of Nevsehir, Acigol ¨ and Kaymakli and by the Miocene volcanic massif of Erdas Dag delineate the inferred source area of the Kavak and Zelve ignimbrites ŽLe Pennec et al., 1994.. This area coincides, in part, with a topographic low called Cardak basin. To the south, the source areas of the Sarimaden, Cemilkoy, and Kizilkaya ign¨ Gordeles, ¨ imbrites were infered to be clustered within the Quaternary basin of Derinkuyu ŽLe Pennec et al., 1994.. This latter basin is delimited by the Erdas massif to the north, by the Derinkuyu fault to the east and by the Miocene volcanic complex of Melendiz to the south ŽFigs. 2 and 3.. The western boundary of the inferred source area is not clearly defined by surface features. However, as we have specified before, the expression ‘source area’ has a statistical meaning rather than a structural meaning as it refers to the area where the vents are most likely to be located and not to a caldera. Despite these controls on source areas, specific calderas could not be identified or associated with individual ignimbrites before our study. The preservation of the topographic expression of a caldera depends upon subsequent erosional, tectonic and volcanic activity. Post-collapse evolution of calderas ŽLipman, 1984. may involve: Ž1. intracaldera volcanic activity; Ž2. marginal volcanic activity fed by ring fractures; Ž3. resurgent uplift of the caldera floor due to renewed rise of magma; Ž4. infilling of the collapsed area by lacustrine sediments and by clastics rocks originating from erosion of caldera walls, resurgent dome and intra-caldera volcanoes; andror Ž5. stripping of the caldera escarpment by erosion. Processes 1, 3, 4 contribute to burying or disguising the collapsed area, while processes 2 and 5 cause the caldera margin to disappear. In Cappadocia, all these processes have probably contributed to erasing and concealing enough the original caldera structures to make their recognition very difficult in the field. However, the structural situation in Cappadocia may have been the major factor in causing the disappearance of calderas from the surface. Indeed, the infilling of the foredeep basin since the Miocene by material derived from the erosion of the Taurus and Kirsehir massif has resulted in embedding the volcanic structures within a thick detrital sedimentary sequence. In addition, the

vigorous Mio-Pliocene tectonic activity has also played a major role in concealing the calderas. Another type of complexity may arise if the formation of the calderas diverges notably from simple elliptical collapse. In a recent review, Lipman Ž1997. stressed the fact that the geometric diversity of calderas represents a continuum of features and processes bounded by few end members such as funnel-shaped calderas, downsagged calderas, incrementally piston collapsed calderas, trap-door calderas or volcano-tectonic depressions. Complex structures could result from nesting or overlapping of two or more of these end members or intermediate members during successive collapses. It is indisputable that identifying such complex structures in the field is significantly more difficult than identifying structures resulting from simple collapse process, especially when they are partly concealed by subsequent processes Žerosion, tectonics, sedimentation.. The failure of geological mapping to unequivocally located calderas in Cappadocia led us to apply indirect methods: gravity surveys, satellite imagery and DEM analysis. The search for concealed calderas using these methods was greatly aided by the work of Le Pennec et al. Ž1994.. It provided us with a constrained framework in which to optimally organise our geophysical surveys. In particular we focus on geophysical analysis within and around the two source areas delineated.

3. Geophysical data 3.1. GraÕity The gravity method is especially well-suited to providing evidence for, and thus studying, ash flow calderas. Although, for some large calderas, the bulk of the fill could be a densely welded tuff that has little or no density contrast with surrounding rocks, many studies demonstrate that ash flow calderas are generally associated with negative gravity anomalies ŽWilliams and Finn, 1985; Rymer and Brown, 1986; Carle and Goldstein, 1988; Barberi et al., 1990; Deplus et al., 1995.. Even long after the surface morphologic expression of the calderas has disappeared, a significant gravity anomaly can remain

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

ŽMorgan et al., 1984; Ratte´ et al., 1984; Steven et al., 1984.. The gravity data used here were collected during three surveys ŽFig. 4.. In 1977, a first survey was conducted by the Geological Survey of Turkey ŽMTA. using a Worden Master Gravimeter. One thousand gravity stations were measured, mostly in the central part of the Nevsehir Plateau, and leveled by a second-order leveling survey. We extended this first data set with two additional surveys in 1992 and 1994. Using a Lacoste and Romberg G gravimeter, we measured 700 gravity stations in the southern, eastern and northern parts of the Nevsehir Plateau. The accuracy of the gravity measurement is about 0.07 mGal. The absolute horizontal positioning of all gravity stations is better than 100 m. Elevation of gravity stations was provided by barometric leveling and by interpolating elevation from a precise digital elevation model. The accuracy of elevation is within a few centimeters for the MTA gravity stations and better than 7 m for our gravity stations. The distribution of the total set of stations is homogeneous, with an average density of one station per 2 km2 . The three surveys have been linked to the International Gravity Standardization Network 1971. Drift, latitude, free air, Bouguer and terrain corrections were applied. We used the algorithm of Longman Ž1959. for the earth tide correction. Terrain corrections were processed numerically by the conic prism method ŽOlivier and Simard, 1981., using a digital elevation model. Curvature correction ŽBullard B. ŽLaFehr, 1991. has been included in the Bouguer correction. More details of the gravity survey and preliminary data processing can be found in

Froger Ž1996.. The choice of the rock densities is a critical step in gravity data processing. In a volcanic context, density anomalies and topography are often correlated ŽWilliams and Finn, 1985., and thus, the method of determining the density for terrain correction proposed by Nettleton Ž1939., may not be wise. We have computed Bouguer anomaly maps using a set of density correction ranging from 1800 to 3000 kgrm3. Although we selected the 2200 kgrm3 anomaly map to illustrate our discussion in this paper, we based our interpretation of gravity data on all the computed Bouguer anomaly maps. The 2200 kgrm3 density value may seem low for a Bouguer correction in a continental area, although the mean density of the thick volcano-sedimentary sequence of the Nevsehir Plateau should be significantly lower than the 2670 kgrm3 generally used in continental context. However, without any data about the density of the deep country rocks, the density of 2200 kgrm3 seems a reasonable choice, as it corresponds to the average density of outcropping rocks within the Nevsehir Plateau ŽTable 2.. Taking a higher correction density will result in changing the amplitude of the anomalies Že.g., negative anomalies would be deeper. but not the general pattern of anomalies. For this density correction, the final accuracy of Bouguer anomaly is better than 1.6 mGal. The 2.5D inverse models were computed for selected anomalies. Density values used in modeling are listed in Table 1. Density measurements were made only for surface samples. In order to take into account the increase of density with depth, Žespecially important for material such as ignimbrites and sediments., the highest density range values have been considered.

Table 2 Densities of main rock units of Nevsehir Plateau Rock type

Density range Žkgrm2 .

Mean density Žkgrm3 .

Number of samples

Andesitic and basaltic lavas Ignimbrites Ophiolitic bedrock Marble Sedimentary rocks

2500–2800 1300–2100 2600–2900

2600 " 100 1670 " 380 2780 " 120 2800 2400 " 350

9 10 5 1 4

1800–2700

107

108

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Fig. 5. Bouguer anomaly map of Cappadocia. Density used for corrections is 2.2 grcm3. Contour interval is 2.5 mGal. A1 to A9 refer to main anomalies. Dashed rectangle locates Figs. 8–10. Coordinates are in UTM Ž36th zone..

The Bouguer anomaly map ŽFig. 5. does not show any obvious regional trend across the whole map area. Therefore, in order to avoid artificially modify-

ing the relative amplitude and geometry of the anomalies, most of the processing was done on a regionally uncorrected Bouguer anomaly.

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Cases of calderas associated with positive anomalies have already been documented ŽBatur caldera, Yokoyama and Suparto, 1970; Masaya caldera, Bonvalot et al., 1992; Tengger caldera, Froger et al., 1992.. The anomalies are mostly associated with deep dense magma chambers whose extension and density are large enough to obliterate a possible gravity effect of caldera filling. However, most calderas formed by silicic ash-flow eruptions are associated with negative Bouguer anomalies ŽWilliams and Finn, 1985.. For this reason, we focused our analysis on the negative anomalies of the Bouguer map. The Bouguer anomaly map exhibits a complex set of relative positive and negative anomalies. We refer here to the most significant of them, numbered A1 to A9 in Fig. 5. In the northern part of the map the A9 positive anomaly is superimposed on the crystalline rocks of Kirsehir Massif. In the west and central part of the map, we consider a large positive anomaly ŽA4. to be the gravity signature of an ophiolitic massif along the Anatolides Taurides suture. This massif was identified in the field by several outcrops of ultrabasic rocks in this area ŽFig. 3.. The Bouguer map shows five negative anomalies that could possibly correspond to calderas. To the northeast of anomaly A4 occur two negative anomalies, A7 and A1. A1 is geographically superimposed on the source area of Kavak and Zelve ignimbrites proposed by Le Pennec et al. Ž1994.. We discuss this anomaly in more detail later. In our opinion, A7 cannot have a volcanic origin because volcanic structures or volcanic products with proximal facies are lacking in this area. However, we think that the crescent-shaped anomaly formed by the coalescence of A1 and A7 has a regional significance as will be discussed later. To the south of anomaly A4 occur three negative anomalies, A3a, A3b and A8. A3a is superimposed partly on the source area of the Sarimaden, Cemilkoy, ¨ Gordeles and Kizilkaya ignimbrites proposed by Le ¨ Pennec et al. Ž1994., and partly on Miocene and Quaternary volcanic structures. As for the A1 anomaly, we will discuss A3a and A3b anomalies in more detail in later sections. Anomaly A8 is not associated with any visible volcanic structures or proximal volcanic product. Therefore we do not believe that it could be a signature of a caldera. Its

109

origin remains obscure to us and will not be discussed further. 3.2. Remote sensing data The use of satellite images as tools for detecting and analyzing tectonic and volcanic structures has frequently been illustrated Že.g., Bellier et al., 1991a,b; de Silva and Francis, 1991; Chorowicz et al., 1992, 1994, in press; Rowland and Munro, 1992; Detourbet et al., 1993.. Such images provide a unique view of large-scale structures which are often difficult to locate in the field. Furthermore, the resolution of some available images Ž10 and 20 m for SPOT, 12.5 m for ERS1. makes it possible to see the details necessary to understand tectonic and volcanic structures. In acidic volcanic context, satellite images are particularly useful to detect faults that are generally difficult to identify in the field because the soft volcanic products do not record very well brittle deformation. In addition, radar images from ERS1 contain textural information which helps discriminate between different types of terrain. For this study, we assembled a mosaic of four multispectral SPOT images and a mosaic of two ERS1 radar images Žpixel: 12.5 m; image dimensions: 100 = 100 km. covering the entire surveyed zone. In some areas, partial overlapping between some images allows, a stereoscopic view. We use the two mosaics for detailed mapping of lineaments and other structural features. The mosaic coverage is outlined in Fig. 2. We did not apply particular processing to the images except dynamic enhancement. 3.3. Digital eleÕation models (DEM) Much information on the physical nature of terrains and associated structures can be obtained from the morpho-structural analysis of topography. DEMs allow one to visualize the topography in unrivalled ways Ž3D representation, vertical exaggeration, artificial shading, views from different directions, etc... Moreover, it is possible to use DEMs with satellite images to procure an overall better description of an area ŽDuffield et al., 1993; Chorowicz et al., 1994; de Chabalier and Avouac, 1994.. In particular, it can

110

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

help discriminate which satellite-deduced lineaments are truly tectonic features. Two DEMs have been produced for this study. The first, at regional scale, was made by digitizing contours on 1r250,000 maps. It covers an area of 260 = 320 km2 and has a resolution of 500 m. The second is more detailed and focuses on the central part of the Nevsehir plateau Ž53 = 69 km2 .. The data come from the digitization of 1r25,000 maps and has a resolution of 50 m. The vertical accuracy is about 8 m. We developed a software based on a Kriging method ŽJournel and Huijbregts, 1978. to interpolate the data and generate the two grids.

4. Interpreted data; caldera locations 4.1. The NeÕsehir–Acigol ¨ caldera complex The Kavak ignimbrites are mostly present in the northern and northeastern parts of the Nevsehir plateau and extend over a surface area of at least 2600 km2 . Their highest elevation Ž1400–1500 m. is just northeast of the Cardak depression Žsee Fig. 3 for localities., from where they dip gently to the north and east. Their thicknesses also decrease in the same directions. Proximal facies are present only in the vicinity of Cardak area Žlithic clasts ) 20 cm, thickest fall depositf 70 cm, Le Pennec et al., 1994.. This evidence suggests that the present Cardak depression corresponds to the vent area of the Kavak ignimbrites. The fact that Kavak ignimbrites are not present to the west and south of this area suggests that a topographic barrier, coinciding with the Erdas massif, prevented flow of the ignimbrites in those directions. Unfortunately, we have no dates for the Erdas massif to evaluate this hypothesis. The Zelve ignimbrite ŽLe Pennec et al., 1994. is mainly exposed in the northern part of the Nevsehir Plateau over an area larger than 4200 km2 . Several observations indicate a source location just SW of Nevsehir, more or less in the same area as the Kavak source: Ž1. isopachs of Zelve fallout pumice suggest a vent location SW of Nevsehir; Ž2. elevation and thickness of the ignimbrite decrease westward, northward and eastward from Acigol–Nevsehir–Cardak ¨ area; Ž3. local paleo-current directions inferred from

a-axis lineations also suggest a source to the south of Nevsehir. In addition, as for the Kavak ignimbrites, the lack of Zelve ignimbrite exposures south of Erdas Dag again leads to the assumption that a pre-existing topographic barrier was present. The vents of Kavak and Zelve ignimbrites are therefore believed to lie approximately in the same area, between Acigol, ¨ Nevsehir and Cardak ŽLe Pennec et al., 1994.. We here propose several arguments to associate these eruptions with a complex of now buried calderas within the Acigol–Nevsehir–Cardak ¨ area. The major evidence is a prominent gravity low ŽA1. on the Bouguer anomaly map ŽFig. 5., located to the south of Nevsehir. The shape of this anomaly is well-constrained by the gravity data in this area ŽFig. 4.. It is sub-circular in form, with a diameter of 15 km and a y35 mGal amplitude. It has a westward extension ŽA2. at the north of Erdas Dag massif ŽFig. 5.. We assume that A1 and A2 are caused by the infilling of a depression by low-density material corresponding to the Kavak and Zelve ignimbrites. Constraints on the geometry of this depression can be given through modeling and first vertical derivative mapping from the Bouguer anomaly map. The 2.5D inverse models, using the Marquardt method ŽWebring, 1985., were calculated along three profiles ŽFig. 6. across anomaly ŽA1.. The use of 2.5D modeling rather than 2D is required as the width of the depression cannot be considered infinite relative to depth. Since no drilling data are available from this area, we have to make assumptions about subsurface rock densities. We assume that the density of the depression fill increases with depth. This is roughly mimicked by a two-layer model. We arbitrarily choose the lowest measured value of the ignimbrite densities for the density of the top layer, and the highest measured value of ignimbrite densities as that of the bottom layer. The two layers were modeled by a set of prismatic bodies of which lengths are equal to the depression lengths. Special care was exercised in obtaining identical structures at the crossing points of the three profiles. According to this model, the depression appears asymmetric with relatively steep slopes to the north, east and south but gently to the west. It is likely that the depression is deeper in its western part than it appears on the

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

X

X

111

X

Fig. 6. Gravity models. Ža., Žb. and Žc. represent 2.5D models along AA , BB and CC profiles, respectively. Žd. shows location of the three profiles and the inferred margin of the Nevsehir–Acigol ¨ Caldera Complex on a topographic map of Cappadocia.

models; the gravity high ŽA4., due to dense ophiolite ŽFigs. 3 and 5., strongly interferes to the west with the gravity low ŽA1–A2.. As a result, the western margin of the negative anomaly ŽA1–A2. appears smoother than it would be without the presence of the ophiolithic massif. Without constraints such as drill-hole data or other geophysical data, it is not reasonable to attempt further refinement of the model. On the basis of the 2.5D model, it is possible to reconstruct approximately the 3D geometry of the depression. The depression appears to result from the coalescence of two structures. The main one, at the east, has a N408W elongation and a depth greater

than 2 km. The other, at the west has a WSW–ENE elongation and, apparently, is less deep. These features are clearly seen on the first vertical derivative map ŽFig. 7.. The first vertical derivative enhances short-wavelength anomalies associated to shallow features, at the expense of long-wavelength trends related to deeper structures. Here, the A2 component of the negative anomaly on Fig. 5 is better individualized and we interpreted it as a separate caldera rather than a secondary feature resulting from landsliding in A1 caldera. The orientations of A1 and A2 are also emphasized on the first vertical derivative. A third anomaly, that was hardly discernible on the

112

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Fig. 7. First vertical derivative of Cappadocia Bouguer anomaly. Contour interval is 2 mGalrkm. A1 to A9 refer to main anomalies. Symbols Žq. mark positive anomalies that outline N458E positive lineaments. White dashed lines approximately outline the margins of the caldera complexes. Coordinates are in UTM Ž36th zone..

Bouguer map becomes visible to the SSW. Another feature refined is the general shape of the entire depression. While it appears sub-circular on the

Bouguer anomaly map, the first vertical derivative map reveals a more complex shape due to the N508E–N608E-trending linear northern and southern

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

boundaries and the N108W-trending linear eastern boundary. This suggests that the depression margins were strongly influenced by regional tectonics. Thus, the gravity data interpretation strongly suggest the presence of a complex shallow depression, comprising of at least two main structures in the Nevsehir–Acigol–Cardak area. We propose that ¨ these structures are calderas formed in response to the eruption of the Kavak and Zelve ignimbrites, but the available geological data do not allow us to relate each ignimbrite to a specific caldera. We refer here to these calderas as the Nevsehir–Acigol ¨ caldera complex. The minimum volume of the complex estimated from the models is about 130 km3. Although this is less than the DRE 240 km3 estimated for the Kavak and Zelve ignimbrites Žvalue includes correction for coignimbrite ash volume., the discrepancy can be easily attributed to uncertainties in both the ignimbrite and calderas volumes and to uncertainties in the densities of caldera infill and surrounding rocks. Surface evidence for these structures is not easily detectable. However some associated features may be recognized a posteriori in the field, on DEM and on SPOT image ŽFigs. 8, 9 and 10.. Ž1. East of anomaly A1, an arcuate feature ŽF1. is visible on both satellite images and the DEM ŽFigs. 8, 9a and 10.. This may represent erosional widening of the caldera wall. Ž2. Fine, and sometimes silicified volcanoclastic and lacustrine deposits found in the Cardak depression, overlaying the Kavak ignimbrites, may represent the infilling of a caldera depression. They are covered by Miocene andesitic lavas ŽGore lava flow, Pasquare, 1968; Batum, 1978a,b. and by subsequent .. ignimbrites ŽCemilkoy, ¨ Gordeles ¨ Ž3. Several normal faults ŽF2. with downthrow to the SW are seen in the SW part of the Oylu Dag massif Žsee Fig. 3 for localities. on both the DEM and satellite images ŽFigs. 8, 9a and 10.. These may represent ring fractures. Ž4. A normal fault ŽF3., with downthrow to the NW, from Nevsehir to Acigol, ¨ uncovers fine lacustine sediments and a succession of andesitic flows ŽGore lavas, Batum, 1978a,b. which we interpret as the top of a caldera fill sequence ŽFigs. 8, 9a and 10.. Ž5. Several lava domes ŽF4. are located along the

113

E and SE boundary of the negative anomaly ŽFigs. 8, 9a and 10.. They could represent post-collapse marginal ring fracture volcanism. Ž6. The north flank of Erdas massif is cut by a 10-km-long curvilinear fault recognized in the field ŽF5. and which appears, on the SPOT image, to be duplicated in the south by at least one other fault ŽF6. ŽFigs. 8, 9a and 10.. At the NE of its eastern tip, an arcuate topographic feature ŽF7. is observable on the DEM ŽFigs. 8 and 10. and on the satellite image ŽFig. 9a.. These features coincide with the limits of the A2 anomaly on the first vertical derivative map ŽFig. 7.. The rampart at the north of the Erdas massif is well-preserved, a fact which leads to the question of whether this fault is related to a Miocene caldera, or associated to the Quaternary eruption of the Acigol ¨ tuff whose source is thought to lie in this area ŽLe Pennec et al., 1991; Druitt et al., 1995.. The products from the Quaternary eruption coat the eroded surface of the fault scarp and therefore the fault predates Quaternary eruptions ŽDruitt et al., 1995.. It is most likely to be associated with the collapse of a Miocene caldera, but could nevertheless have been rejuvenated during the Quaternary eruption. 4.2. The Derinkuyu caldera complex Geological evidence ŽLe Pennec et al., 1994. suggests that the source areas of the Sarimaden, Cemilkoy, and Kizilkaya ignimbrites con¨ Gordeles ¨ centrate in an area located between the Erdas massif to the north and the Melendiz massif to the south ŽFig. 4.. This zone is presently occupied by the Derinkuyu Quaternary basin ŽFigs. 2 and 3.. Gravity, satellite and topographic data, as well as additional field observations, serve to identify a large buried caldera complex, referred to as the Derinkuyu caldera complex. A structure interpreted as a resurgent dome occupies the center of the Derinkuyu caldera complex. The Sarimaden ignimbrite covers an area greater than 3900 km2 , mostly to the west, northeast, and east beyond the Derinkuyu basin. With the scattering of present-day outcrops it is difficult to locate vents precisely. However, thickness of deposit, pumice and lithic sizes, and flow indicators suggest a source located within the Derinkuyu basin ŽLe Pennec et al., 1994..

114

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Fig. 8. Illuminated DEM of central Nevsehir Plateau area. Azimuth of the sun is artificially located to the north. Coordinates are in UTM Ž36th zone.. Resolution is 50 m.

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

The Cemilkoy ¨ ignimbrite is one of the most extensive ignimbrites of the Nevsehir Plateau Ž8600 km2 .. It extends to the N and NE more than 60 km beyond the plateau margins. Thickness variations Žmaximum) 100 m in Selime. and pumice and lithic sizes clearly suggest a source area in the Derinkuyu basin. Because of the lack of exposure within the basin, we are not able, on the basis of field evidence alone, to delineate a more precise source location. The Gordeles ignimbrite covers at least 3600 ¨ km2 , mainly to the W, E and SE of the Nevsehir Plateau. Its mean thickness is 10 to 20 m, with a maximum of 50–100 m on the Sahin Kalesi massif. Constraints for the source of this ignimbrite are more precise than for the other ignimbrites. On the Sahin Kalesi massif, the ignimbrite shows a high content of lithic clasts and lag breccia facies, usually good indicators of vent proximity. Flow indicators show a southward direction, suggesting that the source is located close to the northern edge of the Sahin Kalesi massif ŽLe Pennec et al., 1994.. The Kizilkaya ignimbrite is the largest Cappadocian ignimbrite. It covers about 10,000 km2 and has a mean thickness of 15–20 m Žlocally thickening to greater than 70 m, i.e., Melendizsuyu canyon, see Fig. 3 for localities.. Field criteria and flow directions deduced from AMS measurements ŽLe Pennec et al., 1998. indicate a source in the center of the Derinkuyu basin, near the Quaternary Gollu ¨ Dag Acidic Complex. The most convincing evidence for a caldera complex concealed within the Derinkuyu basin is the gravity data ŽFigs. 5 and 7.. A composite negative anomaly ŽA3a–A3b. occupies a large part of the Derinkuyu basin and extends southwestward over the Sahin Kalesi Massif. This anomaly is flanked to the north by the positive anomaly A4 associated with the dense ophiolite massif. Like A2 Žsee above., A3a– A3b is strongly influenced by the presence of A4. In order to better define A3a–A3b, the map shown in Fig. 11 is corrected for a regional gradient of 1 mGalrkm, oriented N308W, which minimizes the effect of A4 on A3a–A3b. The residual anomaly indicates that A3a and A3b are two parts of a larger anomaly with the shape of a 40 = 20 km rectangle, trending N608E. It is composed of the two lows A3a and A3b of about y25 mGal separated by a relative high ŽA6. of q15 mGal. We assume that the A3

115

anomalies are caused by the low-density infill of a complex set of at least four calderas, the Derinkuyu caldera complex, that formed in response to the eruptions of the Sarimaden, Cemilkoy, and ¨ Gordeles ¨ Kizilkaya ignimbrites. The overall rectangular shape of the anomaly suggests that the boundaries of the Derinkuyu caldera complex were strongly influenced by regional tectonics. As in the case of Nevsehir– Acigol ¨ caldera complex, the first vertical derivative map leads to refinement in the shape of the Derinkuyu caldera complex ŽFig. 7.. It emphasizes the composite shape of the caldera complex, and the linearity of its margins Žtherefore the likely volcanotectonic origin.. Only a qualitative interpretation of the A3 anomaly was carried out, because the distribution and density of gravity measurements, particularly in the southern part, were considered inadequate for quantitative modeling. Structural information from field observations, DEM analysis and satellite images on the Erdas and Sahin Kalesi massifs provide additional evidence that the A3 negative anomaly is due to the presence of a caldera complex. 4.2.1. Erdas Dag massif The Erdas Dag massif is composed essentially of a sequence of northerly dipping andesitic flows. The south side of the massif is a south-facing rampart ŽF8. that cuts these lava flows ŽFigs. 8, 9a and 10.. This morphological evidence indicates that this volcanic massif had a summit located to the south and that the present-day relief is only the remnant of a much larger edifice whose central and southern parts disappeared, probably due to caldera collapse. Thus, the south rampart of Erdas massif would be a caldera wall. This interpretation is supported by structural data from analysis of the DEM ŽFig. 8. and satellite image ŽFig. 9a.. The massif exhibits an obvious pattern of concentric and radial fractures ŽF9. ŽFig. 10.. The concentric faults are normal faults with southernly downthrow indicating a general lowering of the southern part of the massif. On the SPOT image ŽFig. 9a., a large landslide block towards the south is visible ŽF10. ŽFig. 10.. This structure has been confirmed by field observations. In addition, the base of the rampart is characterized by the presence of hydrothermally altered zones and monogenic Quaternary volcanoes.

116

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

The position of the negative gravity anomaly A3a is slightly offset to the south of the southern rampart of Erdas Dag. This suggests that the position of the

main boundary fault of the caldera is also offset to the south of the present topographic wall, or that near the edge of the caldera the infilling material is

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

denser, possibly made of landslide blocks from the wall as commonly observed in other calderas ŽLipman, 1984.. The gravity high ŽA5. ŽFigs. 5, 7 and 11. NE and E of the negative anomaly A3a may mark the eastern flank of the Erdas massif buried within the basin. The top of the lava flow sequence cut by the south rampart of Erdas massif is covered by a welded ignimbrite. We have not been able to establish a clear correlation between this ignimbrite and the four main ignimbrites of the Derinkuyu basin. We suspect however, that the collapse of the Erdas massif may be related to the emission of this ignimbrite. 4.2.2. Sahin Kalesi massif The Sahin Kalesi massif, located to the east of Žsee Fig. 3 for localities., has the general Guzelyurt ¨ shape of a sub-circular dome, 15 km in diameter ŽFigs. 8, 9a and 10.. Its western part consists of a sequence of andesitic flows which are covered by the Cemilkoy, Gordeles and Kizilkaya ignimbrites at the ¨ base of the massif, and only by the Gordeles ign¨ imbrite on the flanks and summit. This ignimbrite is covered by younger andesitic flows. In the NW of the massif, these units overlie metamorphic basement Žmarble. that belongs to the Kirsehir massif. The western flanks of the massif are relatively regular whereas the eastern part is much more chaotic ŽFigs. 8, 9a and 10.. The eastern slopes consist of lacustrine and volcano-clastic sediments overlain by the Gordeles ignimbrite and andesitic lava flows. Pyro¨ clastics related to the Quaternary acidic complex of Gollu ¨ Dag coat the whole area. Analysis of the DEM and satellite images shows that the Sahin Kalesi massif has undergone intense and complex fracturing. Although regional tectonics have contributed partially to the fault pattern, the main origin of the fracturing must be local because the nearby massifs do not show similar fracture patterns.

117

The fracture pattern of the Sahin Kalesi massif and the presence of an apical graben ŽFigs. 8, 9a and 10. mimic the structural pattern obtained in experimental scale modeling of doming ŽKomuro et al., 1984; Komuro, 1987. and are strongly reminiscent of resurgent domes within calderas ŽSmith and Bailey, 1968.. Indeed, several arguments and observations support the interpretation of the Sahin Kalesi massif as being a resurgent dome: Ž1. Lacustrine sediments at the base of the northern part of the massif are tilted outward. Similarly, the Gordeles ignimbrite dips outward in several ¨ places beyond an equilibrium slope, reaching its maximum elevation Ž1900 m with a thicknessf 100 m. at the center of the massif. Ž2. Hot springs, gaseous emissions and intense hydrothermal alteration are present. On the NW flank, the Gordeles ignimbrite is altered to kaolinite. The ¨ hydrothermal activity, and the presence of Quaternary cinder cones and rhyolitic domes in the northern and central parts of the massif argue for the presence of a still partially hot, underlying magma reservoir. Ž3. Finally, the positive gravity anomaly A6 ŽFigs. 5 and 11. inside the A3 anomaly coincides with the Sahin Kalesi massif and may represent uplifted metamorphic basement. In this framework, the metamorphic rocks Žmarble. exposed in the NW part of the massif may be a fragment of the upheaved caldera floor. The measured density of 2.8 for these rocks is compatible with the presence of a relative gravity high. Some constraints can be made on the relative age of the Sahin Kalesi resurgent dome: Ž1. The resurgence postdates the eruption of the Gordeles ignimbrite which has been involved in the ¨ uplift but predates eruption of the Kizilkaya ignimbrite which was deflected around the Sahin Kalesi massif ŽLe Pennec et al., 1994.. A reasonable hy-

Fig. 9. Ža. Mosaic image of two SPOT sub-scenes ŽKJ 113-273 and 114-272. focused on the Quaternary basic volcanic field of central Nevsehir Plateau area, Cappadocia. Main faults are pointed out by white arrows. DF, Derinkuyu Fault; EDM, Erdas Dag Massif; SKM, Sahin Kalesi Massif; DB, Derinkuyu Basin; CD, Ciftlik Depression. Žb. Distribution of monogenic vents Žmostly strombolian cones and some basic maars. within rectangle outlined in white on Ža.. Most of them lie on ; NS lineaments Ždashed lines. that we interpret as tension gashes in a potential sinistral reidel shear system. A N408E gravity fabric on the first vertical derivative of the Bouguer anomaly Žshaded areas. strongly supports the shear system hypothesis.

118

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Fig. 10. Interpretation of the central Nevsehir Plateau area from DEM and SPOT images, showing the main structural features related to caldera complexes. Ž1. Quaternary strombolian centers; Ž2. craters; Ž3. Quaternary acidic centers; Ž4. faults; Ž5. normal faults; Ž6. hydrothermal activity occurrences Žhydrothermal alteration, hot springs, gaseous emissions.; Ž7. lag breccia facies; Ž8. Bouguer anomaly contours ŽmGal.; Ž9. arcuate features; Ž10. inferred margins of the caldera complexes ŽNACC, Nevsehir–Acigol ¨ caldera complex, DCC, Derinkuyu caldera complex.; Ž11. topographic contours; Ž12. Quaternary lacustrine deposits; Ž13. Erdas Dag Massif and Sahin Kalesi Massif; Ž14. uplifted metamorphic basement. Coordinates are in UTM Ž36th zone..

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

119

Fig. 11. Bouguer anomaly map of Derinkuyu caldera complex area, corrected for regional gradient by subtracting a 1 mGalrkm gradient in the NW–SE direction. Contour interval is 2.5 mGal. The Erdas Dag massif is outlined. A3 to A6 refer to the main anomalies. Coordinates are in UTM Ž36th zone..

pothesis, although not unquestionably established, is that the resurgence affects a caldera formed by the emission of the Gordeles ignimbrite. ¨ Ž2. The relatively thick Ž) 100 m. sequence of lacustrine sediments beneath the Gordeles ignimbrite ¨ shows the pre-existence of a depression in this area. We can speculate a pre-existing caldera associated with the eruption of the Sarimaden or Cemilkoy ¨ ignimbrites. This caldera, at least partly at the origin of the A3b anomaly, would have been filled up by the Gordeles ignimbrite. ¨ Ž3. Post-caldera volcanism, in the form of andesitic flows on the Sahin Kalesi massif and by monogenic centers spread over the massif, is associated with resurgence.

The relatively modest thickness of the Gordeles ¨ ignimbrite on the Sahin Kalesi massif Žalthough the maximum values are observed here., poses a problem if we assume that the Sahin Kalesi massif corresponds to a resurgence of the caldera associated with that eruption. Indeed, it is expected that the thickness of the ignimbrite within the caldera would be greater than the observed 100 m or so. Nevertheless, this could be explained if the collapse of the caldera had occurred predominantly after the eruption as it was reported in the case of northern part of Yellowstone caldera ŽLipman, 1984., or if the resurgence had occurred outsideror on the margin of the collapsed area Žremember that the vents of the Gordeles ignimbrite are believed to lie in the north ¨

120

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

of the Sahin Kalesi massif.. Such asymmetrical resurgence was observed for the San Luis caldera in the San Juan field ŽSteven and Lipman, 1976.. In summary, the geological ŽLe Pennec et al., 1994. and geophysical Žthis paper. observations indicate that the eruptions of the Sarimaden, Cemilkoy, ¨ Gordeles and Kizilkaya ignimbrites resulted in the ¨ formation of the Derinkuyu caldera complex. This corresponds to an extensive negative gravity anomaly and its boundaries seem to have been controlled by regional tectonics. The complex is now almost entirely buried, the only visible surface feature being the south rampart of the Erdas massif and the Sahin Kalesi massif which is interpreted here as a resurgent dome possibly associated with the Gordeles caldera. ¨

5. Relationships between tectonics and volcanism The tectonic setting of Central Anatolia remains controversial. Two main models are presently favored. The first ŽPasquare et al., 1988; Toprak and Goncuoglu, 1993. links central Anatolian tectonic ¨ ¨ depressions to a transtensive environment derived from a N–S compressional ŽE–W extensional. regime, active since the Oligo-Eocene ŽArikan, 1975; Scott, 1981; Gorur ¨ ¨ et al., 1984; Jackson and McKenzie, 1988.. The second is the somewhat controversial rifting and spreading model of Pasquare and Ferrari Ž1993. and Borgia et al. Ž1994.. This suggests that the central part of the Nevsehir Plateau is occupied by a 100-km-long N–S rift and that the margins of the plateau Žthe Tuz Golu ¨ ¨ Fault and Sultan Sazligi Basin western border fault ŽFig. 2.. are thrust faults, generated by intrusive pressure from a large batholith underlying the central rift. Our work enables us to refine the interpretation of the regional tectonic setting and of the relationship between tectonism and volcanism in Cappadocia. It is based on the gravity, satellite and DEM data that allow us both to detect new tectonic features and to refine extension and geometry of previously identified faults. We used also microstructural measurements ŽLyberis et al., 1994. and the new interpretation of eastern Anatolia tectonism proposed by Chorowicz et al. Ž1994.. Our interpretation agrees with the ‘transtensive’ model. Our observations do

not provide evidence of thrusting along the Tuz Golu ¨¨ Fault and the Sultan Sazligi Basin western border fault. 5.1. Regional tectonic setting The main lineations and faults deduced from the analysis of DEM ŽFig. 12., ERS1, and SPOT images are shown in Fig. 13. They are divided into two major fault systems. 5.1.1. Ecemis fault system The Nevsehir plateau is cut to the east by the sinistral, strike-slip Ecemis Fault ŽFigs. 12 and 13.. This fault is regarded as having been active since the Eocene ŽScott, 1981; Sengor ¨ and Yilmaz, 1981.. It has a N158E orientation in the Taurus range where it constitutes a major topographic feature. To the north it vanishes in the Sultan Saz basin. However, we believe that it continues to the north with the swarm of NE–SW-oriented faults which are observed from Kayseri to Sivas. Within the Nevsehir plateau itself, several faults parallel to the Ecemis Fault are also evident ŽFigs. 12 and 13.. We refer to this set of faults as the Ecemis Fault System. As suggested by Jackson and McKenzie Ž1988. to explain similar faults in central Anatolia, Ecemis Fault System could be: Ž1. either the record of an older position of the East Anatolian Fault ŽFig. 1., changing from thrusting to strike-slip while the East Anatolian Fault position migrated eastward; or Ž2. a presently active equivalent of the East Anatolian Fault–Karasu Fault–Dead Sea Fault System, contributing to westward extrusion of the Anatolian plate that could no longer be considered as a rigid block. 5.1.2. Tuz Golu ¨ ¨ faults system The N408W Tuz Golu ¨ ¨ fault is one of the major structures in Central Anatolia. It forms the eastern margin of the Tuz Golu ¨ ¨ Quaternary basin. The offsets of features identified on satellite images, as well as by field observations, characterize this structure as a normal fault with a dextral strike-slip component. Arikan Ž1975. and Gorur ¨ ¨ et al. Ž1984. estimate that it has been active since at least the Oligocene. It affects the most recent products of the Quaternary Hasan

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

121

Fig. 12. Illuminated digital elevation model of central Anatolia. TGB, Tuz Golu ¨ ¨ Basin; SSB, Sultan Saz Basin. Box delimits Fig. 13. Coordinates are in UTM Ž36th zone.. Resolution is 500 m.

122

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Dag volcano ŽAydar, 1992; Toprak and Goncuoglu, ¨ ¨ 1993.. On DEM ŽFig. 12. the Tuz Golu ¨ ¨ fault appears as the most prominent lineament of a set of parallel faults that we refer to here as the Tuz Golu ¨ ¨ Fault System. An earthquake occurred along one of these faults in 1938 ŽKirsehir event in Jackson and McKenzie, 1988., indicating that this system is still active. The fault plane solution revealed a right-lateral

motion trending NW–SE ŽJackson and McKenzie, 1988.. Within the Nevsehir plateau itself, the Melendiz Suyu Canyon is superimposed on a fault of the Tuz Golu ¨ ¨ Fault System ŽFigs. 3 and 13.. It is regarded as a normal dextral fault by Toprak and Ž1993.. The Derinkuyu fault ŽFigs. 2, 9a Goncuoglu ¨ ¨ and 13., which limits the Derinkuyu basin to the east, is also a right-lateral normal fault, oriented N108W, cutting Quaternary sediments. The Yesil-

Fig. 13. Structural interpretation of DEM and ERS1 images. DCC, Derinkuyu caldera complex; NACC, Nevsehir–Acigol ¨ caldera complex; TGB, Tuz Golu Graben; DVF, Damsa Valley Fault; YF, Yesilhisar ¨ ¨ Basin; SSB, Sultan Saz Basin; DB, Derinkuyu Basin; GG, Gumuskent ¨ ¨ Fault; MSCF, Melendiz Suyu Canyon Fault; Ž1. faults with a dominant strike slip component; Ž2. faults; Ž3. inferred faults; Ž4. faults with a dominant normal component Žticks on the downthrow side.; Ž5. inferred rims of caldera complexes and other large volcano tectonic collapses; Ž6. stratovolcanoes; Ž7. Quaternary alluvial deposits. Hexagons mark the main Tuz Golu ¨ ¨ Fault System faults; Triangles mark the main Ecemis Fault System faults. Coordinates are in UTM Ž36th zone..

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

hisar fault and the Damsa Valley fault also have the same orientation ŽFigs. 12 and 13.. We consider that these faults are parts of the Tuz Golu ¨ ¨ Fault System. The Tuz Golu ¨ ¨ Fault System slices the Kirsehir massif and Nevsehir Plateau into several SW tilted blocks ŽFigs. 12 and 13.. We believe that the faults

123

of the Tuz Golu ¨ ¨ Fault System are listric with a dextral component leading to a progressive downthrow from the Kirsehir massif–Nevsehir Plateau to the Tuz Golu ¨ ¨ Basin. We suggest that the Tuz Golu ¨¨ Fault System originated in response to a general E–W to ENE–WSW extension, possibly due to the

Fig. 14. Interpreted evolution of the Cappadocian caldera complexes. Ž1. Opening of a pull-apart graben at the intersection of two conjugate strike-slip faults Žwith locally normal component.. Dotted area represent deposit in the subsiding zone. Ž2. Volcanism begins in the distensive area of the pull-apart where magma can ascent more easily. Ž3. Increase of volcanic activity in the pull-apart produces a large stratovolcano. Ž4. Collapse of a first caldera strongly influenced by the faults. Ž5. Subsequent tectonic activity contributes to deform the caldera and to dismember the stratovolcano while a new sequence of volcanism begins. Ž6. Building of a new stratovolcano concurrently with the development of new strike-slip faults. Ž7. Collapse of a second caldera influenced by the faults and partially overlapped with the first caldera. Ž8. Ultimate stage of the evolution. Subsequent tectonic activity result in growing a new pull-apart graben and in deforming and dismembering volcanoes and caldera. The result is a large and complex depression similar to the cappadocian caldera complexes.

124

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

relaxation of the Anatolian continental lithosphere upon moving outside of the compressive east Anatolian zone. 5.2. NeÕsehir–Acigol ¨ and Derinkuyu caldera complexes The Bouguer anomaly map shows a large negative anomaly ŽA7, Fig. 5. in the northern part of Nevsehir plateau, overlapping with the anomaly A1. The NW part of A7 is unknown due to a lack of data. The northern border of A7 can be partially related to a structure identified on the ERS1 image. It corresponds, to a normal fault, north of the Kizilirmak river, previously identified and called the Gumuskent fault ŽMaden Tetkik ve Arama Genel ¨ ¨ Mudurlugu, ¨ ¨ ¨ ¨ 1992.. The gravity data establish that the Gumuskent fault continues to the SE, up to Kizilir¨ ¨ mak river where it bends south, merging with the Derinkuyu fault ŽFigs. 5, 7 and 13.. This confirms that all the N408W and N108W faults belong to the same system. There is no obvious morphologic feature in the present-day surface to be related to the SW border of A7. The north and east borders of A7 have high gravity gradients while the SW border has a relatively low gradient. This leads us to interpret the A7 anomaly as the signature of a half-graben bordered to the north and to the east by the Gumuskent–Derinkuyu listric fault. We refer to it as ¨ ¨ the Gumuskent graben. ¨ ¨ In addition, a subtle but indisputable fabric emerges on the first vertical derivative of the Bouguer anomaly map. It consists of a banded pattern formed by the alternation of N508E–N608E positive and negative lineaments in the central part of the Nevsehir plateau. The most prominent highs of the positive lineaments are underlined by q symbols in Fig. 7. The lineaments lie along a SW continuation of the northern segment of Ecemis Fault System from Kayseri to Sivas. Several positive lineaments coincide with outcrops of ultrabasic basement that seem to be uplifted relative to the neighboring formations. We believe that all these lineaments are the signature of a now concealed sinistral shear zone of the Ecemis Fault System ŽFig. 12.. Alternation of positive and negative trends may be the gravity expression of a succession of en echelon pull-apart grabens. This interpretation is in agreement with the riedel shear model of Druitt et al. Ž1995. for the distribution

pattern of late-Quaternary scoria cones of the Derinkuyu area ŽFig. 9b.. The negative anomaly associated with the Nevsehir–Acigol ¨ caldera complex ŽFig. 5 A1, A2. merges with the larger A7 negative anomaly. The orientation of its NW and SE margins are clearly influenced by the N508E–N608E system described above ŽFig. 7.. Therefore, the Nevsehir–Acigol ¨ caldera complex is probably located at the intersection of the Gumuskent ¨ ¨ half-graben and at least one pull-apart graben connected to Ecemis Fault System ŽFig. 14.. The polygonal shape of Derinkuyu caldera complex documents the role of the regional tectonic system in the formation of this caldera complex. The collapses associated with Derinkuyu caldera complex seem to have occurred mostly along regional faults. The northern and southern margins corresponds to N508E–N608E lineaments and the east and west margins are roughly parallel to the Derinkuyu fault ŽFigs. 7 and 11.. Therefore the two caldera complexes ŽNevsehir– Acigol ¨ caldera complex and Derinkuyu caldera complex. appear to be superimposed upon transtensive grabens at the junction of two conjugate fault systems, the Tuz Golu ¨ ¨ Fault System and the Ecemis Fault System. Such grabens are well-known to be preferential zones for magma ascent through the crust. Moreover, we believe that the calderas forming Nevsehir–Acigol ¨ caldera complex and Derinkuyu caldera complex probably do not conform with the ‘ideal’ elliptical model. As described for the calderas associated with the Great Sumatran Fault zone ŽBellier and Sebrier, 1994. and by many previous studies of calderas in extensional settings, we suggest that Nevsehir–Acigol ¨ caldera complex and Derinkuyu caldera complex each formed incrementally and partly contributed to the subsidence of pull-apart grabens ŽFig. 14.. This hypothesis could reasonably explain the geometric shape of the two caldera complexes.

6. Conclusions Evidence provided by gravity data, satellite images and DEM allow us to identify two major caldera complexes in Cappadocia. The Nevsehir–Acigol ¨ caldera complex is located

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

between the city of Nevsehir and the Erdas Dag stratovolcano. It is believed to be associated with the eruption of the oldest ignimbrites on the Nevsehir Plateau ŽKavak and Zelve.. The Derinkuyu caldera complex is located between the Erdas Dag massif and the Ciftlik depression. It is believed to be associated with eruptions of at least four ignimbrites: Sarimaden, Cemilkoy, and Kizilkaya. In ¨ Gordeles ¨ the two cases, interpreted gravity anomalies indicate the strong influence of regional tectonism rather than simple coalescence of elliptical collapses. Therefore we believe that the evolution model for the two caldera complexes is analogous to that of the one proposed by Bellier and Sebrier Ž1994. for the Toba and Ranau calderas. Nevsehir–Acigol ¨ caldera complex and Derinkuyu caldera complex were built concurrently with incremental growth of transtensive graben complexes. The sedimentation, erosion and tectonic activity induced by the development of these grabens have probably resulted in the burying and dismembering of the calderas. We think that the absence of caldera structures in the present day surface is mostly due to the last process. Moreover, analysis of DEM and remote sensing data provide us with new information about the nature of the Erdas Dag and Sahin Kalesi massifs. It appears that the Erdas Dag Massif is the remnant of a Miocene stratovolcano which was almost totally destroyed by at least two caldera collapses both on its northern and southern flanks. We interpret the Sahin Kalesi massif as a resurgent dome most likely associated with the eruption of the Gordeles ign¨ imbrite. Finally, this work illustrates the efficiency of multisource geophysical data in identifying hidden calderas Ži.e., gravity and DEM data. and its use in understanding the relationship between caldera location and regional tectonism Ži.e., gravity, satellite images and DEM.. We must point out, however, that this efficiency is greatly improved here by prior knowledge of the ignimbrite source areas deduced from previous field work. The location of caldera complexes proposed in this work coincide well with the location of the source areas proposed by Le Pennec et al. Ž1994.. Therefore the two approaches Ži.e., multisource geophysical data and field data. validate and reinforce each others conclusions.

125

Acknowledgements We are especially grateful to Mrs. G. Unalan and E. Sengec¸ and to the Geological Survey of Turkey ŽMaden ve Tetkik Arama. for their invaluable cooperation. This research has been supported by the French Center of Volcanological Research, the Agence Franc¸aise de l’Environement et de la Maitrise de l’Energie, the French Minister of Foreign Affair and the CNRS ‘Programme National de Teledetec´´ ´ tion Spatiale’. The final version of the paper was greatly improved by the careful reviews by P.W. Lipman and S. Bonvalot. We are also grateful to M.T. Gudmundsson, Y. Yilmaz, T. Druitt, H. Diot, M.A. Davies, P. Vincent, O. Merle, and G. Camus for their numerous and constructive comments.

References Arikan, Y., 1975. The geology and petroleum prospects of the Tuz Golu Basin. Maden Tetkik Arama Enst. ŽMiner. Res. Explor. Inst. Turk.. Bull. 85, 17–37, ŽForeign Ed... Aydar, E., 1992. Etude Volcano-structurale et Magmatologique du Strato-volcan Hasan Dagi ŽAnatolie Centrale Turquie.. Thesis, Universite´ Blaise Pascal ŽClermont-Ferrand II., 213 pp. Barberi, F., Cassano, E., La Torre, P., Sbrana, A., 1990. Structural evolution of Campi Flegrei caldera in light of volcanological and geophysical data. J. Volcanol. Geotherm. Res. 48, 33–49. Batum, I., 1978a. Geology and petrography of Acigol ¨ and Golludag ¨ ¨ volcanics at southwest of Nevsehir ŽCentral AnatoliarTurkey.. Yerbilimeri ŽPubl. Inst. Earth Sci. Hacettepe University. 4, 50–69. Batum, I., 1978b. Geochemistry and petrology of Acigol ¨ and Golludag volcanics at southwest of Nevsehir ŽCentral Anato¨ ¨ liarTurkey.. Yerbilimeri ŽPubl. Inst. Earth Sci. Hacettepe University. 4, 70–88. Bellier, O., Pramumijoyo, S., Sebrier, M., 1991. Strike-slip faulting and volcanic calderas along the Great Sumatran Fault. Sixth meeting of the European Union of Geosciences. Terra Abstracts, Vol. 3, pp. 266. Bellier, O., Sebrier, M., Pramumijoyo, S., 1991b. The great Sumatra Fault; geometry, kinematics and amount of displacement shown by satellite imagery. C. R. Acad. Sci. ŽParis. 312 Ž10., 1219–1226. Bellier, O., Sebrier, M., 1994. Relationship between tectonism and volcanism along the great Sumatran Fault Zone deduced by SPOT image analyses. Tectonophysics 233, 215–231. Besang, C., Eckhardt, F.J., Harre, W., Kreuzer, H., Mueller, P., 1977. Radiometrische Altersbestimmungen an neogenen Eruptivgesteinen der Tuerkei. Geol. Jahrb. Reihe B 25, 3–36. ¨ Bigazzi, G., Yegingil, Z., Ercan, T., Oddone, M., Ozdogan, M., 1993. New data for the chronology of Central and northern

126

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

Anatolia by fission track dating of obsidians. Bull. Volcanol. 55, 588–595. Bonvalot, S., Metaxian, J.-P., Albouy, Y., Lesage, P., Perez, O., Remy, D., 1992. Geophysical survey of the Masaya caldera ŽNicaragua.. Abstract, AGU fall meeting, San Francisco, EOS, 73 Ž43. 348. Borgia, A., Ferrari, L., Pasquare, G., 1994. Rifting and spreading of the Cappadocia volcanic plateau, Turkey. Abstract, Int. Volcanol. Congr., IAVCEI, Ankara, 1994. Carle, S.F., Goldstein, N.E., 1988. A three-dimensional structural model for the Long Valley Caldera, CA. Earth Sciences Division annual report 1987. Lawrence Berkeley Laboratory, Berkeley, CA, USA, pp. 99–103. Chorowicz, J., Deffontaines, B., Huaman-Rodrigo, D., Guillande, R., Le Guern, F., Thouret, J.-C., 1992. SPOT satellite follows up the eruption of Nevado Sabancaya Volcano Žsouthern Peru.. Remote Sens. Environ. 42, 1–20. Chorowicz, J., Luxey, P., Lyberis, N., Carvalho, J., Parrot, J.-F., Yurur, N., 1994. The Maras triple junction ¨ ¨ T., Gundogdu, ¨ Žsouthern Turkey. based on digital elevation model and satellite imagery interpretation. J. Geophys. Res. 99 ŽB10., 20225– 20242. Chorowicz, J., Luxey, P., Yurur, ¨ ¨ T., Rudant, J.-P., Gundogdu, ¨ N.M., Lyberis, N., in press. Slip motion estimation along the Ovacik fault near Erzincan ŽTurkey. using ERS1 radar image. Remote Sens. Environ. de Chabalier, J.-B., Avouac, J.-P., 1994. Kinematics of the Asal Rift ŽDjibouti. determined from deformation of Fieale volcano. Science 265, 1677–1681. Deplus, C., Bonvalot, S., Dahrin, D., Diament, M., Harhono, H., Dubois, J., 1995. Inner structure of the Krakatau volcanic complex ŽIndonesia. from gravity and bathymetry data. J. Volcanol. Geotherm. Res. 64, 23–52. de Silva, S.L., Francis, P.W., 1991. Volcanoes of the Central Andes. Springer, Berlin, 1991, 216 pp. Detourbet, C., Bellier, O., Sebrier, M., 1993. SPOT imagery of the Toba volcanic caldera and the Great Sumatran Fault, Indonesia. C. R. Acad. Sci. ŽParis. 316 Ž10., 1439–1445. Dewey, J.F., Hempton, M.R., Kidd, W.S.F., Saroglu, F., Sengor, ¨ A.M.C., 1986. Shortening of continental lithosphere; the neotectonics of eastern Anatolia, a young collision zone. In: Coward, M.P., Ries, A.C. ŽEds.., Collision Tectonics. Imp. Coll., Dep. Geol., Geol. Soc. Spec. Publ., London, UK, Vol. 19, pp. 3–36. Druitt, T.H., Brenchley, P.J., Gokten, Y.E., Francaviglia, V., ¨ 1995. Late-Quaternary rhyolitic eruptions from the Acigol ¨ Complex, central Turkey. J. Geol. Soc. ŽLondon. 152, 655– 667. Duffield, W., Heiken, G., Foley, D., McEwen, A., 1993. Oblique synoptic images, produced from digital data, display strong evidence of a ‘new’ caldera in southwestern Guatemala. J. Volcanol. Geotherm. Res. 55, 217–224. Ekingen, A., 1982. Results of geophysical prospecting in Nevsehir caldera. Abstract, Geol. Congr. Turkey. Ekingen, A., Guven, C., 1978. Jeotermik enerji aramalari Acigol ¨ ¨ ŽNevsehir. bolgesi gravite etudu. ¨ ¨ ¨ Int. Rep. Miner. Res. Explor. Inst. Turk, No. 6289.

Ercan, T., 1985. Orta Anadoludaki senozoyik volkanizmasi. Maden Tetkik ve Arama Dergisi, pp. 119–140. Froger, J.-L., 1996. Apport des donnees multi´ geophysiques ´ sources pour l’identification des caldeiras du plateau de Nevsehir, Anatolie Centrale, Turquie Classification texturale des formations volcaniques a` partir de l’imagerie satellitaire. Thesis, Universite´ Blaise Pascal, Clermont-Ferrand II, 347 pp. Froger, J.-L., Bonneville, A., Mulyadi, E., 1992. Signature gravimetrique des caldeiras et des chambres magmatiques ´ associees. ´ Exemple du volcan Tengger ŽJava Est.. C. R. Acad. Sci. ŽParis. 315, 749–755. Gorur, ¨ ¨ N., Oktay, F.Y., Seymen, I., Sengor, ¨ A.M.C., 1984. Paleotectonic evolution of the Tuzgolu ¨ ¨ basin complex, Central Turkey: sedimentary record of a Neo-Tethyan closure. Geol. Soc. Spec. Publ. ŽLondon. 17, 467–482. Gourgaud, A., Aydar, E., Lyberis, N., Gundogdu, N., 1992. ¨ Petrology of the Quaternary volcanism in central Anatolia ŽTurkey.: association of calk-alkaline and alkaline suites in a collision zone. Abstract, 29th Int. Geol. Congr., Kyoto, Vol. 2, pp. 549. Innocenti, F., Mazzuoli, R., Pasquare, G., Radicati di Brozolo, F., Villari, L., 1975. The Neogene calcalkaline volcanism of central Anatolia; geochronological data on Kayseri–Nigde area. Geol. Mag. 112 Ž4., 349–360. Innocenti, F., Mazzuoli, R., Pasquare, G., Radicati di Brozolo, F., Villari, L., 1976. Evolution of the volcanism in the area of interaction between the Arabian, Anatolian and Iranian plates ŽLake Van, eastern Turkey.. J. Volcanol. Geotherm. Res. 1, 103–112. Innocenti, F., Mazzuoli, R., Pasquare, G., Radicati-di-Brozolo, F., Villari, L., 1982a. Tertiary and Quaternary volcanism of the Erzurum–Kars area Žeastern Turkey.; geochronological data and geodynamic evolution. J. Volcanol. Geotherm. Res. 13, 223–240. Innocenti, F., Manetti, P., Mazzuoli, R., Pasquare, G., Villari, L., 1982b. Anatolia and northwestern Iran. In: Thorpe, R.S. ŽEd.., Andesites; Orogenic Andesites and Related Rocks. The Open University, Department of Earth Science, Milton Keynes, UK, pp. 327–349. Jackson, J., McKenzie, D., 1988. The relationship between plate motions and seismic moment tensors, and the rates of active deformation in the Mediterranean and Middle east. Geophys. J. R. Astron. Soc. 93, 45–73. Journel, A.G., Huijbregts, C.J., 1978. Mining Geostatistics. Academic Press, London, UK, 600 pp. Keller, J., 1983. Potassic lavas in the orogenic volcanism of the Mediterranean area. J. Volcanol. Geotherm. Res. 18, 321–335. Komuro, H., 1987. Experiments on cauldron formation: a polygonal cauldron and ring fractures. J. Volcanol. Geotherm. Res. 31, 139–149. Komuro, H., Fujita, Y., Kodama, K., 1984. Numerical and experimental models on the formation mechanism of Collapse Basins during the Green Tuff Orogenesis of Japan. Bull. Volcanol. 47, 649–666. LaFehr, T.R., 1991. An exact solution for the gravity curvature ŽBullard B. correction. Geophysics 56, 1179–1184. Le Pennec, J.-L., 1991. Recherche de sites d’interet ´ ˆ geothermique ´

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128 en Anatolie centrale ŽTurquie.: approche geologique. Intern ´ Rep Agence Franc¸aise pour la Maıtrise de l’Energie, Univerˆ site´ Blaise Pascal, 33 pp. Le Pennec, J.-L., Froger, J.-L., Olanca, K., Camus, G., 1991. La caldeira du Koca Dag ŽTurquie.: produits, intrusions et protrusions associes. ´ Bull. Volcanol. Soc. Geol. ´ de France 24, 9–13. Le Pennec, J.-L., Bourdier, J.-L., Froger, J.-L., Temel, A., Camus, G., Gourgaud, A., 1994. Neogene ignimbrites of the Nevsehir Plateau Žcentral Turkey.: stratigraphy, distribution and source constraints. J. Volcanol. Geotherm. Res. 63, 59–87. Le Pennec, J.-L., Chen, Y., Diot, H., Froger, J.-L., 1998. Interpretation of anisotropy of magnetic susceptibility fabric of ignimbrites in terms of kinematic and sedimentological mechanism: an Anatolian case study. Earth Planet Sci. Lett. 157, 105–127. Lipman, P.W., 1984. The roots of ash flow calderas in western north America; windows into the tops of granitic batholiths. Calderas and associate igneous rocks. J. Geophys. Res. 89 Ž10., 8801–8841. Lipman, P.W., 1997. Subsidence of ash-flow calderas: relation to caldera size and magma-chamber geometry. Bull. Volcanol. 59, 198–218. Longman, I.M., 1959. Formulas for computing the tidal acceleration due to the moon and the sun. J. Geophys. Res. 64, 2351–2355. Lyberis, N., Yurur, N.M., Chorowicz, ¨ ¨ T., Temel, A., Gundogdu, ¨ J., 1994. New tectonic data from Cappadocian Volcanic Province, Central Anatolia, Turkey. Abstract, Int. Volc. Cong., IAVCEI, Ankara, 1994. Maden Tetkik ve Arama Genel Mudurlugu, ¨ ¨ ¨ ¨ 1992. In: Suroglu, F., Emre, O., Kuscu, I. ŽEds.., MTA, Active Fault Map of Turkey, 1:1,000,000 Žprepared bySuroglu, F., Emre, O. and Kuscu, I... Maden ve Tetkik Arama Enstitusu, ¨ ¨ MTA, 1989. Morgan, L.A., Doherty, D.J., Leeman, W.P., 1984. Ignimbrites of the Eastern Snake River Plain: evidence for major caldera-forming eruptions. J. Geophys. Res. 89 Ž10., 8665–8678. Nettleton, L.C., 1939. Determination of density of reduction of gravimeter observation. Geophysics 4, 176–183. Olivier, R.J., Simard, R.G., 1981. Improvement of the conic prism model for terrain correction in rugged topography. Geophysics 46, 1054–1056. Pasquare, G., 1966. Outlines of the Neogene and Quaternary volcanism of Asia Minor. Accad. Naz. Lincei, Atti, Ser. 8 Rend., Cl. Sci. Fis., mat. natur. 40, 1077-1085 Žincl. Ital. sum... Pasquare, G., 1968. Geology of the Cenozoic volcanic area of central Anatolia. Atti Accad. Naz. Lincei 9, 53–204. Pasquare, G., Ferrari, L., 1993. Rifting and Spreading of Cappadocia Volcanic Plateau, Turkey. Abstract, V22F-10 Am. Geophys. Union Fall Meeting, San Francisco, 1993. Pasquare, G., Poli, S., Vezzoli, L., Zanchi, A., 1988. Continental arc volcanism and tectonic setting in central Anatolia, Turkey. Tectonophysics 146, 217–230. Pearce, J.A., Bender, J.F., De Long, S.E., Kidd, W.S.F., Low, P.J., Guner, Y., Saroglu, F., Yilmaz, Y., Moorbath, S.,

127

Mitchell, J.G., 1990. Genesis of collision volcanism in eastern Anatolia, Turkey. J. Volcanol. Geotherm. Res. 44, 189–229. Ratte, ´ J.C., Marvin, R.F., Naeser, C.W., Bikerman, M., 1984. Calderas and ash flow tuffs of the Mogollon Mountains, Southwestern New Mexico. J. Geophys. Res. 89 Ž10., 8713– 8732. Rowland, S.K., Munro, D.C., 1992. The caldera of Volcan Fernandina: a remote sensing study of its structures and recent activity. Bull. Volcanol. 55, 97–109. Rymer, H., Brown, G.C., 1986. Gravity fields and the interpretation of volcanic structures; geological discrimination and temporal evolution. J. Volcanol. Geotherm. Res. 27, 229–254. Scott, B., 1981. The Eurasian–Arabian and African continental margin from Iran to Greece. J. Geol. Soc. ŽLondon. 136, 269–282. Sengor, ¨ A.M.C., Yilmaz, Y., 1981. Tethyan evolution of Turkey; a plate tectonic approach. Tectonophysics 75, 181–241. Shumacher, R., 1994. Kizilkaya ignimbrite, an exeptional medium volume pyroclastic flow deposit from central Anatolia. Abstract, Int. Volc. Cong., IAVCEI, Ankara, 1994. Shumacher, R., Keller, J., Bayhan, H., 1990. Depositional charasteristics of ignimbrites in central Anatolia, Turkey. In: Savasc¸in, Eronat ŽEds.., Proc. Int. Earth Sci. Cong. on Aegean Regions, Vol. 2, pp. 435–449. Smith, R.L., 1979. Ash-flow magmatism. Spec. Pap. Geol. Soc. Am. 180, 5–27. Smith, R., Smith, L., Bailey, R.A., 1968. Resurgent cauldrons. Mem. Geol. Soc. Am. 116, 613–662. Steven, T.A., Lipman, P.W., 1976. Calderas of the San Juan volcanic field, southwestern Colorado. U.S. Geol. Surv. Prof. Pap. 958, 35. Steven, T.A., Rowley, P.D., Cunningham, C.G., 1984. Calderas of the Marysvale volcanic field, West Central Utah. J. Geophys. Res. 89 Ž10., 8751–8764. Temel, A., 1992. Kapadokya Eksplozif Volkanizmasinin Petrolo¨ jik ve Jeokimyasal Ozellikleri. Doktora Tezi, Hacettepe Univ., Ankara, 209 pp. Toprak, V., Goncuoglu, M.C., 1993. Tectonic control on the ¨ ¨ development of the Neogene–Quaternary central anatolian volcanic province, Turkey. Geol. J. 28, 357–369. Vezzoli, L., Pasquare, G., 1994. Stratigraphy and volcanology of the Cappadocia ignimbrites, Turkey. Abstract, Int. Volc. Cong., IAVCEI, Ankara, 1994. Vincent, P., 1993. Recherche de sites d’interet en ´ ˆ geothermique ´ Anatolie centrale ŽTurquie.: discussion et propositions. Intern. Rep. of Agence de l’Environement et de la Maitrise de l’Energie, Universite´ Blaise Pascal, 13 pp. Webring, M., 1985. Saki: a Fortran program for generalized linear inversion of gravity and magnetic profiles. U.S. Geol. Surv. Open-File Rep. OF 85-122, 108 pp. Williams, H., 1942. The geology of Crater Lake National Park, OR, with a reconnaissance of the Cascade Range southward to Mount Shasta. Carnegie Inst. Washington, Pub. 540, 162 pp. Williams, D.L., Finn, C., 1985. Analysis of gravity data in volcanic terrain and gravity anomalies and subvolcanic intrusions in the Cascade Range, USA, and at other selected

128

J.-L. Froger et al.r Journal of Volcanology and Geothermal Research 185 (1998) 99–128

volcanoes. In: Hinze, W.J. ŽEd.., The Utility of Regional Gravity and Magnetic Anomaly Maps. Purdue Univ., Dep. Geosci., West Lafayette, IN, USA, pp. 361–374. ¨ ¨ R., 1979. Acigol Yildirim, T., Ozgur, ¨ ŽNevsehir. kalderasi jeomorpholoji etudu. ¨ ¨ M.T.A. Report no. 6809, Ankara.

Yilmaz, Y., 1990. Comparison of young volcanic associations of western and eastern Anatolia formed under a compressional regime: a review. J. Volcanol. Geotherm. Res. 44, 69–87. Yokoyama, I., Suparto, S., 1970. A gravity survey on and around Batur caldera. Bali. Bull. Earthq. Res. Inst. 48, 317–329.