Geo-Marine Letters

, Volume 23, Issue 3, pp 258–268

Mud and fluid migration in active mud volcanoes in Azerbaijan

Authors

    • Volcanic Basin Petroleum Research (VBPR)
    • Physics of Geological ProcessesUniversity of Oslo
  • H. Svensen
    • Physics of Geological ProcessesUniversity of Oslo
  • M. Hovland
    • Statoil
  • D. A. Banks
    • School of Earth SciencesUniversity of Leeds
  • B. Jamtveit
    • Physics of Geological ProcessesUniversity of Oslo
Original

DOI: 10.1007/s00367-003-0152-z

Cite this article as:
Planke, S., Svensen, H., Hovland, M. et al. Geo-Mar Lett (2003) 23: 258. doi:10.1007/s00367-003-0152-z

Abstract

Mud volcanic eruptions in Azerbaijan normally last for less than a few hours, and are characterized by vigorous extrusion of mud breccias, hydrocarbon gases, and waters. Recent fieldwork and mapping on four active mud volcanoes show that dormant period activity ranges from quiet to vigorous flow of mud and fluids. Geochemical analyses of expelled waters show a wide range in solute concentrations, suggesting the existence of a complex plumbing system. The mud and fluids have a deep origin, but are sometimes stored in intermediate-depth mud chambers. A mixing model between deep-seated saline waters and shallow meteoric water is proposed.

Introduction

Piercement structures, such as mud volcanoes and hydrothermal vent complexes, are common in many sedimentary basins. Hydrothermal vent complexes are numerous in sedimentary basins on the NE Atlantic margins and in the Karoo basin in South Africa (Jamtveit et al. 2003). These vent complexes were formed as a consequence of the intrusion of mafic melts in sedimentary basins in late Paleocene and mid-Jurassic times, respectively. In contrast, mud volcanoes are formed by tectonic processes, e.g., by overpressure buildup in compressional settings, or by maturation and degassing of rapidly buried organic-rich sediments (e.g., Hedberg 1974; Higgins and Saunders 1974; Sjögren 1886; Brown 1990; Milkov 2000; Kopf and Behrmann 2000; Fowler et al. 2000; Kopf 2002).

The piercement structures will clearly have an important impact on fluid migration in sedimentary basins when they are formed (e.g., Dimitrov 2002). However, it has been shown that they may also have an important long-term impact on the fluid-flow history in sedimentary basins (Svensen et al. 2003, this volume). Currently, it is difficult to incorporate such piercement structures in basin modeling theories, and they are thus commonly ignored when studying fluid flow in sedimentary basins.

An important issue regarding mud volcanoes and seeps is if the seep fluids have the same source during the dormant and eruptive phases, and the degree of mixing of deep waters (i.e., from mud reservoir level) with shallow waters. Geochemical data from mud volcanoes suggest that, even during dormant periods, water seeps are sourced from the deep mud reservoir (e.g., Dia et al. 1999). Mixing between deep and shallow waters during ascent is to be expected, and may be related to the presence of intermediate-depth mud chambers (cf. Cooper 2001). Thus, the waters expelled at mud volcanoes may represent complex mixtures of deep and shallow waters, with chemistries affected by processes like mineral dehydration, adsorption and desorption on clay minerals, precipitation and dissolution, redox reactions, and degradation of organic material (e.g., Lagunova 1976; You et al. 1993; Martin et al. 1996; Dia et al. 1999; Kopf and Deyhle 2002).

The South Caspian Basin (Fig. 1) provides a unique possibility to study mud volcanism and fluid flow in an active pierced basin. More than 400 active mud volcanoes are present in this region, both onshore and offshore (Jakubov et al. 1971; Aliyev et al. 2002). These mud volcanoes are commonly associated with hydrocarbon fields (e.g., Guliev and Feizullayev 1996; Fowler et al. 2000), and may provide important channels for petroleum migration (Katz et al. 2000; Guliyev 2002). The aim of this paper is to document the main processes of mud volcano eruptions, mud breccia emplacement, and fluid migration based on fieldwork on four active mud volcanoes in Azerbaijan and geochemical analysis of seep waters from two of these.
Fig. 1

The distributions of active mud volcanoes in Azerbaijan (Az). The numbered mud volcanoes are studied in this work. Modified from Jakubov et al. (1971) and Nadirov et al. (1997)

Mud volcanic processes and terminology

Mud volcanism shows many similarities to magmatic volcanism, and a substantial part of the terminology used to describe volcanic processes and deposits (e.g., Sigurdsson 2000) can be applied to mud volcanoes and mud volcanism. There is, however, also a separate terminology related to mud volcanoes (e.g., Hovland et al. 1997; Milkov 2000; Aliyev et al. 2002; Dimitrov 2002; Kopf 2002). Although the terminology of mud volcanoes has become more rigorous and consistent (e.g., Milkov 2000; Kopf 2002), there is still a need to make a clearer distinction between the active and dormant stages of mud volcanoes. The importance of the active/dormant distinction is crucial when estimating gas and fluid fluxes from mud volcanoes, and when determining their impact on the release of methane to the atmosphere. What are commonly referred to as active mud volcanoes in the literature are in many cases only manifestations of seep activity during the dormant period (e.g., gryphons and salses). Most of the flux estimates from mud volcanoes are calculated from dormant volcanoes (see Dimitrov 2002). The bulk of the fluid and mass is, however, released during eruptions (cf. Kopf and Deyhle 2002; Dimitrov 2002), but quantitative measurements of eruption parameters are scarce. According to Dimitrov (2002), the integrated annual fluxes of methane from mud volcanoes worldwide during eruptions are 7 times (average value) higher than the seep activity in the dormant periods. The estimated value for Azeri mud volcanoes is a 16 times higher annual flux during eruptions (Guliyiev and Feizullayev 1994, in Dimitrov 2002).

Mud volcanoes can be classified as active, extinct, or buried, and mud volcanism may represent a major land-building process. A mud volcano eruption can be explosive or effusive, and may occur both in subaerial and subaqueous environments. Hydrocarbon gases are commonly erupted. These gases may self-ignite, and up to 1-km-high fire-columns have been observed during mud volcanic eruptions in Azerbaijan (Aliyev et al. 2000). The erupted solid material is present as fall or flow deposits. Breccia flows are commonly emplaced during mud volcanic eruptions. These flows consist of angular boulders and clasts (‘xenoliths’) derived from the country rocks that are cut by the plumbing system, being embedded in a mud-dominated matrix.

Active mud volcanoes show a variable degree of dormant activity. Many are quiet, with no surface seep activity or ground deformation. However, seep activity is common on many mud volcanoes, expelling muds, liquids, and gases. The seep activity leads to the formation of scenic features such as gryphons (<3 m high, steep-sided cones extruding mud), mud cones (<10 m high cones extruding mud and rock fragments), salses (water-dominated pools with gas seeps), springs (<0.5 m small, water-dominated outlets), burning fires, scoria cones formed by heating of mud deposits during the fires (e.g., Jakubov et al. 1971; Hovland et al. 1997; Guliyev and Feizullayev 1997), and hydrocarbon deposits (Higgins and Saunders 1974; Aliyev et al. 2002). The seep activity is dominantly effusive, but small explosive events do also occur.

South Caspian Basin

The South Caspian Basin is a Tertiary back-arc basin with an up to 25–30 km thick sedimentary package (e.g., Guliyev and Feizullayev 1996; Abrams and Narimanov 1997; Devlin et al. 1999; Guliyev et al. 2002). The South Caspian Basin was a depression from the mid Jurassic, with deposition of marine sediments upon Lower Jurassic oceanic crust formed during back-arc spreading (Abrams and Narimanov 1997). Further collision in the Alpine-Himalayan zone led to regional uplift in the Miocene–Pliocene, with associated rapid deposition of 10 km of deltaic and lacustrine sediments in the South Caspian region. Sedimentation rates during the Quaternary were as high as 2.4 km/106 years (Nadirov et al. 1997), and 5–8 km of sediments has been deposited the last 5×106 years (Tagiyev et al. 1997).

The geothermal gradient in the South Caspian Basin is low (10–18 °C/km), providing hydrocarbon maturation down to great depths (14 km to the onset of the gas window; Abrams and Narimanov 1997; Nadirov et al. 1997). The typically 1–2 km thick Maykop Fm. is regarded as the major source of both the extruded mud and the petroleum (Inan et al. 1997; Fowler et al. 2000). However, mud breccias from the mud volcanoes often contain clasts from formations below the Maikop Fm., suggesting that some of the mud may have an even deeper source (Inan et al. 1997). The Maikop Fm. is locally very deeply buried—it is located between 8.5- and 11-km depth offshore Baku, and at 5.5-km depth underneath the offshore Shah Deniz structure (Fowler et al. 2000). The mud reservoirs for the extrusive volcanism may be as deep as 14 km, with intermediate mud chambers at 2–4 km depth (Cooper 2001).

The rapid Miocene–Pliocene sedimentation and burial led to increased maturation of organic material, created structural traps, and caused initiation of mud volcanism (Abrams and Narimanov 1997). Almost 300 historic small and large mud volcano eruptions are documented in Azerbaijan, occurring on 76 mud volcanoes (Aliyev et al. 2002). The mud volcanoes, and associated hydrocarbon reservoirs, are often found within large anticlines (Jakubov et al. 1971; Guliyev and Feizullayev 1996; Narimanov et al. 1998; Fowler et al. 2000; Guliyev et al. 2002).

Materials and methods

Fieldwork was performed at four active mud volcanoes—Bakhar, Dashgil, Koturdag, and Lokbatan, during October 2002 (Fig. 1). Two of the mud volcanoes, Bakhar and Dashgil, are good examples of mud volcanoes with a high seep activity in the dormant period. In contrast, the active Kotyrdag and Lokbatan mud volcanoes are quiet, and recently eruptive deposits and structures are well exposed. Finally, a satellite vent 1 km west of the Bakhar mud volcano was visited.

Detailed mapping of the Dashgil and Lokbatan mud volcanoes was done using a Garmin etrex GPS with a relative accuracy of four meters. Water samples were collected from Dashgil and Bakhar, both from salses, springs, and gryphons. Temperatures were measured in-situ with a thermocouple thermometer. Mud and water was separated by settling (to minimize release of elements adsorbed to clay minerals; cf. You et al. 1996), filtered using a 2-μm mesh, and analyzed in January 2003 at the School of Earth Sciences, Leeds, UK.

Cations and iodide were measured by ICP-MS (Agilent 7500c) on filtered water samples. Anions were measured by ion chromatography using a Dionex 600. Accuracy and precision of both methods was checked by external standards. Precision in all instances was better than 5% RSD (residual standard deviation) and accuracy better than 4%.

Eruptions

The Lokbatan mud volcano

The Lokbatan mud volcano is located in the middle of an oil and gas field just south of Baku (Figs. 1, 2 and 3). The mud volcano pierces the crest of an anticline, where several mud volcanoes are aligned along strike within 11 km (e.g., Sjögren 1891; Jakubov et al. 1971; Guliyev 2002; Kadirov et al. 2002). Twenty-two eruptions have been recorded from the Lokbatan mud volcano since the early 1800s (Aliyev et al. 2002). These eruptions are associated with seismic activity, fracture formation, ground deformation, emplacement of mud breccia flows, and ignition of hydrocarbons flowing with the mud.
Fig. 2

Simplified geological map of the upper part of the Lokbatan mud volcano (vent, crater, and associated graben; October 2002)

Fig. 3A–F

Lokbatan mud volcano eruption, deposits, and tectonic structures. The mud volcano (A) is located on the crest of an E-W-trending anticline (view from the south). Its last eruption was on 24 October 2001 (B; photo by Phil Hardy, BBC 2001). Red, burned mud breccia was present at the eruption vent (C; see person in circle for scale). Ejected mud breccia covers the horsts on the northern and southern sides of the vent. However, most of the ejected mud breccia was emplaced as a 400-m-long flow in the graben extending along the axis of the regional anticline (D). Several horst structures are present within this graben (E), which extends for more than 1 km from the vent (F). See Fig. 2 for photo locations

The most recent Lokbatan eruption occurred on 24 October 2001, lasting for about 30 min (Aliyev et al. 2002). It was associated with ground tremors, eruption of mud breccia, and extensive fires lasting for more than one year after the eruption (Figs. 2, 3). A graben extends from the vent along the crest of the anticline (Fig. 2). The vent is flanked by two horst blocks, which were both covered by ejected mud breccia material during the most recent eruption. However, most of the mud breccia was emplaced as a flow within the central part of the graben. The estimated total volume of the flow is approximately 0.0003 km3 (Aliyev et al. 2002).

Extensive faulting is seen along the summit of Lokbatan. Displacement of these faults occurred both during the 2001 eruption (Fig. 3F) and previous eruptions (Jakubov et al. 1971). The subsidence pattern is circular in the summit area, forming a ring fault with gas escape features (burned mud). The displacement on the main graben-forming faults is decreasing away from the summit area, from 10–20 m to zero meters one kilometer to the west.

The westward-trending graben collapse structure suggests the presence of an elongated, shallow mud breccia chamber within the crest of the anticline. During an eruption, mud breccia from this chamber is drained through the vent. We suggest that removal of this material caused subsidence and collapse of the roof rocks, forming the graben structure. Subsequent infilling of the graben by the erupted mud breccia increased the load on the mud chamber, causing further surface subsidence. A shallow mud chamber hypothesis is also consistent with the observation of a negative gravity anomaly above the mud volcano, suggesting the presence of light, buoyant material within the crest of the anticline (Kadirov et al. 2002).

The Kotyrdag mud volcano

The cone-shaped Koturdag mud volcano is located about 50 km south of Baku (Figs. 1 and 4; Jakubov et al. 1971). This dormant mud volcano is currently quiet. A single circular crater is found on the summit of the volcano. No faults were identified outside the crater.
Fig. 4A–D

Eruptive deposits of the 200-m-high Kotyrdag mud volcano (A). A 1-km-long channelized mud breccia flow (B) extends from the main crater down the flank of the volcano. Large boulders and clasts are common in the breccia flow (C; 1.5-m-high cliff). The eruption ended by slow extrusion of viscous, plastic mud breccia (D). Note well-defined striations and the presence of burned breccia at the edge of the extruded mud breccia sheet. Photo locations are shown by arrow on A

The geometry of Koturdag is similar to a simple, cone-shaped composite magmatic volcano of intermediate composition. The morphology of composite volcanoes is controlled by a complex interaction of aggradation (i.e., eruption and emplacement) and degradation (i.e., erosion and gravity flow) processes (e.g., Davidson and De Silva 2000). Whereas composite volcanoes have been extensively studied, no comprehensive studies of mud volcanic eruptions (detailed seismicity surveys, ground deformation measurements, or gas and fluid emission sampling) are available. The understanding of the ascent and eruption processes of composite mud volcanoes is therefore limited.

The most recent breccia flow was erupted from the summit crater and emplaced on the northeastern flank of the volcano. The flow is 1 km long and can be subdivided into two main parts—an upper channelized part on the steep flanks of the volcano, and a lower lobate part on the less steep, lower parts of the volcano. The flow is blocky in the channel, and the presence of red, oxidized mud breccia and scoria shows that escaping gases have ignited and burned during the emplacement (Fig. 4B). The lower lobe is characterized by a smoother surface and the presence of curved pressure ridges. The eruption terminated by slow extrusion of very viscous, plastic mud breccia (Fig. 4D). The emplacement rate of similar extruded mud breccia has been measured to be 2–15 m/year, lasting for several years after the main eruption event (e.g., Guliyev and Feizullayev 1997).

Dormant period

The study of the seep activity during the dormant periods may provide important insight into the mud volcanic process, and is particularly important because no mud volcanic eruptions have been monitored in detail. In addition, studies of seep gases are important because methane and other gases may contribute significantly to the global greenhouse gas budget (e.g., Hovland et al. 1997; Milkov 2000; Dimitrov 2002), and because mud volcanoes are important conduits for petroleum migration (e.g., Katz et al. 2000).

Vigorous seep activity was taking place in October 2002 on both the dormant Dashgil and Bakhar mud volcanoes (Figs. 1, 5, and 6). In contrast, the dormant Kotyrdag and Lokbatan mud volcanoes were quiet, with the exception of a small fire in the Lokbatan vent (Fig. 6D).
Fig. 5

Simplified geological map of the Dashgil mud volcano crater field (October 2002). Numbers show the sampling locations of the Dashgil water samples in Table 1

Fig. 6A–D

Seep structures and deposits on dormant mud volcanoes. A Salse A at the crater field of the Dashgil mud volcano, with the gryphon field to the west (B). C Hydrocarbons (black mud) in a gryphon at Bakhar. D Burning hydrocarbon gas in the vent at Lokbatan. The fire has been burning for more than a year since the October 2001 eruption (Figs. 2 and 3)

Seep structures on the Dashgil crater field were described by Hovland et al. (1997), and include a gryphon cluster, mud cones, salses, and sinter cones (Figs. 5 and 6). Natural oil seeps are further found on the flanks of the mud volcano. A re-mapping of the crater field in October 2002 (Fig. 5) reveals the same seep features as present in 1995, but some of the gryphons and mud cones are up to 50% larger. The only major difference is the addition of a mud cone field formed during a small eruption in May 2001 (Aliyev et al. 2002).

Active gryphons, mud cones, and springs were also found at the Bakhar mud volcano and a nearby satellite vent. The circular satellite crater is about 10 m deep with a diameter of about 50 m, and is located in an area with natural oil seeps. The crater was formed by an explosive eruption in 1998 on a gryphon field.

Waters from several of the gryphons, mud cones, salses, and springs were sampled. The salses and springs are dominated by water and gas, whereas the gryphons and mud cones are filled with intermediate- to high-viscosity fluids consisting of mud, water, gas, and oil. The temperatures of these fluids are up to 2–3 °C above the ambient temperature (Table 1).
Table 1.

Composition of water seeps from mud volcanoes (all values in ppm)

Sample

AZ02-2

A202-3

AZ02-4

AZ02-7

AZ02-8

AZ02-12

AZ02-14

AZ02-16

Seawaterc

Caspian Sea

Locality

Bakhar

Bakhar

Bakhar

Bakhar s. v.b

Dashgil

Dashgil

Dashgil

Dashgil

Type

Springa

Spring

Spring

Spring

Salse A

Salse B

Spring

Gryphon

Temp. (°C)

20.3

20.3

21.5

22.5

21.6

18.5

19.9

N

40°00′03.93″

40°00′03.97″

40°00′03.86″

39°59′53.53″

39°59′43.81″

39°59′39.56″

39°59′53.80″

39°59′46.50″

E

49°28′13.60″

49°26′13.91″

49°28′14.67″

49°27′18.71″

49°24′23.03″

49°24′30.29″

49°24′20.19″

49°24′09.68″

Li

0.2

0.2

0.15

0.25

0.2

0.25

0.2

0.26

0.18

B

49

122

201

254

<2

16

<2

125

4.5

Na

8,056

12,170

13,230

22,340

5,842

14,560

7,876

14,010

10,763

3,250

Mg

116

121

135

153

160

504

306

270

1,292

817

Al

85

87

86

88

86

66

66

89

0.01

K

40

100

119

154

90

323

175

341

399

90

Ca

170

175

155

178

189

196

198

184

411

387

Cr

101

102

102

105

102

102

102

102

5E−05

Fe

279

284

320

296

324

497

295

299

0.01

Mn

41

42

42

44

42

42

42

42

0.002

Ni

91

92

93

93

91

93

93

95

0.002

Cu

85

91

90

93

84

91

92

100

0.003

Zn

19

29

26

31

1.6

92

61

62

0.01

As

17

17

17

18

17

18

19

17

0.003

Rb

0.03

0.04

0.03

0.04

0.04

0.03

0.04

0.04

0.12

Sr

0.5

0.5

0.6

0.8

0.9

6.3

2.9

1.4

8.0

Cd

0.03

0.07

0.05

0.07

0.02

0.03

0.08

0.07

0.0001

Ba

0.2

0.2

0.2

0.4

0.3

2.2

0.2

0.1

0.03

Pb

0.003

0.02

0.014

0.015

0.004

0.02

0.03

0.015

3E−05

U

0.12

0.32

0.2

0.22

0.09

0.13

0.37

0.17

0.003

F

<1

<1

<1

<1

<1

<1

<1

<1

1.3

Cl

13,608

20,534

21,450

33,464

11,457

27,168

13,895

23,938

19,354

5,650

Br

69

108

110

193

58

134

59

150

67

9

NO3

6

19

13

<2

<2

<2

<2

252

0.5

SO4

84

247

235

161

2

57

832

2,985

2,710

3,167

I

<5

<5

<5

<5

<5

<5

<5

<5

0.06

Cl/Br

197

190

195

173

198

203

236

160

291

628

Na/Br

117

113

120

116

101

109

133

93

162

361

CI/B

278

168

107

132

>5,729

1,698

>6,948

192

278

aThe springs are water-dominated with varying degrees of mud

bBakhar satellite vent

cSeawater composition is compiled from Fontes and Matray (1993) for major elements, and Mason (1966) for trace elements. The seawater data are recalculated from mg/l values to ppm using a density of 1,022 kg/l (Fontes and Matray 1993). The (central) Caspian Sea data are from Balakishiyeva and Rashidova (1980)

The gases seeping through the mud volcanoes in Azerbaijan during the dormant periods are dominated by methane (>90%; Sokolov et al. 1968; Valyaev et al. 1985; Guliev and Feizullayev 1996; Aliyev et al. 2002). The isotopic composition of the seep methane overlaps with the isotopic composition of methane from petroleum fields in the Caucasus (Valyaev et al. 1985; Guliev and Feizullayev 1996), emphasizing the intimate relationship between petroleum formation, secondary migration, and mud volcanism.

Composition of expelled waters

The major element composition of the expelled waters is controlled by the depositional environment (e.g., marine/non-marine, presence of evaporites), diagenetic processes (e.g., Carpenter and Miller 1969; Hanor 1994; Worden 1996), temperature (e.g., Fournier and Truesdell 1973; Hanor 1994), and mixing (e.g., Dia et al. 1999). Elements like Cl, Br, I, and B may give important information both about fluid source, depth, and fluid–rock interactions, and are commonly used as tracers (e.g., Rittenhouse 1967; Harder 1970; Lagunova 1976; You et al. 1993; Worden 1996; Kopf and Deyhle 2002). A full geochemical discussion of all the elements listed in Table 1 is beyond the scope of this contribution, and would need a bigger dataset. Although our dataset is limited, we may draw important conclusions about the main fluid reservoirs that hosted the expelled waters.

The expelled waters have a composition with Na and Cl as dominant ions (Table 1). The Cl content range is from 11,000 to 33,000 ppm, from the Dashgil Salse A and the Bakhar satellite vent, respectively. There is a considerable range in concentration of major solutes (Cl, Na) within both the Dashgil and Bakhar localities (Table 1), consistent with what is documented from other mud volcanoes in Azerbaijan (Jukabov et al. 1971; Aliyev et al. 2002) and elsewhere (e.g., Lagunova 1976; Dia et al. 1999). Na and Cl correlate linearly (Fig. 7), and the Na–K–Cl relations group according to locality. Furthermore, the most K-rich waters (175–341 ppm) have the highest Zn concentrations (61–92 ppm). The sulfate content of all the waters is low, the one exception being water from a water–mud mixture from a gryphon at Dashgil (SO4 value of 3,000 ppm). This is in accordance with published data on both expelled waters and reservoir waters (Aliyev et al. 2002). Charge balance considerations do not suggest that bicarbonate is abundant in the waters, also in agreement with published data from the region (Jakubov et al. 1971).
Fig. 7

Variation of selected elements compared to seawater composition (line seawater evaporation trend) for the waters from gryphons, salses, and springs. Seawater (SW) evaporation data (straight lines from SW) are taken from Fontes and Matray (1993), and Caspian Sea (CS) compositional data are from Balakishiyeva and Rashidova (1980). Open symbols represent samples from Dashgil, and closed symbols from Bakhar

The analyzed waters are enriched in B and metals (Al, Cr, Fe, Mn, Ni, Cu, Zn) compared to seawater (Table 1), oilfield brines (Collins 1975), and expelled waters from other mud volcanoes worldwide (e.g., Dia et al. 1999; Kopf and Deyhle 2002). The concentrations of metals like Cu, Ni, and Cr, which are soluble in the reduced state, are very high (84–105 ppm), and independent of the Cl concentration in the waters (Table 1). Metals like Cu, Cr, and Ni are usually present in parts per billion in oilfield brines (Collins 1975). Intriguingly, elements like Na, Cl, Mg, K, and B vary considerably in the analyzed waters, whereas most of the metals show no concentration variations (Table 1). The standard deviation for the average of all analyses in Table 1 is less than 5% for Cr, Mn, Ni, and Cu.

Furthermore, the B content is high (50–254 ppm; Fig. 8), with the exception of two waters from Dashgil (<2 ppm). A boron concentration of 250 ppm represents a 55-fold enrichment in comparison to seawater. Expelled waters from mud volcanoes and oilfield brines from Azerbaijan are indeed associated with high B concentrations (Aliyev et al. 2002), as are mud volcanoes in the Crimea area of the Black Sea (Kerch-Taman), where values up to 915 mg/l are documented from expelled waters (Lagunova 1976).
Fig. 8

The Cl–B–Br systematics of the analyzed brines from Dashgil and Bakhar, showing that their composition can be explained by a simple two-component mixing model. Type 1 brines have Cl and B concentrations typical of oilfield brines and high Br concentrations (low Cl/Br ratios). Type 2 brines are defined by Salse A at Dashgil. These fluids have a salinity that is lower than seawater and low B concentrations. The Guneshli production water Cl concentration is from Statoil (M. Hovland, personal communication), and taken from about 1,500-m depth in the Productive Series. The compositional range of oilfield brines is from Collins (1975). Seawater (SW) evaporation data (straight lines from SW) are taken from Fontes and Matry (1993). Open symbols represent samples from Dashgil, and closed symbols from Bakhar

Fluid sources and mixing trends

Figure 8 shows the Br and Cl relations of the analyzed waters, revealing a good linear correlation (r2=0.95). The slope of the trend line for the Cl–Br correlation is less steep than that of seawater evaporation, making it unlikely that the waters acquired their salinity by surface evaporation of parent water in the vents. Although Cl/Br versus Na/Br plot close to the seawater evaporation trend, the salinities are far too low to correspond to possible surface evaporation (~146,000 ppm Cl at the onset of halite precipitation; Fontes and Matray 1993).

The Cl–Br–B relationships in Figs. 8 and 9 suggest two different mixing models: (1) mixing between two saline waters (Fig. 8), and (2) mixing between deep-seated saline waters and shallow meteoric waters (Figs. 8 and 9). The two end-member waters in the Cl–Br system (Fig. 8) were taken from the Bakhar satellite vent (high Cl, high Br: type 1) and from Dashgil Salse A (low Cl, low Br: type 2). Type 2 water has a salinity resembling production waters from the Guneshli oilfield reservoir at about 1,500-m depth below the seafloor (M. Hovland, personal communication). A mixing model between the two Cl–Br end-member fluids is also supported by the Cl–B relations (Fig. 8). The Dashgil Salse A water is low in Cl and low in B, whereas the Bakhar satellite vent is high in both Cl and B.
Fig. 9

The Cl–Br systematics of the waters can be explained by dilution of a saline fluid by a low-salinity component (meteoric water). The halite compositional range (“salt” rectangle) is from McCaffrey et al. (1987)

The high B contents of mud volcano waters in Azerbaijan suggest that the waters have a deep origin from strata affected by the smectite–illite transformation (e.g., You et al. 1996; Kopf and Deyhle 2002), where the B could originate from marine clays or organic material (e.g., Ishikawa and Nakamura 1993; Williams et al. 2001). The high Br concentration and lower Cl/Br ratio in the type 1 component do not support a Br source related to organic material, as the I concentrations are too low (cf. Worden 1996). Interestingly, the B and Li concentrations in expelled waters from a Kerch-Taman mud volcano was increased following an eruption (Lagunova 1974), suggesting that these elements are enriched in the source region. However, our data do not show any correlation between B and Li as described from the Kerch-Taman area (Lagunova 1976). In the case of B release during the smectite–illite transformation, an accompanying salinity reduction due to dehydration should occur as well (e.g., Moore and Vrolijk 1992; Fitts and Brown 1999). However, the B and Cl concentrations are positively correlated (Fig. 7), suggesting a more complex relationship between Cl–B concentration and smectite dehydration.

The low Cl/Br ratios of the waters (160–236 by mass) suggest that their salinity was not derived from dissolution of evaporates in the source strata or during ascent, which would have resulted in higher Cl/Br ratios. The narrow Cl/Br ratio range of the waters suggests that all waters represent dilution of one, high-salinity water reservoir by low-salinity waters. The low B concentrations in the type 2 water component suggests that it has undergone different processes than type 1 waters, possibly originating from shallow reservoir levels with fluids not significantly affected by clay mineral diagenesis (cf. the Guneshli water composition), or from meteoric water. The latter is more likely, considering the narrow Cl/Br range of the waters.

The metal concentrations in the analyzed waters do not vary with the chlorine concentration, and may be explained by enrichment after mixing. Both metal desorption from clay minerals (e.g., Helios Rybicka et al. 1995) and redox-controlled precipitation–dissolution are possible mechanisms that may occur in the upper part of mud volcanoes. High Fe/Mn ratios (6.7–11.8) do not suggest loss of iron due to oxidation, and can be considered as indicative of a reduced fluid (Bottrell and Yardley 1991); hence, the waters may carry transition metals soluble in the reduced state, as shown by the water analyses.

Type 2 waters have the lowest salinity and possibly the shallowest origin, and represent the most water-dominated parts of the mud volcanoes (i.e., the salses). The salses apparently have higher runoff rates than the springs and the gryphons. The waters plotting along the mixing line away from the type 2 component were all sampled from mud-dominated systems (gryphons and springs). These observations can be explained if the flux from the shallow reservoirs (meteoric water) is high in the salses, whereas the gryphons and springs represent a low-flux system originating from the deep-seated mud source.

Conclusions

A shallow mud chamber is interpreted to be present below the summit of the Lokbatan mud volcano (Fig. 10). Inflation and deflation of this elongated mud chamber prior to, and during eruptions causes the formation of a graben by subsidence and fracturing of the roof. A more deep-seated mud chamber is suggested for the composite Koturdag mud volcano.
Fig. 10

Conceptual drawing summarizing the main elements of Azeri mud volcanism and the results from the geochemistry presented in this paper. The deep type 1 waters may have originated from the clay-rich Maikop Fm., and deeper mud-rich layers. The type 1 reservoir is located below the main depth of initiation of smectite dehydration (see Ransom and Helgeson 1995), and has probably acquired the high B concentration from desorption from clays (cf. Kopf and Deyhle 2002). The presence of a mud reservoir at deeper levels than the Maikop Fm. is evident from Cretaceous–Eocene clasts in mud breccias (Inan et al. 1997). Furthermore, the main oil source rock is believed to be located deeper that the Eocene (stippled line; Inan et al. 1997). Saline water, mud, and oil migrate through the mud volcano conduit, and mud is in many cases stored in intermediate mud chambers. Low-salinity meteoric waters mix with the more saline deeper waters, and there is considerable variation in the degree of mixing between the two. High-water runoff features like salses can be explained by being fed by a shallow meteoric reservoir, whereas waters from gryphons may represent deeper fluids slowly percolating to the surface. The figure is modified from Guliyev and Feizullayev (1997), and the stratigraphic information is from Inan et al. (1997)

New data on water geochemistry, integrated with published gas data (Guliev and Feizullayev 1996; Katz et al. 2000), support the hypothesis of a complex plumbing system beneath the mud volcanoes during the dormant periods. We propose that the mud volcano seep fluids represent leakage from a deep source mixed with a component of meteoric waters. This is consistent with the recognition of the deep-seated Maikop Fm. as the main source of the extruded mud, and the existence of intermediate mud chambers at 2–4 km depth (Cooper 2001).

Dormant mud volcanoes show a large variation in activity, from mainly quiet (e.g., Koturdag, Lokbatan) to very active (e.g., Dashgil, Bakhar). We suggest that the water-dominated salses represent expulsions of shallow waters, whereas deep-seated fluids percolate slowly to the surface and are expelled as mud–water mixtures in gryphons and springs.

Acknowledgments

This work was partly financed by the Norwegian Research Council grant 120897 to B. Jamtveit/H. Svensen. We would like to thank Statoil for logistic assistance during fieldwork, and Achim Kopf for a constructive review.

Copyright information

© Springer-Verlag 2003