Geochemistry, Geophysics, Geosystems

A geomechanical approach for the genesis of sediment undulations on the Adriatic shelf

Authors


Abstract

[1] This study is among the first to examine the genesis of the seafloor and subsurface undulations on the Adriatic continental shelf by integrating stratigraphic information and in situ and laboratory geotechnical measurements. Interpretation of sediment behavior is based on a 32-m-long borehole crossing (1) a possible shear plane and (2) a silty clay layer at about 20 m below seafloor (mbsf) on which sediment undulations are rooted and could be interpreted as a potential weak layer succession. Our main results in terms of triggering mechanism for the observed undulations show that under an earthquake, liquefaction and/or failure of the silty-clay sediments (weak layer) leading to deformation of the upper more cohesive sediments is possible only when such a layer is buried by less than 5 m. For greater burial thicknesses, this silty clay becomes stable under the confining lithostatic pressure exerted by the overlying sediment. This work shows that the seafloor and subsurface undulations observed in the study area are most probably the result of an early deformation process of the seafloor followed by a depositional process.

1. Introduction

[2] Undulated sediment features are commonly observed on the seafloor deep below the wave base in muddy prodeltas [Correggiari et al., 2001; Lee et al., 2002; Mosher and Thomson, 2002; Cattaneo et al., 2004; Urgeles et al., 2007]. Some authors argue that those features are sediment waves induced by bottom currents and/or hyperpycnal flows [e.g., Trincardi and Normark, 1988; Bornhold and Prior, 1990; Lee et al., 2002; Berndt et al., 2006; Urgeles et al., 2007], others identify those areas as sediment deformation structures, creep and/or early signs of slope instability [e.g., Lee et al., 1981; Field and Barber, 1993; Baraza and Ercilla, 1996; Chiocci et al., 1996; Gardner et al., 1999; Correggiari et al., 2001], or as a result of a combination of deformation and depositional processes [Faugères et al., 2002; Gonthier et al., 2002; Cattaneo et al., 2004]. This debate has been especially intense in areas such as the “Humboldt Slide” offshore California [Lee et al., 2002] illustrating how little is known about the origin and evolution of these undulated sediment features since none of the proposed theories can be easily confirmed or refuted.

[3] Sediment undulations on continental shelves are interesting because they are characterized by relatively recent sedimentation and high human impacts. A correct interpretation and understanding of such features is necessary for a proper risk evaluation (in case of sediment instability) and safe offshore development. Noteworthy is that in continental shelf settings these features occur in areas off river outlets, such as prodeltas, characterized by high sedimentation rates and gas-charged sediments as, for example, the Tiber River prodelta off Rome [Trincardi and Normark, 1988; Chiocci et al., 1996], the Noeick River prodelta [Bornhold and Prior, 1990], the Gulf of Cadiz [Baraza and Ercilla, 1996; Lee and Baraza, 1999], and the Llobregat River prodelta off Barcelona [Urgeles et al., 2007]. In the Adriatic, offshore Ortona, the sediment undulations are not only accompanied by free gas in the sediment and relatively high-sedimentation rates, but are also located in an area of frequent earthquake activity that might have acted as a triggering mechanism for deformation [Correggiari et al., 2001].

[4] In many areas of the Western Mediterranean Sea, sediment undulations have been described as being rooted at the last Maximum Flooding Surface (mfs) [Díaz and Ercilla, 1993; Ercilla et al., 1995; Chiocci et al., 1996; Correggiari et al., 2001; Cattaneo et al., 2004; Urgeles et al., 2007], and this remarkable sedimentary surface, marking the onset of the present sea level highstand at the base of modern prodeltas, could represent a change in the physical properties of the sediment and a possible explanation for the origin of the undulations. The maximum flooding surface at the Adriatic site is particularly well imaged on seismic reflection profiles, has been correlated regionally and was sampled at several distal locations where prodelta deposits thin out. Here we show, for the first time and thanks to a drilling operation (PROMESS 1, June–July 2004), a continuous sedimentary record through this surface at exactly the site where sediment undulations have their maximum expression.

[5] We used undisturbed samples from the PROMESS1 borehole and cone penetration tests to evaluate the mechanical properties of the sediment within the undulated sediment section and at its base. We then modeled the effect of an earthquake of plausible magnitude for this site to reconstruct the mechanical behavior of the sedimentary units within and below the undulations.

[6] The drilling site PRAD2 was selected in the central segment of western Adriatic margin, immediately off Ortona (Figure 1), in an area characterized by seafloor and subsurface undulations. At this site, sediment deformation and/or submarine currents are likely to have a major impact on the observed sedimentary features. In situ geotechnical measurements and sediment samples were recovered from this site in order to characterize (1) possible shear planes affecting the Adriatic prodelta mud wedge and (2) the base of the mud wedge, the maximum flooding surface interpreted as a potential weak layer along which deformation of the above sedimentary units might occur [Trincardi et al., 2004].

Figure 1.

(a) Location of PRAD2 site and thickness map in TWTT (ms) of the late Holocene mud wedge [from Cattaneo et al., 2004; Trincardi et al., 2004] and (b) stratigraphy at the site PRAD2 showing seafloor and subsurface irregularities (ISMAR data).

2. Geological Context and Seafloor Features

[7] The Adriatic region represents a foreland basin formed during the Cenozoic, as a consequence of the convergence between the African and European plates [Channell et al., 1979]. The Apennine chain was built in this geodynamic context and consists of an arcuate thrust belt with convexity toward the Adria-Africa foreland, where the thrusts show different size and curvature that progressively change their orientation [Cinti et al., 2004]. Seismicity is concentrated in the central and southern Apennines. At about 200 km west of the study area, a highly seismically active central Apennine zone (Umbria) is characterized by moderate magnitude (M < 6) and rare large magnitude (M > 6) earthquakes [CPTI Working Group, 1999, 2004]. At about the same distance toward the SE (Figure 1), historical and recent seismicity is documented north of the Gargano promontory with low-magnitude (M < 4) earthquakes [Console et al., 1993]. An evaluation of the tectonic activity recorded offshore in the central Adriatic is given by Ridente and Trincardi [2006].

[8] Late Quaternary deposits on the Adriatic shelf record glacio-eustatic cycles; the most recent of these sequences formed during the last ∼20 ka, when a rapid sea level rise shifted the shoreline from the lowstand position of the Last Glacial Maximum to the modern highstand location. In the central Adriatic, several transgressive and highstand deposits have elongated depocenters along the coast as a consequence of the location of the main sediment entry points (the Po River and several smaller Apennine rivers) and a counterclockwise circulation [Cattaneo and Trincardi, 1999].

[9] During the late Holocene, a progradational mud wedge up to 35 m in thickness deposited along the western side of the Adriatic basin. This mud wedge has an overall clinoform geometry with a submerged offlap break in ∼25 m water depth and deposited on a flat surface, whereas the average slope angle of the foresets is 0.5 degree. The mud wedge is composed of a basal unit (∼1 m thick) overlain by three sigmoidal prograding units; the base of the mud wedge is a regional downlap surface that represents the time of maximum sea level highstand dated circa 5.5 ka B.P. [Correggiari et al., 2001; Cattaneo et al., 2003]. Shallow gas of biogenic provenance was sampled at several locations along the mud wedge, in association with acoustic masking of seismic reflection profiles [García-García et al., 2007].

[10] Over large areas (more than 250 km parallel to the coast, between ∼30 and 110 m water depth), the basal unit is acoustically transparent and topped by a discontinuous reflector showing lateral variations in seismic amplitude, likely because of the presence of shallow gas and fluid escape [Trincardi et al., 2004]. In these areas, seafloor and subsurface undulations affect the whole stratigraphic section of the mud wedge or selective sub-units [Correggiari et al., 2001]. Seafloor and subsurface undulations occur in water depths of ∼30 to 70 m with strikes that are sub-parallel to the regional bathymetric contours [Correggiari et al., 2001; Cattaneo et al., 2004]. These undulations are associated with small-scale mud reliefs in water depths of 70 to 110 m, with preferred crest orientations that are perpendicular to regional contours [Marsset et al., 2003].

3. Geotechnical Measurements and Methods

3.1. In Situ Testing: CPTU

[11] In the Cone Penetration Test (CPTU) a cone, with an instrumented sleeve above it, is pushed through a series of rods into the sediment layers at a constant rate. A continuous measurement is made of the cone resistance qc, the sleeve friction fs, and the pore pressure u2 measured by means of a porous filter located immediately behind the cone (called U2 type cone). The electric cones used by Fugro on board the R/V Bavenit during the PROMESS cruise gave a continuous measurement over successive lengths of 3 m. The geometry of the used cone penetrometer with tip, sleeve and pore pressure filters follows the International Reference Test Procedure for Cone Penetration Testing [International Society of Soil Mechanics and Geotechnical Engineering, 1999].

[12] The primary objective of hole PRAD2-3 was to mechanically characterize one probable shear plane (from a series of possible shear planes) that were identified from seismic reflection data (Figure 1). For this, the soil hydro-mechanical parameters of the first 32 m of the sediment column were determined using in situ CPTU measurements (for location see Table 1). Eleven CPTU sequences around 3 m each were carried out (Figure 2 and Figure 3). Unfortunately, instrument failure resulted in invalid data for the upper 15 m and a new, deep CPTU hole (PRAD2-6) was drilled at the same site. To guarantee continuous recovery between holes PRAD2-3 and PRAD2-6, the latter was drilled to 18 mbsf. Thus, six CPTU sequences were obtained at this hole.

Figure 2.

(a) Corrected cone resistance qt versus depth below seafloor. (b) Unit sleeve friction resistance fs versus depth. (c) Pore pressure u2 versus depth (Figures 2a, 2b, 2c, and 2d from boreholes PRAD2-3 (15–32 mbsf) and PRAD2-6 (0–15 mbsf)). (d) Unit weight from γ-density compared to the unit weight determined from the water content values (PRAD2-5). (e) Compressional wave velocity Vp versus depth from PRAD2-5. The sudden decrease of the Vp below 21 mbsf is probably related to gas exsolution.

Figure 3.

Between 20.2 m and 20.9 m below seafloor, the CPTU measurements from borehole PRAD2-3 have detected the existence of a layer characterized by (a) a relatively high cone resistance, (b) a relatively high friction, and (c) a decrease of the excess pore pressure. These in situ measurements confirm what was observed in laboratory concerning the existence of a silty layer at this depth.

Table 1. Location, Water Depth, and Depth Below Seafloor of the Different Geotechnical Boreholes From the Adriatic Site and Considered in This Work
HoleLatLonWater Depth, mDepth Below Seafloor, mType
PRAD2-342°27′20.39″N14°25′54.34″E56.330.0CPTU
PRAD2-542°27′20.21″N14°25′54.08″E56.332.8cores
PRAD2-642°27′20.23″N14°25′54.32″E56.318.0CPTU
PRAD2-642°27′20.23″N14°25′54.32″E56.319.8→21.3cores

3.2. Laboratory Testing

[13] An experimental program on undisturbed marine sediments from holes PRAD2-5 and PRAD2-6 (for location see Table 1 and Table 2) was also carried out. Its specific aim was to identify the key mechanical and physical parameters of the sediments that form the undulations in the study area so as to determine whether a genesis by deformation of the sediment column is possible or not. The detailed laboratory geotechnical investigations included (1) classification tests, (2) strength tests under static and dynamic loading, and (3) consolidation/permeability tests.

Table 2. Depth Below Seafloor of the Different Samples Tested in Laboratory in the Present Work
HoleSampleDepth (Top of the Sample), mbsf
PRAD2-5S302.45
PRAD2-5S806.45
PRAD2-5S1008.05
PRAD2-5S1310.45
PRAD2-5S1814.40
PRAD2-5S1915.25
PRAD2-5S2016.05
PRAD2-5S2317.90
PRAD2-5S2620.85
PRAD2-5S2923.25
PRAD2-5S3124.85
PRAD2-5S3628.85
PRAD2-5S3830.45
PRAD2-5S4032.05
PRAD2-6S120.85

3.2.1. Index Properties

[14] Classification tests included unit weight and moisture content determinations, grain size analysis and Atterberg limit tests. The results of the classification tests are presented in Figure 4. The unit weight profile presented in Figure 4a is obtained from the GEOTEK core logging device (MSCL, http://www.geotek.co.uk) based on a gamma ray source and detector for measuring the attenuation of gamma rays through the core. The P wave velocity profile of Figure 4c was obtained from a celerimeter device allowing direct measurement of the P wave velocity by insertion of two transducers spaced by a known distance into the sediment. The Atterberg limits and plasticity index were determined using a fall cone according to the method of Feng [2001].

Figure 4.

Data from PRAD2-5 hole: (a) unit weight, (b) grain size distribution, (c) P wave velocity, (d) water content, (e) plasticity index, and (f) liquidity index versus depth.

3.2.2. Strength Tests

[15] Shear strengths determined from laboratory tests were regularly performed using the torvane, the fall cone and, the shear vane devices (UU: Unconsolidated Undrained tests). In addition to these tests, static and cyclic triaxial tests (CU: Consolidated Undrained) were carried out on undisturbed samples from holes PRAD2-5 and PRAD2-6 (Appendix A, sections A1 and A2).

3.2.2.1. Static Triaxial Tests

[16] Shear-strength parameters were measured to assess whether drained or undrained instability could be at the origin of the observed sediment undulations. Intact values of c′ (effective cohesion) and ϕ (internal friction angle) were determined from consolidated undrained (CU) triaxial shearing tests [e.g., Germaine and Ladd, 1990] made at various confining pressures (20–100 kPa) on samples from holes PRAD2-5 and PRAD2-6 (Appendix A, section A1).

3.2.2.2. Cyclic Triaxial Tests

[17] Assessing the potential for triggering or initiation of sediment liquefaction and degradation of soft clays under cyclic loading has been a problem of major concern since the early 1960s. Under the effect of an earthquake, the sediment dynamic behavior is influenced by the intensity and duration of the cyclic loading and the state of the sediment (the grain size distribution, the presence or absence of a clay fraction, the consolidation state, and the degree of saturation). Cyclic loading may lead to degradation or cyclic softening failure of soft clays [e.g., Pestana et al., 2000] and the liquefaction of sandy silty sediments [Ishihara, 1985]. Liquefaction failure over gentle slope enhances lateral spreading, ground settlement and sometimes generates sand boils [Varnes, 1978].

[18] Boulanger and Idriss [2006] have recently defined a new criteria based on the stress-strain behavior for distinguishing between silts and clays that are susceptible to liquefaction versus cyclic softening failure. Boulanger and Idriss [2006] show that the transition between liquefaction (sediments that behave more like sands) and cyclic softening failure (sediments that behave more like clays) depends strongly on the plasticity indices (PI) of the sediment. Boulanger and Idriss [2006] found that clay-like behavior (cyclic softening failure) occurs for fine-grained soils that have PI greater or equal to 7. Sediments from PRAD2 borehole (Figure 4) indicate that both liquefaction and cyclic softening failure may occur within the studied sedimentary column.

[19] In order to evaluate the liquefaction/failure potential, two primary seismic variables are required. These variables are the level of cyclic stress induced by the earthquake on a sediment layer, expressed in terms of cyclic stress ratio (CSR), and the capacity of a sediment layer to resist liquefaction and softening failure, expressed in terms of cyclic resistance ratio (CRR).

[20] Evaluation of the cyclic resistance ratio (CRR) has developed along two specific areas of research: methods based on the results of laboratory tests, and methods based on in situ tests and field observations of liquefaction behavior in past earthquakes. In laboratory testing, the number of shear stress cycles is the basis for expressing the resistance of sediment to the initiation of liquefaction and cyclic softening failure.

[21] Using the cyclic triaxial test, the Cyclic Resistance Ratio (CRR) corresponds to the cyclic stress ratio amplitude (=σd,cyc/2σ30 where σd,cyc is the cyclic deviator stress and σ30 is the effective minor principal stress at the end of consolidation). The potential for liquefaction can then be evaluated by comparing the earthquake loading (CSR) with the liquefaction resistance (CRR). The ratio between both values is the factor of safety against liquefaction.

[22] Fifteen cyclic triaxial tests were carried out at different Cyclic Stress Ratios (CSR) (Table 3) on samples from cores PRAD2-5 and PRAD2-6. Samples were isotropically consolidated to different effective confining pressures. The cyclic tests, carried out in Fugro-France's laboratory, aimed to investigate the potential that liquefaction and pore pressure buildup during cyclic loading might have in generating deformation of the prodeltaic sediments. Tests were performed on undisturbed samples from different sedimentary units: the lower sandy layer (level 3; Figure 4), the silty clay layer above the maximum flooding surface (level 2; Figure 4) and the surrounding matrix clayey sediment in order to understand stratigraphic controls in the genesis of the observed features (for more details see Appendix A, section A2). In this work, liquefaction is considered to occur for excess pore pressure equal to 90% of the initial confining stress: according to Ishihara [1993], silty sands or sandy silts containing some amount of fines may behave as liquefiable materials with an excess pore pressure values equal to 90 to 95% of the initial confining stress.

Table 3. Summary of the Cyclic Tests Carried out on Samples From the PRAD2-5 and PRAD2-6 Sites
HoleSampleDepth, mbsfσd,cyc, kPaσ30, kPaCSRCycles to Liquefaction or Cyclic Softening Failure
PRAD2-5s806.4550500.50no failure/liquefaction
PRAD2-5s1008.0530650.23no failure/liquefaction
PRAD2-5s1310.4550750.33no failure/liquefaction
PRAD2-5s2016.051001000.50116
PRAD2-5s2620.851451650.4410
PRAD2-5s2620.85801650.24no failure/liquefaction
PRAD2-5s2923.251301850.35200
PRAD2-5s3124.851502500.30no failure/liquefaction
PRAD2-5s3628.851452500.29170
PRAD2-5s3830.452453500.3510
PRAD2-5s4032.052703500.387
PRAD2-5s4032.053003500.435
PRAD2-6s120.851902500.3882
PRAD2-6s120.851752500.35100
PRAD2-6s120.851502500.3200

3.2.3. Consolidation/Permeability Tests

[23] Nine oedometer-consolidation tests were carried out at Site PRAD2 in order to characterize the consolidation state and pore pressure in the sediment column. Oedometer tests were conducted according to the ASTM D-2435 method [ASTM International, 1993]. The determination of the hydraulic conductivities and permeability coefficients were also possible using the falling head method.

4. Correlation Between Geotechnical Data and the Sedimentary Layers From the PRAD2 Site

4.1. From CPTU

[24] Figure 2a presents the corrected cone resistance qt versus depth below seafloor of holes PRAD2-3 (from 0 to 15 mbsf) and PRAD2-6 (from 15 mbsf to around 32 mbsf). The geotechnical data set obtained from the PRAD2 site appears consistent and of good quality. The qt profile (Figure 2a) shows (1) a linear increase with depth until 20.2 mbsf, (2) relatively high qt values between 20.2 m and 20.9 mbsf, (3) again a linear increase between 20.9 m and 27.5 mbsf and (4) a sudden increase in qt values at 27.5 mbsf followed by highly oscillating values to the base of the borehole. Figure 2b presents the unit sleeve friction resistance, fs, versus depth below seafloor of holes PRAD2-3 (from 0 to 15 mbsf) and PRAD2-6 (from 15 mbsf to around 32 mbsf). The fs profile in Figure 2b shows a similar trend to that observed from the qt profile. Figure 2c shows the pore pressure u2 generated by the rod penetration versus depth below seafloor for holes PRAD2-3 and PRAD2-6. The pore pressure profile shows a generally linear increase with two major reductions in the intervals between 20.2 m and 20.9 m and below 27.5 mbsf. The simultaneous variation of qt, fs and qc is a typical indicator of the presence of silty sediment at these intervals. Figure 3 shows an enlargement of the interval between 18 m and 24 m below seafloor of CPTU measurements from hole PRAD2-3. CPTU data show a layer between 20.2 mbsf and 20.9 mbsf characterized by relatively high cone resistance, high friction and low excess pore pressure. According to Robertson [1990], these features are characteristic of coarser material (silty sediment). In order to recover additional sediment from this particular layer, 1.6 m of sediment were cored at hole PRAD2-6 between 19.75 mbsf and 21.35 mbsf.

4.2. From Index Properties

[25] Figure 2d shows the unit weight profile obtained from the core-logger γ-density compared to the unit weight determined from water content values (assuming 100% saturation, PRAD2-5). There is a maximum shift of 0.7 kN/m3 between the two profiles from the seafloor to 23 mbsf, followed by a very good agreement between the two profiles. The unit weight profiles (Figure 2d) show a sudden increase at around 25.9 mbsf, which corresponds to a change in sediment type (from finer to coarser). The boundary between these two sediment types was observed at 27.5 mbsf, based on CPTU data at hole PRAD2-3, which is only 9 m distant from PRAD2-5.

[26] The increase in unit weight with depth at around 26 mbsf is supported by the grain size distribution profile presented in Figure 4b (level 3). At this level (level 3), the clay content decreases to around 16% with a silt content of around 60% and a sand content of around 24%. Two other levels can also be identified from the grain size distribution profile. The first level (level 1) at around 6 mbsf is characterized by a silt content of around 62% and a sand content of around 13%. The second level at around 20.5 mbsf contains around 53% silt and 10% sand.

[27] The water content profile presented in Figure 4c shows a linear decrease with depth over the first 26 mbsf followed by an important decrease of the water content in the sandy clay layer reaching 20% at around 28.5 mbsf. The plasticity index profile (Figure 4e) shows that the sediment from PRAD2 site is characterized by medium plasticity whereas the sediment from level 1 is just slightly plastic. The plasticity index was not determined for level 2 and level 3. The liquidity index profile showing the plastic behavior of the sediment (liquidity index values between 0 and 1) is presented in Figure 4f and compared to the analytical expression given by Lévesque [2005].

[28] Figure 2e shows the core-logger compressional wave velocity, Vp, versus depth at hole PRAD2-5. The profile shows a sudden decrease below 21.5 mbsf that is probably related to gas exsolution due to the change in pressure and temperature between in situ and laboratory conditions. The signal of the P wave was lost below 22.5 m, also probably due to this process. Above 22.5 m, the P wave velocity values vary between 1481 m/s and 1532 m/s.

[29] Two main reflectors defined from seismic profiles and identified as the transgressive surface (TS) and the maximum flooding surface (mfs) seem to match a sharp increase in qt at the PRAD2 site (Figure 5), although these increases are slightly less pronounced for the mfs than for the TS. The increase in qt at the level of the two main reflectors was accompanied by a decrease of the excess pore pressure (Figure 2 and Figure 3), which indicates coarser sediment.

Figure 5.

Correlation between geophysical data (seismic profile) and geotechnical properties.

4.3. From Shear Strength

[30] The undrained shear strength profile presented in Figure 5a and Appendix A, section 1 (Figure A1) is clearly disrupted at about 11 mbsf with an increase of about 12 kPa. At this depth, drilling penetrated from one sediment undulation to the one immediately upslope, and from the “lee” side of one undulation to the “stoss” side of the next one. A depth offset of about 1 to 2 m can be observed between the Su peak and the depth of the interface between undulations as identified from the seismic data in Figure 5. This interface did not have an expression on the CPTU measurements.

[31] Figure 6 shows the variation of CRR as a function of the cycles to liquefaction or cyclic softening failure. Comparison between the three curves in Figure 6 shows clearly three different behaviors as explained in Appendix A, section A1. Figure 6 illustrates the example of an earthquake of magnitude 6.8 at 50 km epicentral distance with 16 significant cycles according to the empirical regression equations given by Liu et al. [2001]. Figure 6 shows that failure may occur under CSR of 0.36 for level 3 and CSR of 0.46 for level 2. The surrounding clayey sediment cannot fail under an earthquake equivalent to 16 loading cycles.

Figure 6.

Potential liquefaction diagram: cyclic resistance ratio as a function of the cycles to liquefaction.

5. Discussion

5.1. Sedimentation Rate, Excess Pore Pressure, and Consolidation State

[32] In normally pressured geological formations, the sediment is permeable and the fluid can communicate through the different layers. The pore water is free to escape during consolidation; thus the fluid pressure is hydrostatic. For overpressurized layers, the permeability of the sediment is low and restricts fluid circulation. In these layers, an increase in sediment loading is transferred in part from the sediment matrix to the pore water. Thus, the pore water partially supports the overburden pressure, which prevents the pores from compressing under the weight of the overburden. The normal consolidation phenomenon is retarded and the sediment is in an underconsolidated state.

[33] Figure 7 shows the Over-Consolidation Ratio (OCR is defined as the ratio of the preconsolidation stress to the effective stress calculated from the unit weigh profile) derived from the oedometer tests indicating that the sediment is in an underconsolidated state. Two different underconsolidation states can be clearly identified from Figure 7 where the OCR is between 0.9 and 1 for the upper 4 points (between the seafloor and 8 mbsf) and underconsolidated for the lower 3 points (below 14 mbsf). The results are qualitatively supported by consolidation state estimates carried out using Skempton's equation [Skempton, 1954], which relates the consolidation state to the undrained shear strength and plasticity indices. Using this Skempton [1954] approach, the consolidation profile shows nearly normally consolidated to slightly overconsolidated sediments within the upper part of the sedimentary column, whereas underconsolidated sediments are present in the lower part of the profile. The consolidation state profile derived from Skempton's equation shows overconsolidated sediment (OCR about 2.7) at about 11 mbsf, coinciding with the location of the plane that separates one sediment undulation from another (Figure 1).

Figure 7.

Overconsolidation ratio obtained from the oedometer tests and derived from the undrained shear strength and the plasticity index (Skempton's equation).

[34] From the sedimentation rate, the porosity, the permeability and the bulk unit weight of the sediment, it is also possible to estimate theoretically the evolution of excess pore pressure and stress state over the sediment column [see, e.g., Gibson, 1958; Gordon and Flemings, 1998; Dugan and Flemings, 2000]. In order to evaluate the origin of overpressures in the Adriatic sedimentary column, we used the SeCo software [Sultan et al., 2004], which solves consolidation equation using a finite difference model (Eulerian approach).

[35] The SeCo software uses an upper moving boundary, simulating continuous sedimentation, in combination with the principle of effective stress (σv), which in porous media is the difference between the total stress (σv) and the pore fluid pressure (u). The link between the void ratio e and the vertical effective stress is considered through the compression index (see Figure 8b), whereas the permeability coefficient depends on the void ratio through the theoretical permeability curve (see Figure 8c).

Figure 8.

(a) Excess pore pressure versus depth derived from oedometer tests and calculation (SeCo software) using two different sedimentation rates (10 m ka−1 and 5 m ka−1), (b) void ratio versus vertical effective stress obtained from 9 different sediment samples, and (c) permeability versus void ratio obtained from 9 different samples and at different vertical effective stresses. The SeCO software used the theoretical compressibility curve shown in Figure 8b and the theoretical permeability curve shown in Figure 8c.

[36] The process of consolidation is directly related to the rate of excess pore pressure dissipation and the rate of sedimentation. The key equation used to evaluate the evolution of the excess pore pressure during the sedimentation process is the consolidation equation [Terzaghi and Peck, 1967]. Figure 8a shows the excess pore pressure at site PRAD2 versus depth derived from oedometer tests and that calculated using the SeCo software for two different sedimentation rates characterizing the study area (5 m ka−1 and 10 m ka−1 [from Vigliotti et al., 2008]). The results from this theoretical analysis (Figure 8) show that the sedimentation rate in the study area is too low to generate excess pore pressures as high as those measured in the consolidation tests. The excess pore pressure predicted from the observed sedimentation rate is too low to trigger alone instability and/or deformation that could account for the observed mud reliefs. However, an important mechanism that could contribute significantly to the excess pore pressure is earthquake shaking, which would also promote sediment remolding.

5.2. Earthquakes and Liquefaction Development as Possible Source of Sediment Deformation

5.2.1. Historical Seismicity

[37] The seismicity of the area is well known both from reports of strong historical earthquakes and from seismic sequences that have been recorded over the last few decades [Trincardi et al., 2004]. Focal regions on land and offshore cluster along the main tectonic structures described above, reaching magnitudes typically between 5 and 6 on the Richter scale [Tinti et al., 1995; Tinti and Armigliato, 2003]. Earthquakes located offshore of the Gargano Promontory display very high energy releases with peaks greater than Richter magnitude 6.6; the largest seismic events have an estimated return interval of 228 years [Tinti et al., 1995]. Major destructive earthquake shocks in historical times occurred in 1223, 1627 and 1731 A.D. [Postpischl, 1985]. According to Trincardi et al. [2004] and because much of the seismic activity is located offshore, the central Adriatic has been affected by large historical tsunamis, of which the 1627, 1646, 1731 A.D. and 8 December 1889 were the most devastating [Tinti et al., 1995].

[38] Figure 9a summarizes the historical seismicity in the area during the last 400 years. Only significant magnitudes (>5) have been considered and plotted in Figure 9, showing that maximum earthquake magnitude during the last 400 years is around 7. Paleoseismic studies of faults in the central Apennine region suggest that this value was not exceeded during the Holocene [Pantosti et al., 1996; Galadini and Galli, 1999]. From the earthquake magnitudes and their respective distance to the study area, the Peak Ground Acceleration (PGA) was evaluated using Idriss's method [Idriss, 1993], which proposes relationships for PGA for various magnitudes (M) in the range from 4.5 to 8.5. The relationship depends on epicentral distance for M ≥ 6 and hypocentral distance in km otherwise. In Figure 9b the calculated PGA for two different fault mechanisms (reverse and strike slip) is presented. The two maximum PGAs that the study area has been subject to during the last 400 years are 0.08 g in 1706 and 0.075 g in 1881. The calculated PGA fit with the range (0.08 g – 0.16 g) given by the Global Seismic Hazard Assessment Program (http://www.seismo.ethz.ch/gshap/adria/) for the Adriatic Sea and for a 475 year return period.

Figure 9.

(a) Distance from epicenter to the study area of the main earthquakes from the last 400 years and (b) Peak Ground Acceleration derived using the Idriss [1993] relationship. The magnitude and the PGA generated by the moderate 2001 earthquake are added to both diagrams.

5.2.2. A Representative Seismic Ground Motion Record

[39] Within the frame of the European project COSTA (2000–2003), an Ocean Floor ObServatory (OFOS) was deployed in order to investigate the effect of earthquakes on pore pressure in the upper soft sediment of the Adriatic continental shelf [Mienert et al., 2002]. With the OFOS two instruments were deployed: a 3-component Ocean Bottom Seismometer (OBS) for recording the seismic events and a PUPPI (Pop-Up Pore Pressure Instrument) to measure the pore pressure transients that might occur in the sediments in relation to a possible earthquake event [Schultheiss et al., 1985].

[40] The OFOS deployment (42°25.1701′N and 14°26.4872′E) took place about 8 km offshore the port of Ortona (Italy) in 31.4 m water depth (Figure 1). The OFOS recorded in situ data for about one year (from 16 May 2001 to 15 April 2002). However, after recovery, it turned out that the PUPPI had been damaged, and only the OBS seismic data were properly recorded. Due to the sensitivity of the OBS, it was only possible to record events within a range of 250 km from deployment location. One of the most important seismic events that occurred within that range was an earthquake on 2 July 2001 (Table 4). The magnitude and PGA generated by the earthquake of July 2001 are shown in Figure 9a and Figure 9b. The distance from the epicenter to the OFOS instrument was approximately 88 km. In this work, the PGA for a return period of 475 years and the seismogram of the event of 2 July 2001 was used to simulate the effect of earthquake shaking on the sedimentary column. It is important to mention that the use of a small earthquake event (M = 4.2), even normalized to an accurate PGA, represents a source of uncertainty in the calculation results which is mainly related to the frequency and duration of an earthquake. Biscontin and Pestana [2006, p. 105] have shown the importance of “selecting representative ground motions that include realistic combinations of distance, maximum horizontal acceleration and duration.”

Table 4. Location and Magnitude of the Event of 2 July 2001
DateTimeLatitudeLongitudeDepth, kmMagnitudeDistance, km
2 Jul 200110:04:42.0241.94615.293104.288.04

5.2.3. Effect of Earthquake Loadings on Excess Pore Pressure and Sediment Deformation

[41] Simulations of the effect of earthquake shaking on possible deformation from buildup of the pore pressure within the different sedimentary layers were carried out using Cyclic1D software (http://cyclic.ucsd.edu). Cyclic1D is a nonlinear finite element program for execution of one-dimensional site amplification and liquefaction simulations (for level as well as mildly inclined sites). Finite Elements are employed within an incremental plasticity coupled solid-fluid formulation. The liquefaction model employed in Cyclic1D is formulated within the framework of multiyield-surface plasticity (for more details, see Elgamal et al. [2002] and Yang et al. [2004]).

[42] The model parameters needed in Cyclic1D are the shear wave velocities Vs determined from the cyclic triaxial tests (Vs is a function of the shear modulus G and the unit weight [e.g., Locat and Beauséjour, 1987]), the friction angle, the Poisson's ratio, the permeability coefficient, the unit weight and the excitation signal. The only geometry parameter used in the calculation is the mean slope angle, which was taken equal to 1.5 degree. The seismogram of July 2001 was normalized to a PGA of 0.08g and used to simulate the effects of earthquake shaking on the sedimentary column at site PRAD2.

[43] The sedimentary column at the PRAD2 site was divided into 6 layers, according to the hydro-mechanical parameters obtained from the different in situ and laboratory geotechnical tests (Table 5). The initial pore pressure was considered equal to the hydrostatic pressure. Figure 10 shows the seismogram recorded by OFOS [Mienert et al., 2002] and used in the calculation (Figure 10a) and the calculation results in terms of excess pore pressure (Figure 10b), effective stress (Figure 10c) and CSR (Figure 10d) versus depth. CRR values obtained from the cyclic trixial tests carried out on sediments from level 2 and level 3 are added to Figure 10d. Sediments from level 1 (as defined in Figure 4b) were not tested under cyclic triaxial tests, however from the plasticity index and the grain size distribution, its behavior was considered similar to that of level 2. Under the earthquake seismogram recorded by OFOS [Mienert et al., 2002] and shown in Figure 10a, the most sensitive layer seems to be level 1 where the excess pore pressure at the end of the earthquake shaking raised above 90% of the lithostatic stress. The critical CRR values obtained from the triaxial cyclic tests carried out on samples from level 2 and level 3 are added to Figure 10d and show that the excess pore pressure generated within levels 2 and 3 remains too low to generate liquefaction and/or the failure of those two levels. The initial excess pore pressure considered equal to the hydrostatic pressure for the 6 sedimentary layers could be a source of uncertainty in the present calculation results. However, in the absence of accurate measurements of the in situ excess pore pressure it is not possible to evaluate the uncertainty in the modeling results. The simulation results presented in Figure 10 and the comparison between CRR and CSR values presented in Figure 10d show that for the present-day stratigraphy and for the maximum historical earthquake, only the upper silty layer (level 1) could liquefy. On the other hand, it seems that the sediment undulations are rooted at the mfs (top of level 2; Figure 1b), not at level 1. This indicates that deposition of level 1 most probably occurred after formation of the undulations took place.

Figure 10.

(a) Horizontal acceleration time history obtained from OFOS [Mienert et al., 2002] and the final profile of (b) excess pore pressure, (c) vertical effective stress, and (d) CSR. CRR values obtained from the cyclic trixial tests for level 2 and level 3 (as defined in Figure 4b) are added to Figure 10d.

Table 5. Mechanical Parameters Used in the Cyclic1D Software to Study the Effect of an Earthquake on the Sedimentary Behavior From the PRAD2 Site
LayerDepth, mTypeShear Wave Velocity, m/sFriction Angle, deg, or Undrained Shear Strength, kPaSubmerged Unit Weight, kN/m3
10.0→5.5cohesive soft80Su = 107.5
25.5→6.0cohesionless loose silt140ϕ = 308.0
36.0→19.5cohesive medium100Su = 208.5
419.5→20.5cohesionless loose silt165ϕ = 308.5
520.5→26.0cohesive medium150Su = 309.0
526.0→40.0cohesionless medium silt200ϕ = 3610.0

[44] Three additional analyses were carried out using the Cyclic1D software to define the critical depth of level 2 during deposition history at which an earthquake of similar frequency content and duration to that of July 2001 [Mienert et al., 2002], but scaled to the maximum PGA observed during the last 400 years, would be able to produce deformation of the sedimentary column above it. Figure 11 shows the simulation results in terms of normalized excess pore pressure with respect to the vertical effective stress for the three investigated depths of level 2 (Figure 11a: 10.5 mbsf, Figure 11b: 5.5 mbsf and Figure 11c: 3 mbsf). For the three calculations the ratio of the excess pore pressure to the vertical effective stress after the earthquake shaking was equal to 0.8, 0.87 and 0.94, respectively. The simulation results presented in Figure 11 show that under an earthquake similar to that presented in Figure 10a the liquefaction and/or failure of layer 2, inducing deformation of the upper clayey sediments, could only occur for as long as this layer was buried with less than 5 m of sediment. For greater burial thicknesses, this silty clay layer becomes stable because of the confining lithostatic pressure of the overlying sediment. Moreover, a classical deformation process (Mohr-Coulomb failure) related to gravitational loading cannot explain the observed undulations in the area [Berndt et al., 2006]. These observations and calculation results, show that the seafloor and subsurface undulations are most probably the result of an early deformation that has predisposed the seafloor for a subsequent sediment wave style deposition.

Figure 11.

Excess pore pressure profile for three different positions of the silty clay layer (level 2; Figure 4): (a) at 10.5 mbsf, (b) at 5.5 mbsf, and (c) at 3 mbsf.

[45] The potential liquefaction of sediments from level 2 cannot explain, however, the origin of the excess pore pressure identified from the in situ measurements and laboratory testing. An open question remains about the role of free gas identified from seismic data (Figure 1b) in generating the observed excess pore pressures.

6. Conclusion

[46] The integration of stratigraphic information (geometry, sedimentary facies, chronology), in situ geotechnical measurements (CPTU), laboratory measurements of physical and mechanical sediment properties (classification tests, oedometer/permeability, static and cyclic triaxial compression tests) allowed a rigorous analysis of the mechanical behavior of the Adriatic prodeltaic sediments and an assessment of its response to seismic ground motions. The main conclusions that can be drawn from this study are as follows:

[47] 1. The existence of a boundary at 9 mbsf between low and high undrained shear strength (Su). This limit fits well with the interface (at around 8 mbsf) between one package of undulations (belonging to a seismic unit showing high values of reflection amplitude on CHIRP sonar profiles) and an underlying package.

[48] 2. The existence at 20.5 mbsf of a silty clay layer, interpreted as the basal unit of the late Holocene mud wedge immediately above the mfs, with coarser grain size than the underlying and overlying units. This layer was identified from in situ measurements as well as laboratory testing.

[49] 3. Oedometer tests have shown that sediment from the PRAD2 site is slightly underconsolidated to normally consolidated in the upper part of the borehole (above 8 m) and highly underconsolidated in the lower part (below 14 m).

[50] 4. Cyclic triaxial tests show two different dynamic behaviors characterizing the Adriatic prodeltaic sediment: one for granular silty-clay, silty or sandy sediment (liquefaction), another for cohesive clay sediment (cyclic softening failure). From the triaxial cyclic tests and the in situ effective stress measurements it is clear that the silty/sandy sediment is the most sensitive to earthquake loading.

[51] 5. Modeling results indicate that the origin of the excess pore pressure identified from in situ measurements and laboratory testing seems unrelated to the high sedimentation rate and/or to the high seismicity in the area. Therefore, an open question remains about the role of the free gas in generating the observed excess pore pressure.

[52] 6. Calculation of the potential for liquefaction and degradation of sediments from PRAD2 site, under an earthquake similar in frequency and duration to the that of July 2001 and for a maximum PGA of 8% g, shows that sediment liquefaction within level 2 (layer above the mfs at which the undulations are rooted), and deformation of the above sedimentary column, could only be possible up to a maximum level 2 burial of 5 m. For greater burial thicknesses, level 2 silty clay becomes stable because of the confining lithostatic pressure of the overlying sediment. This work shows that in the study area the seafloor and subsurface undulations are most probably the result of an early deformation of the seafloor that has predisposed the seafloor for a subsequent sediment wave style deposition.

Appendix A

A1. Shear Strength Under Static Loading

[53] The undrained shear strength, whether measured by torvane, fall cone or vane shear strength tests, shows consistent trends (Figure A1). The torvane measurements were performed during the cruise shortly after the cores arrived aboard, whereas the fall cone and vane tests were conducted in the laboratory about 4 months after the cruise. This indicates that the cores suffered little dewatering and disturbance during the transport and storage process.

Figure A1.

Undrained shear strength Su acquired from PRAD2-5 using (a) Torvane, (b) Fall cone, and (c) Shear Vane. A limit at around 10 mbsf corresponding to an increase of the undrained shear strength was identified from the torvane and the fall cone.

[54] The undrained shear strength follows an almost linear increase profile with depth. Near the seafloor the undrained shear strength shows values of 5 to 6 kPa, whereas near the bottom of the borehole values range between 28 and 40 kPa. This linear trend is only disrupted at about 11mbsf where a sudden increase in shear strength is observed both in the Torvane and fall cone data and, in a more subdued way, in the vane tests. Such an increase divides the strength profile in an upper section (seafloor to 11 mbsf) where values increase from 5 to 13 kPa, and a lower section where undrained shear strength ranges from 25 to 40 kPa. The strength gradients are therefore similar in the upper and lower profile sections with a shift between 10 and 11 mbsf of about 12 kPa.

[55] Static triaxial tests show that sediment from level 3 is characterized by a high internal friction angle corresponding to 36 degree (Figure A2). The internal friction angle from the surrounding clayey sediment is 30 degree (Figure A2).

Figure A2.

Static triaxial tests: Stress paths in the shear stress-mean effective stress diagram.

A2. Shear Strength Under Cyclic Loading

[56] Figures A3 through A5 present three typical cyclic triaxial tests carried out on samples from the three different levels: the lower sandy layer (level 3; Figure 4), the silty clay layer above the maximum flooding surface (level 2; Figure 4) and the surrounding matrix clayey sediment in order to understand stratigraphic controls in the genesis of the observed features. For each cyclic test, three diagrams are presented showing (1) the applied cyclic shear stress versus the mean effective stress, (2) the applied cyclic shear stress versus the shear strain, and (3) the excess pore pressure generated by the cyclic loading normalized with respect to the initial effective confining pressure (σ30) as a function of the number of cycles.

Figure A3.

Undrained cyclic triaxial test: PRAD2-5–S20 (clayey layer: 16.05 mbsf).

[57] For sample S20 (Figure A3), taken from a clayey layer (Table 1), the sediment was set to an effective confining pressure of 100 kPa and the applied cyclic shear stress was equal to 100 kPa. Cyclic softening failure occurred for sample S20 (shear strain greater than 20%) after 115 uniform cycles (Figure A3). For sample S1 from hole PRAD2-6 (Figure A4), obtained from the silty clay layer above the mfs (level 2), sediment was confined under an effective stress of 250 kPa and was loaded under an applied cyclic shear stress of 190 kPa. Liquefaction (excess pore pressure greater than 90% of the initial effective confining pressure) occurred for S1 after 84 uniform cycles (Figure A4). The last example shows the cyclic tests carried out on sample S40 (Table 1), recovered from the lower sandy layer (level 3). The sample was confined under an effective confining pressure of 350 kPa and cyclically loaded under a shear stress of 270 kPa. Liquefaction occurred for sample S40 after only 6 uniform cycles (Figure A5).

Figure A4.

Undrained cyclic triaxial test: PRAD2-6–S1 (silty clay layer: 20.85 mbsf).

Figure A5.

Undrained cyclic triaxial test: PRAD2-5–S40 (sandy layer: 32.05 mbsf).

Notation
Symbol

Definition

α

effective cone section ratio.

Bq

pore pressure parameter.

c

effective cohesion.

CSR

Cyclic Stress Ratio.

CRR

Cyclic Resistance Ratio.

Cv

hydraulic diffusivity.

Δu

excess pore pressure.

e

void ratio.

Fr

normalized friction ratio.

fs

sleeve friction.

ϕ

internal friction angle.

g

gravitational acceleration.

G

shear modulus.

γw

water unit weight.

k

permeability coefficient.

OCR

Over-Consolidation Ratio.

PGA

Peak Ground Acceleration.

PI

Plasticity Index.

qc

tip resistance.

qnet

net cone resistance.

qt

corrected cone resistance.

Qt

normalized cone resistance.

Su

undrained shear strength.

σv

vertical total stress.

σv

vertical effective stress.

σ30

effective confining pressure.

σd,cyc

cyclic deviator stress.

u0

hydrostatic pressure.

u2

pore pressure measured immediately behind the cone.

ui

hydrostatic pore pressure at the borehole base.

Vp

compressional wave velocity.

Vs

shear wave velocity.

Acknowledgments

[58] This study has been possible thanks to EC project PROMESS1 (contract EVR1-CT-2002-40024) and EURODOM (contract RTN2-2001-00281). OFOS data were acquired within the EC project COSTA (EVK3-CT-1999-00006) and were made available by J. Mienert and C. Berndt. R.U. acknowledges “Ramón y Cajal” contract by the Spanish “Ministerio de Educación y Ciencia.”

Ancillary