Thermal history interpretation of AFTA data

AFTA is based on the measurement of radiation damage trails (fission tracks) created by the spontaneous fission of ²³⁸U within the crystal lattice of apatite grains (a common detrital constituent of many sandstones). Because spontaneous fission is a form of radioactive decay, in principle the number of tracks per unit area of a polished surface of an apatite grain will be governed by uranium content and time. A fission-track age can be calculated by counting the tracks and measuring the uranium content of an apatite grain. In the absence of other factors, the fission track age provides a direct measure of the time over which tracks have accumulated. However, although fission tracks form with lengths within a narrow range (c. 16 μm), they are progressively shortened as the radiation damage is repaired at a rate which increases with temperature (annealing). This reduction in length results in a corresponding reduction in the measured fission track age. All damage is repaired above c. 120°C and no tracks are preserved. A measured fission track age therefore represents the balance between production of tracks and loss because of annealing. Green et al. (2002) provide further details of the AFTA methodology.

The annealing kinetics of fission tracks in apatite are dependent on the chlorine content of the apatite grains (Green et al. 1986; Barbarand et al. 2003), and incorporation of the variation in wt% Cl content within each sample is essential in extracting accurate thermal history information from AFTA data (Crowhurst et al. 2002; Green et al. 2002). The kinetic model employed in this study is based on a series of parallel ‘fanning Arrhenius plot’ equations similar to the model presented by Laslett et al. (1987) (except for the inclusion of a non-zero 1/T intercept), with constants that vary with wt% Cl. Some studies (e.g. Barbarand et al. 2003) have suggested that other elements also influence annealing kinetics in apatite. However, chlorine remains the dominant compositional control (Green et al. 2005) and, in practical application, inspection of the variation of fission track age and length with wt% Cl allows any grains influenced by second-order effects to be identified and isolated.

It is important to emphasize that because AFTA data are dominated by maximum palaeotemperature, this technique can reveal little information on the thermal history prior to the onset of cooling. For this reason, we do not attempt to constrain the entire thermal history of each sample. Instead, we focus on key aspects of the thermal history that control the development of the AFTA parameters (i.e. fission track age and length distribution). These aspects are the maximum palaeotemperature of each sample and the time at which cooling from that palaeotemperature began.

Extraction of thermal history information from AFTA data begins with construction of a ‘default thermal history’, derived from the burial history defined by the preserved sedimentary section and the present-day thermal gradient. This is the history that applies if the sample has never been any hotter than the present-day temperature at any time since deposition. Because of the nature of the system response (e.g. Green et al. 2002), if the AFTA data can be explained by this history, then no further thermal history information can be extracted. However, if the AFTA data show a greater degree of annealing (i.e. fission track age and/or track length reduction) than expected from the default history, then the sample must have been hotter in the past and information on the magnitude and timing of palaeothermal events can be extracted from the data.

By comparing measured AFTA parameters with values predicted from a range of possible scenarios, maximum likelihood theory (similar to that described by Gallagher (1995)) is used to define best-fit values of maximum palaeotemperature and the time at which cooling begins. The systematic variation of the timing of the onset of cooling and the peak palaeotemperature about the best-fit values enables rigorous definition of the range of conditions giving predictions that are consistent with the measured data within 95% confidence limits. Green et al. (2002), Japsen et al. (2007) and Turner et al. (2008) provide examples of this approach.

Measured fission track age and mean track length data for the six AFTA samples from Larne No. 2 are plotted against depth in Figure 4. These are compared with fission track ages and mean track lengths predicted from the default thermal history, for apatites with a range of Cl contents. The default thermal history was initially constructed using a present-day geothermal gradient of 29.9°C km⁻¹, based on a corrected temperature at the base of the borehole of 96.2°C (Geotrack unpublished report 403) and a present-day surface temperature of 10°C. However, this default history predicts a much greater degree of annealing in the deepest AFTA sample (GC403–57, depth = 2876 m) than that shown by the measured AFTA parameters. For example, the predicted age for 0.0–0.1 wt% Cl apatites in this scenario is close to zero, whereas the measured fission track age (in apatites dominated by this compositional range) is almost 100 Ma (Fig. 4a). In addition, the measured track length distribution contains a distinct population of long tracks that are not present in the predicted distribution. Both these observations suggest that present-day temperatures have been overestimated. The AFTA parameters predicted from a revised default thermal history using a geothermal gradient of 24°C km⁻¹ can better explain both the observed lack of severe age reduction and the longest tracks within the track length distribution, while the shorter tracks are the result of an earlier heating episode (Fig. 4b).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4

AFTA parameters (fission-track age and length data) plotted against sample depth and present-day temperature for Larne No. 2. The variation in stratigraphic age with depth, together with the variation in fission-track age, as predicted by the ‘default thermal history’ for a range of apatite compositions (0.0–0.1, 0.4–0.5, 0.9–1.0 and 1.5–1.6 wt% Cl) are also shown. Separate default thermal histories were calculated assuming present-day geothermal gradients of (a) 29.9°C km⁻¹ (as indicated by BHT measurements) and (b) 24°C km⁻¹, suggested by AFTA. The scenario shown in (a) predicts fission-track ages of close to zero in the deepest samples, in contrast to observations, suggesting that a present-day gradient with a value of 29.9°C km⁻¹ is too high. For reasons discussed further in the text, a value of 24°C km⁻¹ was therefore used when estimating the present-day geothermal gradient.

The observed fission track ages and mean fission track lengths in AFTA samples from Larne No. 2 are generally much younger and shorter than those predicted by the (revised) default thermal history (Fig. 4b). This indicates that all parts of the preserved section were hotter at some point in the past. The fission track ages in the two deepest AFTA samples are less than 100 Ma, which indicates that the major phase of cooling at Larne No. 2 occurred during the Mesozoic rather than the Cenozoic.

Track length distributions for the six AFTA samples are shown in Figure 5. A schematic illustration of the preferred thermal history for that sample is shown next to each histogram. All samples require either two or three episodes of heating and cooling, and maximum palaeotemperatures in all samples were reached during the Mesozoic. The fact that three episodes are not observed in all samples suggests that there are two early episodes that are relatively closely spaced in terms of temperature and time, therefore making it difficult to separate the timing of the discrete episodes. One of the two samples in which three cooling episodes are observed is a Lower Permian sandstone sample (GC403–56, present-day depth = 2019 m). This sample cooled from a maximum post-depositional palaeotemperature in excess of 110°C at some time between 240 Ma and 135 Ma. This was followed by a second phase of cooling from a temperature of 95°C to 110°C sometime between 145 Ma and 30 Ma. The final phase of cooling began during the mid-late Cenozoic (35 Ma to 5 Ma) from a temperature of between 75°C and 95°C. As indicated in Figure 5, this latest phase of cooling is observed in all samples.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5

Fission-track length distributions with accompanying thermal history solutions for each of the six AFTA samples analysed at Larne No. 2. Black rectangles indicate the stratigraphic ages of the samples.

Figure 6 shows a comparison of the cooling episodes identified following thermal history analyses of individual AFTA samples, in an attempt to define a series of synchronous events that affected the entire borehole. The overlap between the timing estimates for individual samples enables the identification of three dominant phases of cooling. The earliest episode, in which the samples cooled from their maximum post-depositional palaeotemperatures, began during the Jurassic–early Cretaceous (between 200 Ma and 135 Ma). The subsequent phases of cooling began during the Cretaceous to early Cenozoic (between 140 Ma and 60 Ma) and the mid- to late Cenozoic (between 35 Ma and 5 Ma). In a previous, comprehensive thermal history study utilizing AFTA data collected from across the Irish landmass, Green et al. (2000) identified a series of regional cooling episodes beginning during the following intervals: 300–250 Ma (late Carboniferous to Permian), 180–170 Ma (mid-Jurassic), 125–110 Ma (early Cretaceous), 65–55 Ma (early Cenozoic) and 25–15 Ma (late Cenozoic). Cooling episodes at broadly similar times were identified by Allen et al. (2002) in an independent study of thermal histories and exhumation across onshore Ireland. We suggest that the 200–135 Ma and 35–5 Ma cooling episodes identified at Larne No. 2 represent local expressions here of the mid-Jurassic (180–170 Ma) and late Cenozoic (25–15 Ma) cooling episodes described by Green et al. (2000). Following this approach, the Cretaceous–early Cenozoic (140–60 Ma) episode identified at Larne No. 2 could be a local expression of the early Cretaceous (125–110 Ma) or early Cenozoic (65–55 Ma) regional cooling episodes. For reasons outlined in more detail in a later section, the 140–60 Ma cooling episode observed at Larne No. 2 is interpreted to correlate with the early Cretaceous cooling episode, which has also been identified in the sedimentary basins offshore eastern (Green et al. 2001a; Holford et al. 2005a) and western Ireland (Corcoran & Mecklenburgh 2005).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6

Estimates of the time at which cooling episodes began (see also Table 1) for each of the AFTA samples from Larne No. 2, as determined by thermal history interpretation of AFTA data. The horizontal ranges correspond to 95% confidence intervals for each sample, whilst the vertical bars represent the degree of the overlap of timing constraints for each cooling episode (as with Fig. 5, the black rectangles indicate the stratigraphic age of each AFTA sample). This analysis allows the identification of three phases of cooling, beginning during the Jurassic (200–135 Ma), Cretaceous (140–60 Ma) and Cenozoic (35–5 Ma), respectively.

Estimating deeper burial using AFTA data

Figure 7 shows palaeotemperature estimates from the six AFTA samples plotted against depth. The palaeotemperature constraints for all of the cooling episodes can be satisfied by linear palaeotemperature profiles that have similar gradients to the calculated present-day geothermal gradient 24°C km⁻¹. This suggests that we can interpret each of the episodes in terms of deeper burial followed by exhumation. The AFTA data thus indicate that the preserved section at Larne No. 2 achieved maximum burial depths prior to exhumation, which began during the Jurassic, with subsequent phases of cooling and exhumation during the Cretaceous and late Cenozoic. The absence of post-Triassic sediments at this location means that it is not possible to determine whether Larne No. 2 experienced a continuous exhumation history since the Jurassic, with accelerations in the Cretaceous and late Cenozoic, or whether the exhumation episodes were separated by intervening periods of burial. The fact that Cretaceous sediments and ALG basalts crop out within a few kilometres of the borehole, however (Fig. 3), suggests that a history of vertical motions, involving multiple phases of burial and exhumation, provides the most likely scenario. Therefore, we can use the regional geology to guide the interpretation of the AFTA data.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7

AFTA-derived palaeotemperature constraints plotted against depth. Palaeotemperature profiles for each episode are approximately sub-parallel to the present-day geothermal gradient, indicating that the causes of heating and cooling for each palaeothermal episode were burial and exhumation, respectively. This analysis indicates that maximum burial of the preserved Permian–Triassic succession at Larne No. 2 was achieved prior to the Jurassic cooling and exhumation episode.

We apply maximum likelihood theory to define the range of values of palaeogeothermal gradient and the thicknesses of section eroded during the Jurassic, Cretaceous and Cenozoic cooling episodes, consistent with the measured palaeotemperature constraints within 95% confidence limits. This approach entails fitting a linear profile to the palaeotemperature data to provide an estimate of the palaeogeothermal gradient. This is extrapolated to an assumed palaeosurface temperature to estimate the amount of missing section eroded during exhumation (Bray et al. 1992).

Figure 8 shows estimates of additional section and palaeogeothermal gradients prior to the Jurassic and Cretaceous cooling episodes. The shaded region defines the allowed range of values for each parameter within 95% confidence limits. Assuming that the Jurassic and Cretaceous palaeogeothermal gradients were comparable to the present-day value of 24°C km⁻¹, and using a Mesozoic palaeosurface temperature of 20°C (cf. Yalçin et al. 1997), the results indicate that the preserved Permian–Triassic succession at Larne No. 2 was buried by an additional 2–2.45 km section prior to Jurassic cooling, and 1.45–1.87 km of section prior to Cretaceous cooling.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 8

Amounts of additional section and palaeogeothermal gradients required to explain the Jurassic and Cretaceous palaeotemperatures observed in the Larne No. 2 borehole. For each of the cooling episodes, an assumed palaeosurface temperature of 20°C was used on the basis of palaeoclimatic evidence presented by Yalçin et al. (1997). The shaded regions define the parameter ranges consistent with the respective palaeotemperature constraints within 95% confidence limits, whilst the black dots indicate the maximum likelihood estimates for each episode. Estimates of the amount of additional section prior to each cooling episode, assuming that the palaeogeothermal gradient was equivalent to the present-day gradient, are also shown. Independent estimates of the magnitude of deeper burial following analysis of Mercia Mudstone Group compaction data (see Fig. 10) are shown for comparative purposes. Note the excellent agreement between the estimates from these separate techniques.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 10

Density-log-derived porosities for the Mercia Mudstone Group of the Larne No. 2 borehole plotted against depth and the porosity–depth trend for North Sea shales determined by Sclater & Christie (1980). The calculated trends are considerably lower (by c. 20–50%) than those predicted by the porosity–depth relationship of Sclater & Christie (1980), which indicates that all the Mercia Mudstone Group and underlying units are not currently at their maximum burial depths.

We show constraints on palaeogeothermal gradients and additional section required to explain the late Cenozoic palaeotemperature constraints from AFTA in Figure 9. Again, assuming a palaeogeothermal gradient equivalent to the present-day gradient of 24°C km⁻¹ and a surface temperature of 20°C, between 1.1 km and 1.3 km of additional section are required to explain the Late Cenozoic paleotemperatures define from AFTA.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 9

Amounts of additional section and palaeogeothermal gradients required to explain the Cenozoic palaeotemperatures observed in the Larne No. 2 borehole. A palaeosurface temperature of 15°C was assumed in this analysis.

Estimating deeper burial using compaction data

The estimates of the amount by which the Permian–Triassic succession at Larne No. 2 borehole has been more deeply buried based on AFTA data were independently corroborated by the analysis of the amount of burial-driven overcompaction within the fine-grained units of the Mercia Mudstone Group encountered by the borehole. Compaction-driven porosity reduction is a commonly used measure of former burial depths in sedimentary basins, and sedimentary successions in exhumed basins exhibit anomalously low porosities (Hillis 1991; Thomson & Hillis 1995; Japsen 1998, 2000; Holford et al. 2005a; Japsen et al. 2007). The amount by which a sedimentary succession has been more deeply buried prior to exhumation can be estimated by comparison with a reference compaction curve for a sedimentary succession at maximum burial depth at the present day under conditions of hydrostatic pressure (Japsen et al. 2002).

We used a density log to estimate the porosities of mudstones and siltstones of the Mercia Mudstone Group, following the approach described in Holford et al. (2005a). Fine-grained siliciclastic sediments, such as muds and silts, are generally preferred to coarse-grained lithologies (e.g. sandstones) when estimating former burial depths because they exhibit relatively simple compaction trends and are less susceptible to anomalous compaction behaviour, e.g. due to diagenesis (Japsen et al. 2002). We compare the calculated porosities with Sclater & Christie's (1980) compaction trend for shales in the Central Graben of the North Sea. Before porosities were calculated, we edited the density log to ensure lithological consistency with the reference compaction trend. We used core descriptions to edit the log to separate out fine-grained units from other lithologies. The Mercia Mudstone Group in Larne No. 2 mainly consists of mudstones and siltstones but there are subordinate interbedded sandstones, several thick halite horizons and numerous igneous intrusions of presumed Palaeogene age (Penn 1981; Penn et al. 1983) that have been removed. Following this step, a gamma-ray log (representing a proxy for shale volume, e.g. Rider 1996) was used to identify and remove parts of the density log where the gamma-ray response was less than 40 API units (i.e. GR >40 API = shale).

We then manually removed spurious data points (i.e. anomalously low or high densities). We calculated porosities by sampling and averaging the log at 20 m intervals and using equation (1) (Rider 1996): ϕ=(ρma−ρb)/(ρma−ρf) (1) where ρ_ma is the matrix (grain) density, ρ_b is the bulk density and ρ_f is the density of the pore fluids. The highly variable nature of shale matrix density can introduce considerable errors when converting density data to porosities. For this reason, two sets of porosities were calculated using end-member matrix densities of 2.67 g cm⁻³ and 2.72 g cm⁻³ (Rider 1996). We used a pore fluid density of 1.01 g cm⁻³ for both sets of calculations.

Calculated porosities show a general reduction from c. 24–28% to c. 8–14% over the c. 1 km thick Mercia Mudstone Group succession (Fig. 10), although there is a restricted zone of higher porosities (i.e. c. 16–20%) just above the boundary with the underlying Sherwood Sandstone Group. It is apparent though that over all depths, the observed porosities are much lower than those predicted for equivalent depths by the Sclater & Christie (1980) shale compaction trend. This indicates that the Mercia Mudstone Group at Larne No. 2 is not currently at its maximum burial depth, confirming the results from AFTA data.

Figure 10 shows net exhumation estimates (see Corcoran & Doré 2005) relative to the Sclater & Christie (1980) shale trend. For a matrix density of 2.67 g cm⁻³, the density data yield a net exhumation estimate of 2.43 km (standard deviation 0.49 km), and, for a matrix density of 2.72 g cm⁻³, the data yield a net exhumation estimate of 2.09 km (standard deviation 0.39 km). Collectively, therefore, the compaction data indicate deeper burial of the Mercia Mudstone Group and underlying units by an additional 2.09–2.43 km of removed section. This estimate is in good agreement with that of Shelton (1997), who compared sonic velocity data from the Mercia Mudstone Group against the shale compaction trends of Marie (1975) and Bulat & Stoker (1987) and estimated that between 1.93 and 2.27 km of additional section had been removed by exhumation. As demonstrated in Figure 8, the estimates based on density data are also highly consistent with the estimates of the amount of deeper burial prior to Jurassic exhumation from AFTA data (2–2.45 km). There are thus a number of lines of evidence to suggest that the preserved Permian–Triassic succession at Larne No. 2 was more deeply buried by >2 km prior to exhumation which began in the Jurassic.

Jurassic exhumation

The AFTA and compaction results described above indicate that the preserved c. 2.9 km thick Permian–Triassic succession at Larne No. 2 reached maximum burial depths in the Jurassic as a result of an additional c. 2–2.45 km of section which subsequently has been eroded. AFTA indicates that the cooling associated with this first exhumation episode began between 200 Ma and 135 Ma, which implies that a thick late Triassic–Jurassic succession had accumulated at this location prior to exhumation. Although no Jurassic rocks were encountered by Larne No. 2, Lower Jurassic rocks do crop out within a few kilometres of the borehole (e.g. at Waterloo; Mitchell 2004) and the Larne No. 1 borehole encountered 51.52 m of Hettangian sediments (Manning & Wilson 1975). The maximum-recorded thickness of Jurassic rocks in Northern Ireland is 248 m in the Port More borehole on the northern coastline (Mitchell 2004). The observation that the UWLF Chalk sequence rests unconformably on tilted and faulted Lower Jurassic strata (e.g. George 1967; Roberts 1989) provides strong independent support for the burial and subsequent exhumation of the preserved Lower Jurassic succession in Northern Ireland.

The timing of this exhumation is constrained only loosely by AFTA data from the Larne No. 2 borehole, but the interpretation of a more extensive AFTA database containing samples from across onshore Ireland identified a widespread mid-Jurassic cooling episode which began between 180 Ma and 170 Ma (Green et al. 2000). Tate & Dobson (1989) identified an angular unconformity of pre-Bathonian age (i.e. before 167.7±3.5 Ma; Gradstein et al. 2004) on seismic data from the North Porcupine Basin, offshore western Ireland, and also noted the existence of a number of angular unconformities and hiatuses dated as latest Toarcian (183.0±1.5 to 175.6±2.0 Ma; Gradstein et al. 2004) in the Sea of the Hebrides and Cardigan Bay Basins, offshore eastern Ireland. These observations led Tate & Dobson (1989) to propose an E–W-trending phase of mid-Jurassic uplift and erosion focused on Northern Ireland and extending outwards to the North Porcupine and Irish Sea basins. To the north and south of this proposed high, the Sea of the Hebrides and Cardigan Bay basins contain some of the thickest preserved Bajocian–Bathonian successions (c. 600 m and c. 850 m, respectively) in NW Europe, and the direction of sediment supply in these basins indicates deposition on northerly- and southerly-orientated palaeoslopes, respectively (Tate & Dobson 1989).

The timing of the proposed episode of mid-Jurassic uplift and erosion (Toracian to Bathonian, i.e. c. 183 to c. 168 Ma) is highly consistent with that of the mid-Jurassic cooling episode (180–170 Ma) revealed by AFTA data from onshore Ireland (Green et al. 2000). We suggest here that the earliest cooling episode identified by AFTA data from Larne No. 2 borehole records this regional mid-Jurassic exhumation event, and that the 2–2.45 km of additional section required in order to explain the palaeotemperature and compaction results therefore comprised sediments of Upper Triassic to Middle Jurassic age. It is possible to estimate the amount of section removed during mid-Jurassic exhumation by comparing the estimates of additional section prior to Jurassic and Cretaceous cooling provided by AFTA data. Assuming a constant geothermal gradient of 24°C km⁻¹ (equivalent to the value of the present-day gradient), Figure 8 indicates that the preserved section at Larne No. 2 was more deeply buried by 2–2.45 km prior to Jurassic cooling, and by 1.45 to 1.87 km prior to Cretaceous cooling. If no deposition took place between these separate episodes, minimum and maximum estimates of exhumation are 0.13 km and 1 km, respectively. The actual amount could have been higher if intervening burial occurred between the two exhumation episodes. Assuming no deposition, a figure tending towards the upper end of this range of possible estimates is most likely in order to elicit a sufficient response in the fission-track data to enable the recognition of the cooling episode.

The timing of the mid-Jurassic exhumation episode described here correlates well with the major intra-Aalenian ‘mid-Cimmerian’ unconformity attributed to plume-related crustal doming in the central North Sea (Ziegler 1990; Underhill & Partington 1993). Estimates of the amount of erosion near the centre of the North Sea dome vary between 0.25 km and 2 km (Hesselbo 2000). Tate & Dobson (1989) noted that their proposed mid-Jurassic Irish uplift axis lay on a comparable palaeolatitude to the similarly E–W-aligned North Sea dome and, thus, suggested that the mid-Jurassic uplift of Northern Ireland may have occurred either as a westward extension of the North Sea dome or as a separate entity. More work is required to establish a topographic link between these areas, but we note here that AFTA data from the more proximal East Midlands Shelf and southern North Sea, where the Jurassic succession is more complete, provide no evidence for significant Jurassic cooling and exhumation (Green et al. 2001b; Green 2005).

Cretaceous exhumation

AFTA results indicate that the intermediate cooling episode at Larne No. 2 began between and 140 Ma and 60 Ma (Fig. 6) and that, prior to cooling, the preserved section was more deeply buried by 1.45–1.87 km (Fig. 8). The broad timing interval for this cooling episode could allow an exhumation episode during either the early Cretaceous or early Cenozoic, and major exhumation at these times is recognised across onshore Ireland and the Irish Sea basin system (Green et al. 2000, 2001a; Holford et al. 2005a, 2009b). Based on regional geological and thermochronological evidence, we propose that the 140–60 Ma cooling episode at Larne No. 2 most likely records early Cretaceous exhumation. The preservation of the Upper Cretaceous limestones of the UWLF (youngest member is the c. 68 Ma early Maastrichtian Ballycastle Chalk) (Mitchell 2004) beneath the Paleocene basalts of the ALG (earliest flows dated as c. 58 Ma (Bell & Jolley 1997) but possibly as old as c. 61 Ma (Mitchell 2004)) implies only limited early Palaeogene exhumation. A striking unconformity between the karstified Chalk and basalt ( Fig. 11) shows that the UWLF was uplifted and eroded prior to basalt emplacement (Simms 2000). The UWLF sequence is condensed, indicating shelfal (as opposed to deep-sea) deposition (cf. Oakman & Partington 1998) and Maastrichtian water depths around the British Isles are estimated at 100–150 m (Cope 2006). Withdrawal of a sea 150 m deep would result in c. 60 m of isostatic uplift (Cope 2006), which means that it is not essential to invoke a significant amount of tectonic uplift to explain the subaerial exposure of the UWLF.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 11

Photograph of the distinctive unconformity between the Upper Cretaceous UWLF Chalk and the Palaeogene ALG basalts at the White Rocks, near Portrush. Courtesy of Ian Cowe.

The fact that our AFTA data do not require significant early Palaeogene cooling indicates that any uplift that occurred at this time was not accompanied by significant exhumation (uplift in the absence of erosion will not result in cooling). Isostatic calculations indicate that 150 m of uplift caused by either underplating or lithospheric shortening should result in 600 m of exhumation (Brodie & White 1995), but it is unlikely that 600 m of Chalk has been removed from Northern Ireland because the UWLF succession is thought to be effectively complete (Cope 1997). Based on local and regional geological constraints, Corcoran & Mecklenburgh (2005) proposed that the maximum thickness of Chalk eroded from the area around the Corrib Gas Field in the Slyne Basin is of the order of 50–150 m. The absence of overwhelming evidence for major early Palaeogene exhumation across Northern Ireland is surprising, given that several workers have made an explicit link between Paleocene magmatism and permanent uplift in the Irish Sea region (Brodie & White 1994; Rowley & White 1998; Al-Kindi et al. 2003; Tiley et al. 2004).

We believe that the intermediate cooling episode is more likely to have begun during the early Cretaceous, predating the preserved Chalk sequence. The oldest preserved Cretaceous rocks in Northern Ireland belong to the Cenomanian Belfast Marls Member of the Hibernian Greensand Formation. (Mitchell 2004). The base of the Cenomanian stage is dated as 99.6±0.9 Ma (Gradstein et al. 2004). Combining this age with the constraints on the onset of Cretaceous cooling from AFTA data indicates that Cretaceous exhumation at Larne No. 2 most likely began between 140 Ma and c. 100 Ma. Similarly timed cooling and exhumation episodes have been identified across the Irish landmass (beginning between 125 Ma and 110 Ma; Green et al. 2000), Central Irish Sea Basin (120 Ma and 115 Ma; Green et al. 2001a), East Irish Sea Basin (140 Ma and 110 Ma; Holford et al. 2005b) and Cardigan Bay Basin (150 Ma and 80 Ma; Holford et al. 2005a). Combining timing constraints from all AFTA data across the Irish Sea region enables the definition of a regional early Cretaceous exhumation episode which began between 120 Ma and 115 Ma (Holford et al. 2009b). AFTA, vitrinite reflectance (VR) and compaction data indicate that the preserved section across parts of the Irish Sea basin system were more deeply buried by as much as 1.5 km prior to this exhumation episode (Holford et al. 2009b), which is consistent with the amount of additional section at Larne No. 2 prior to Cretaceous exhumation (1.45–1.87 km).

Similarly timed exhumation was identified within the Slyne Basin, offshore northwest Ireland (Corcoran & Mecklenburgh 2005) and across southern England (McMahon & Turner 1998). The exhumation in the Irish Sea is also broadly synchronous with the development of the Jurassic–Cretaceous unconformity complex in the North Sea Basin, where numerous stratigraphic breaks of Oxfordian to Albian age are observed (Kyrkjebø et al. 2004). As discussed further by Holford et al. (2009b), the timing of the 120–115 Ma exhumation episode was coeval with continental break-up and the onset of North Atlantic seafloor spreading southwest of the British Isles, indicating a genetic relationship between these processes.

The hardness of the Chalk in Northern Ireland

The Chalk of the UWLF has a similar age and composition to the Chalk of southern England but is considerably more indurated (Scholle 1974; Simms 2000). The porosity of UWLF Chalk varies between 2% and 10% (Maliva & Dickson 1997) compared to 35–47% for the Chalk of southern England (Simms 2000). Figure 12 shows the porosities of the Irish and English Chalks compared against a number of Chalk porosity–depth trends based on data from various parts of NW Europe (Sclater & Christie 1980; Bulat & Stoker 1987; Hillis 1995; Japsen 1998; Mallon & Swarbrick 2002). Assuming that burial diagenesis is the primary control on porosity reduction, it is clear from Figure 12 that considerable burial of the Chalk is required to explain such low porosities.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 12

Comparison of the average porosities of the Chalk formations of Northern Ireland and southern England with a number of porosity–depth trends proposed for the Chalk of NW Europe. Modified after Mallon & Swarbrick (2002).

The origin of the hardness of the UWLF Chalk has been a source of discussion for many years (e.g. Hancock 1963; Wolfe 1968). There are two main hypotheses to account for the observed induration. The first hypothesis attributes the hardness to porosity reduction caused by mechanical compaction and pressure solution following deep burial diagenesis beneath a thick (up to c. 2 km) basalt cover (Wolfe 1968; Scholle 1974; Maliva & Dickson 1997). The second hypothesis proposes that intense cementation of the Chalk occurred in a freshwater environment during early diagenesis prior to basalt emplacement, therefore not requiring the need for compaction during burial (Mimran 1978; Simms 2000). This hypothesis is supported by the observation that the UWLF has been extensively karstified (Simms 2000). Simms (2000) has identified striking pinnacle karst topography with up to 15 m of relief along the unconformable contact between the UWLF and the ALG, (Fig. 11), and a number of cave passages filled with basalt lava flows within the UWLF. These observations led Simms (2000) to conclude that cementation and porosity reduction must have occurred prior to the emergence and karstification of the UWLF.

The UWLF Chalks have experienced a very complex history of diagenesis (Wolfe 1968; Mimran 1978), which makes it difficult to attribute their hardness to a single cause or process. There is evidence for extensive recrystallization, e.g. from overgrowths of calcite cement on coccoliths and microcrystalline calcite particles (Mimran 1978; Maliva & Dickson 1997; Mitchell 2004), but there is also abundant evidence for mechanical compaction (Wolfe 1968; Scholle 1974). Geochemical investigations of the diagenetic conditions under which recrystallization and cementation of the Chalk took place have produced conflicting results. Based on trace element concentrations and oxygen and carbon isotopic ratios, Mimran (1978) argued for cementation during early diagenesis in a freshwater environment. A similar study by Maliva & Dickson (1997) led these workers to propose that Chalk cementation and recrystallization occurred during deep-burial diagenesis in a low water–rock ratio system with marine or mixed marine-meteoric pore waters. This view is supported by Wolfe (1968), who suggested that the calcite responsible for secondary cementation of pore spaces was derived from pressure solution during compaction.

Chalks are deposited with initial porosities as high as c. 70–80%, and early dewatering reduces porosities to c. 50–60% within the first few tens to hundreds of metres of burial (Scholle 1977). Mechanical compaction can reduce Chalk porosity to c. 40% but, below this level, the dominant controls on porosity reduction are chemical compaction and cementation (Jones et al. 1984). Wolfe (1968) and Scholle (1974) have provided convincing petrographic evidence (i.e. modifications of primary sedimentary structures and deformation of microfossils) indicating that mechanical compaction has caused significant porosity reduction in the UWLF Chalk. Although there is variation in the amount of compaction on a sample-to-sample basis (Wolfe 1968), for the Chalk succession as a whole, mechanical compaction appears to have caused porosity reduction of up to c. 30% (Scholle 1974). The compilation of Chalk porosity–depth trends in Figure 12 show that a c. 30% reduction in porosity corresponds to burial depths of c. 1 km. This is consistent with the AFTA results that indicate deeper burial of c. 1.1–1.3 km at Larne No. 2 prior to mid-late Cenozoic exhumation. We submit that the hardness of the Chalk can be explained by a combination of early cementation (which can account for the distinctive landforms associated with the karstification of the UWLF; Simms 2000) followed by burial-driven mechanical compaction beneath a thick basalt cover.

The former extent and thickness of the Antrim Lava Group

There is no real consensus as to the original extent of the ALG prior to mid-late Cenozoic exhumation (Walker 1995; Corcoran & Mecklenburgh 2005). The present-day extent of the ALG is >4300 km² and the maximum known thickness is 780 m (Walker 1995). Some palaeogeographical reconstructions depict the basalts of the ALG extending far beyond their present-day erosional limits (Preston 1982) and perhaps forming a continuous field with the contemporaneous basalts preserved on the island of Skye (Murray 1992) with which the ALG shares geochemical affinities (Bell & Jolley 1997). Other workers have suggested that the original lava pile is essentially intact due to the close correlation between the preserved limits of the ALG and the known margins of the underlying Mesozoic basins (Simms 2000). This suggestion is unlikely given the presence of basalt outliers that indicate the minimum extent of the original lava field at Knocklayd, Slieve Gallion and Slieve Gullion, some distance from the margins of the main lava field outcrop (Mitchell 2004).

Walker (1995) presented some estimates for the original thickness of the ALG based on the distribution of zeolites. Noting that (a) the top of the analcime zeolite zone occurs at a depth of 600 m in the basaltic pile of eastern Iceland, and (b) in the peripheral parts of the ALG, analcime characterizes an almost continuous zone along the base of the lavas that is c. 100 m thick, Walker (1995) concluded that the original thickness of the lava pile was at least 700 m. Using a similar approach, Walker (1995) estimated that the original thickness of the ALG at the Langford Lodge borehole was c. 1060 m. This borehole encountered 780.2 m of basalts (McCaffrey & McCann 1992), implying that c. 300 m of basalt has been removed at this location.

Given that AFTA data indicate deeper burial at Larne No. 2 by 1.1–1.3 km of now-removed overburden prior to exhumation which began between 35 Ma and 5 Ma, it seems entirely plausible that c. 0.7 km of basalt could have accumulated in the Larne area during the early Palaeogene. The principal factors that influence the length of a lava flow are the rates of effusion and cooling, the total volume and initial viscosity of the lava, and the nature of the underlying topography (Walker 1973; Keszthelyi & Self 1998). Constraining the first four factors is beyond the scope of this study, but the area around Larne at the present day is characterized by extremely low relief, which means that topography is unlikely to have been a significant obstacle to flowing lavas. The thickness of individual basalt flow units is sensitive to the angle of the ground-slope. Walker (1995) reported morphometric analyses of lava flows in Hawaii, which suggested that on slopes >4° average flow units are c. 1 m thick, whereas on slopes <2° average thicknesses are 5 m or more. The average thickness of flow units in the ALG is c. 7 m, indicating that the lavas formed on slopes <2°. Despite the fact that the ALG lavas appear to have been erupted onto an essentially horizontal surface (cf. George 1967), the present-day dips of the lavas vary between 5° and 15°, which suggests significant post-emplacement tilting of the ALG (Fig. 1) (Walker 1995). The dip of the ALG is largely centripetal, delineating a saucer-shaped structure that is elongated NW–SE (Walker 1995). This saucer shape can be explained partly by post-volcanic extension and subsidence associated with the Lough Neagh pull-apart basin. The Lough Neagh basin was formed by dextral motions along NNW-trending right-stepping faults that accommodated the accumulation of Oligocene non-marine (lacustrine) clays and lignites of the Lough Neagh Group, which overly the ALG with a maximum known thickness of 381 m encountered by the Derryinver borehole (Cunningham et al. 2003; Quinn 2006).

The Lough Neagh Group (LNG) enables further constraint to be placed on the timing of the 35–5 Ma mid-late Cenozoic exhumation episode that removed 1.1–1.3 km of overburden from Larne No. 2, including, as discussed above, up to 700 m of basalt. Palynological data indicate a Chattian age (28.4±0.1 to 23.03 Ma (Gradstein et al. 2004)) for the LNG (Wilkinson et al. 1980), which is similar to the ages of other non-marine Cenozoic sediments in the western British Isles deposited in basins that were also influenced by strike-slip faulting (e.g. the Cardigan Bay and Bovey basins) (Turner 1997; Cunningham et al. 2004; Williams et al. 2005). The LNG unconformably overlies the ALG, indicating that some erosion of basalt occurred prior to the accumulation of the LNG (Mitchell 2004). For example, provenance studies indicate that the ALG was the dominant source for the c. 240 m thick LNG succession in the Ballymoney pull-apart basin (Parnell et al. 1989). However, for reasons outlined below, we prefer a post-Oligocene age for the mid-late Cenozoic exhumation episode at Larne No. 2.

The timing of mid-late Cenozoic exhumation at Larne No. 2

In their regional study of post-Variscan exhumation across Ireland, Green et al. (2000) identified a late Cenozoic cooling episode beginning between 25 Ma and 15 Ma in the majority of AFTA samples they analysed. Cooling and exhumation of comparable age is observed across much of the southern British Isles (Hillis et al. 2008; Holford et al. 2009b), while Holford et al. (2008) present AFTA, VR and seismic data from the southern Irish Sea basins which show that the main phase of Cenozoic exhumation occurred between 20 Ma and 15 Ma. Detailed mapping of the distribution of exhumation shows that areas with the highest amounts of removed section (up to 2.5 km in the Wessex–Weald and southern Irish Sea basins) are consistently associated with late Cenozoic-age compressional structures, and this localized shortening-related exhumation is superimposed upon more regional uplift of Neogene age (Hillis et al. 2008). A number of lines of evidence indicate that late Cenozoic compressional deformation affected Northern Ireland. Roberts (1989) demonstrated that amplitudes of present-day structure contours on the base of the ALG vary by more than 500 m over wavelengths of c. 20 km, and the LNG sequence in the Lough Neagh Basin shows similar warping, which indicates a post-Oligocene age for deformation ( Fig. 13) (cf. George 1967). Roberts (1989) also suggested that faults which displace the basalts, such as the Tow Valley Fault (Fig. 1), are probably also related to post-Oligocene deformation, noting that isoclinal folding of basalt is observed along splays of this fault on Rathlin Island. Seismic data from the offshore Larne Basin c. 10 km north of Larne No. 2 indicate reversal of motion along the Ballytober Fault, which probably occurred during the late Cenozoic (Fig. 13a) (Mitchell 2004). George (1967) suggested that the primary landforms of the Northern Ireland region formed following Neogene uplift, whilst Dewey (2000) and Cunningham et al. (2004) describe numerous examples of late Cenozoic deformation and uplift from across the Irish landmass.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 13

Interpreted seismic sections from Northern Ireland, showing evidence for folding of the ALG and, thus, post-Paleocene compressional deformation. There is also some evidence for minor reverse-reactivation of the Ballytober Fault in (a). Modified after Mitchell (2004).

We propose here that the NNW–SSE-trending anticline which brings Triassic rocks to outcrop at Larne No. 2 formed during late Cenozoic compressional deformation which, based on regional stratigraphic and thermochronologic data, began between 20 Ma and 15 Ma (Holford et al. 2009b). This deformation caused 1.1–1.3 km of exhumation that ensured the complete removal of as much as 700 m of ALG basalt from Larne No. 2, and this localized deformation can account for the removal of a greater thickness of basalt than at Langford Lodge, where c. 300 m may have been removed.

Assuming that the ALG basalts originally covered an area at least twice their current extent of c. 4300 km², as suggested by Walker (1995), and that a minimum thickness of 300 m has been removed due to late Cenozoic exhumation, this indicates that at least c. 2600 km³ of basalt has been eroded from Northern Ireland. The most likely destinations for the products of this erosion are the sedimentary basins west of Ireland that contain thick Neogene successions (Stoker et al. 2005).

An important outstanding question regarding the LNG is how these deposits survived the late Cenozoic exhumation of Northern Ireland. One possibility is heterogeneous exhumation, whereby less section was removed across most of Northern Ireland relative to areas of locally high exhumation, such as Larne No. 2. The difficulty in reactivating steep strike-slip faults, such as those that bound the Lough Neagh Basin under a compressional stress regime (Turner 1997), may have enabled this basin to escape the effects of exhumation. Another possibility is that the original LNG succession was considerably thicker prior to exhumation. We are not aware of any published studies investigating whether the Lough Neagh Group has been more deeply buried in the past. However, by comparison with the Oligocene–Miocene clays and lignites in the Mochras borehole, NW Wales, which AFTA, VR and compaction data show to have been buried beneath an additional c. 1.5 km of section removed during late Cenozoic exhumation (Holford et al. 2005a), we think it likely that some of the original thickness of Lough Neagh Group was removed during late Cenozoic exhumation.